Improvement in measurement accuracy

Abstract

One aspect of the present invention is to provide a system and method for improving measurement accuracy, which includes at least one or more parameters each having a random error.

Description

[0001] Cross - reference to Related Applications

[0002] This application claims the priority benefit of U.S. Provisional Patent Application No. 62 / 724,025, filed on August 28, 2018, and U.S. Provisional Patent Application No. 62 / 742,247, filed on October 5, 2018, the contents of which are incorporated herein by reference in their entirety. The entire disclosure of any publication or patent document mentioned herein is incorporated by reference in its entirety. Technical Field

[0003] In particular, the present invention relates to devices and methods for performing biological and chemical assays, and more particularly to improving the accuracy and reliability of assays when performed using simple devices and limited resource settings. Background Art

[0004] When assaying biomarkers in a sample from a subject (e.g., a human) for diagnosing a disorder or disease, the accuracy of the assay is essential. Incorrect results can be harmful to the subject. Traditionally, the accuracy of an assay has been achieved through a "perfect protocol paradigm", i.e., precisely performing all operations including sample handling. This approach requires complex machinery, specialized operations, ideal environments, etc. to ensure "perfect" assay devices and "perfect" assay performance and operations. However, there is a great need to develop systems and methods for improving assay accuracy that involve at least one parameter each having a random error. Summary of the Invention

[0005] The following brief summary is not intended to include all features and aspects of the present invention.

[0006] One aspect of the present invention is to provide systems and methods for improving assay accuracy, where the assay involves at least one parameter each having a random error. In addition to biochemical measurements of the analyte in the sample, the accuracy is improved by examining the credibility of the assay.

[0007] Another aspect of the present invention is to provide systems and methods for improving assay accuracy by using monitoring structures distributed within the sample, where the assay involves at least one parameter each having a random error.

[0008] Another aspect of the present invention is to provide systems and methods for improving assay accuracy by using the credibility of the sample used in the assay, where the assay involves at least one parameter each having a random error.

[0009] In some embodiments, as shown in FIG. 1, the present invention provides a method for improving the accuracy of determining an analyte in or suspected of being in a sample, wherein the apparatus or operation of the determination has one or more parameters each having a random variation, and the method comprises: (a) detecting the analyte in the sample containing or suspected of containing the analyte using the determination, wherein the detection comprises: (i) placing the sample into the determination apparatus; and (ii) detecting the analyte using the determination apparatus to produce a detection result; and (b) determining the credibility of the detection result in step (a), comprising: (i) using an imager to capture one or more images of at least a portion of the sample and / or at least a portion of the determination apparatus, wherein the images substantially represent the conditions for measuring at least the portion of the sample when the detection result is produced in step (a); and (ii) determining the credibility of the detection result in step (a) and generating a credibility score by analyzing the images using an algorithm, wherein the algorithm comprises comparing the one or more images with training data, and wherein the training data comprises a random variation of one of the one or more parameters and / or a statistical response of the accuracy of the determination to the random variation of the one or more parameters; and (c) reporting the detection result and the credibility score.

[0010] In some embodiments, the present invention provides a method for improving the accuracy of determining an analyte in or suspected of being in a sample, wherein the apparatus or operation of the determination has one or more parameters each having a random variation, and the method comprises: (a) placing the sample into the determination apparatus; (b) using an imager to capture one or more images of at least a portion of the sample and at least a portion of the determination apparatus; (c) analyzing the one or more images to detect the analyte in the sample and produce a detection result; and (d) using an algorithm for analyzing the one or more images to determine the credibility of the detection result and generate a credibility score, wherein the algorithm comprises comparing the one or more images with training data, and wherein the training data comprises a random variation of one of the one or more parameters and / or a statistical response of the accuracy of the determination to the random variation of the one or more parameters; and (e) reporting the detection result and the credibility score.

[0011] In some embodiments, as shown in FIG. 2, a method for improving the accuracy of determining an analyte in or suspected of being in a test sample, wherein the apparatus or operation of the determination has one or more parameters each having a random variation, the method comprising: (a) placing the sample into the determination apparatus, wherein the determination apparatus has at least one sample contact area for contacting the sample, and wherein the at least one sample contact area includes one or more monitoring structures; (b) using an imager to capture one or more images of at least a portion of the sample and at least a portion of the monitoring structures; (c) analyzing the one or more images to detect the analyte in the sample and generating a detection result; and (d) using an algorithm for analyzing the one or more images to determine the credibility of the detection result and generating a credibility score, wherein the algorithm includes comparing the one or more images with training data, and wherein the training data includes a random variation of one of the one or more parameters, the monitoring structures, and / or the statistical response of the accuracy of the determination to the random variation of the one or more parameters; (e) reporting the detection result and the credibility score; wherein the monitoring structures include structures for monitoring the operation of the determination and / or the optical properties of the quality of the determination apparatus.

[0012] In some embodiments, the present invention provides a method for improving the accuracy of an assay for an analyte in or suspected to be in a sample, wherein the assay apparatus or operation has one or more parameters each having a random variation, and the method comprises: (a) detecting the analyte in the sample containing or suspected to contain the analyte using the assay, wherein the detection comprises: (i) placing the sample into the assay apparatus; and (ii) detecting the analyte using the assay apparatus to generate a detection result; and (b) determining the credibility of the detection result in step (a), comprising: (i) using an imager to capture one or more images of at least a portion of the sample and / or at least a portion of the assay apparatus, wherein the images substantially represent the conditions for measuring at least the portion of the sample when generating the detection result in step (a); and (ii) determining the credibility of the detection result in step (a) and generating a credibility score by analyzing the images using a first algorithm, wherein the first algorithm comprises comparing the one or more images with training data, and wherein the training data comprises a random variation of one of the one or more parameters and / or a statistical response of the accuracy of the assay to the random variation of the one or more parameters; and (c) if the credibility score exceeds a threshold, reporting the detection result, otherwise proceeding to step (d); (d) if the credibility score exceeds the threshold and the detection result is reported, repeating steps (a), (b), (c) using a different assay apparatus; otherwise proceeding to step (e); (e) determining an average detection result and an average credibility score by analyzing all the detection results and all the images generated prior to step (e) using a second algorithm to generate an average detection result and an average credibility score, wherein the second algorithm comprises comparing the one or more images with training data, and wherein the training data comprises a random variation of one of the one or more parameters and / or a statistical response of the accuracy of the assay to the random variation of the one or more parameters; and (f) if the average credibility score exceeds the threshold, reporting the detection result, otherwise proceeding to step (d).

[0013] In some embodiments, the algorithm is machine learning.

[0014] As shown in Figure 3,

[0015] In some embodiments, the assay apparatus has at least one sample contact area for contacting the sample, and wherein the at least one sample contact area comprises one or more monitoring structures; wherein the monitoring structures comprise structures for monitoring the optical properties of the operation of the assay and / or the quality of the assay apparatus, and wherein the training data comprises the monitoring structures.

[0016] In some embodiments, the sample comprises at least one of the parameters having a random variation.

[0017] In some embodiments, the sample comprises at least one parameter having random variations, wherein the parameter comprises dust, bubbles, non-sample material, or any combination thereof.

[0018] In some embodiments, the sample comprises at least one parameter having random variations, wherein the parameter comprises dust, bubbles, non-sample material, or any combination thereof.

[0019] In some embodiments, the method further comprises the step of maintaining or rejecting the test result using the credibility score.

[0020] In some embodiments, the assay is a device for detecting an analyte using a chemical reaction.

[0021] In some embodiments, the assay is an immunoassay, a nucleic acid assay, a colorimetric assay, a luminescence assay, or any combination thereof.

[0022] In some embodiments, the assay device comprises two plates facing each other with a gap therebetween, wherein at least a portion of the sample is within the gap.

[0023] In some embodiments, the assay device comprises a QMAX, which comprises two plates movable relative to each other and a spacer for adjusting the spacing between the plates.

[0024] In some embodiments, some of the monitoring structures are periodically arranged.

[0025] In some embodiments, the sample is selected from cells, tissues, body fluids, and feces.

[0026] In some embodiments, the sample is amniotic fluid, aqueous humor, vitreous humor, blood (e.g., whole blood, fractionated blood, plasma, serum, etc.), breast milk, cerebrospinal fluid (CSF), earwax (cerumen), chyle, chyme, endolymph, perilymph, feces, gastric acid, gastric juice, lymph, mucus (including nasal drainage and sputum), pericardial fluid, peritoneal fluid, pleural fluid, pus, inflammatory exudate, saliva, sebum (skin oil), semen, sputum, sweat, synovial fluid, tears, vomit, urine, and exhaled condensate.

[0027] In some embodiments, the analyte comprises molecules (e.g., proteins, peptides, DNA, RNA, nucleic acids, or other molecules), cells, tissues, viruses, and nanoparticles.

[0028] In some embodiments, the sample is a non-flowable but deformable sample.

[0029] In some embodiments, the method further comprises the step of discarding the test result generated in step (a) if the value determined in step (b) is below a threshold.

[0030] In some embodiments, the method further comprises discarding the detection result generated in step (a) if the value determined in step (b) is below a threshold.

[0031] In some embodiments, the present invention provides an apparatus for improving the accuracy of an assay having one or more unpredictable and random operating conditions, comprising: (1) a detection device that detects an analyte in a sample to generate a detection result, wherein the sample contains or is suspected of containing the analyte; (2) an inspection device that inspects the credibility of a particular detection result generated by the detection device, comprising: (i) an imager that is capable of acquiring one or more images of (1) a portion of the sample and / or (2) a portion of the detection instrument surrounding the portion of the sample, wherein the image substantially represents the conditions under which the portion of the sample was measured when the detection result was generated in step (a); and (ii) a calculation unit having an algorithm capable of analyzing features in the image taken in step (b)(i) to determine the credibility of the detection result; (c) discarding the detection result generated in step (a) if step (b) determines that the detection result is not credible; wherein step (a) has one or more unpredictable and random operating conditions.

[0032] In some embodiments, the algorithm is machine learning, artificial intelligence, statistical methods, or a combination thereof.

[0033] In some embodiments, the present invention provides a method for assaying a sample having one or more operating conditions with random variations, comprising: (a) providing a sample containing or suspected of containing an analyte; (b) depositing the sample onto a solid surface; (c) after step (b), measuring the sample to detect the analyte and generate a detection result, wherein the result can be affected by one or more operating conditions when performing the assay, and wherein the operating conditions are random and unpredictable; (d) imaging a portion of the sample area / volume where the analyte in the sample was measured in step (c); and (e) determining the error risk probability of the result measured in step (c) by analyzing one or more operating conditions shown in one or more images generated in step (d).

[0034] In some embodiments, if step (e) determines that the result measured in step (c) has a high error risk probability, the result will be discarded.

[0035] In some embodiments, the present invention provides an apparatus for determining an analyte present in a sample under one or more operating variables, comprising: (a) a solid surface having a sample contact area for receiving a thin layer of the sample, the sample comprising the analyte to be measured; (b) an imager configured to image a portion of the sample contact area that measures the analyte; and (c) a non-transitory computer-readable medium having instructions that, when executed, perform a determination of the credibility of the measurement result by analyzing the operating variables shown in an image of the portion of the sample.

[0036] In some embodiments, the present invention provides a method for determining a sample having one or more operating variables, comprising: (a) depositing a sample containing an analyte between a first plate and a second plate; wherein the sample is sandwiched between the first plate and the second plate that are substantially parallel; (b) measuring the analyte contained in the sample to produce a result, wherein the measurement involves one or more random and unpredictable operating variables; (c) imaging a region portion of the first plate and the second plate to produce an image, wherein the region portion contains the sample and measures the analyte contained in the sample; and (d) determining whether the result measured in step (b) is credible by analyzing the operating variables shown in the image of the region portion containing the sample.

[0037] In some embodiments, if the analysis in step (d) determines that the result measured in step (b) is not credible, the result is discarded.

[0038] In some embodiments, the present invention provides a method for assaying a sample having one or more operating variables, comprising: (a) depositing a sample containing or suspected of containing an analyte between a first plate and a second plate, wherein the sample is sandwiched between the first plate and the second plate, and the first plate and the second plate are movable relative to each other into a first configuration or a second configuration; (b) measuring the analyte in the sample to produce a result, wherein the measurement involves one or more random and unpredictable operating variables; and (c) imaging a region portion of the sample where the analyte is measured; and (d) determining whether the result measured in step (b) is credible by analyzing the operating variables shown in the image of the region portion, wherein the first configuration is an open configuration where the two plates are partially or fully separated, the spacing between the plates is not adjusted by a spacer, and the sample is deposited on one or both of the plates, and wherein the second configuration is a closed configuration, which is configured after the sample is deposited in the open configuration and the plates are forced into the closed configuration by applying an imprecise pressing force on the force-bearing region; and in the closed configuration: at least a portion of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, wherein the uniform thickness of the layer is defined by the sample contact regions of the two plates and is adjusted by the plates and the spacer.

[0039] In some embodiments, if step (d) determines that the result measured in step (b) is not credible, the result is discarded.

[0040] In some embodiments, the present invention provides a method for assaying a sample having one or more operating variables, comprising: (a) depositing a sample containing or suspected of containing an analyte in a region in a device of any of the embodiments described in the present disclosure; (b) measuring the analyte in the sample, wherein the measurement involves one or more random and unpredictable operating variables; and (c) imaging a portion of the sample region, wherein the portion is the location where the analyte is measured; and (d) determining whether the result measured in step (b) is credible by analyzing the operating variables shown in the image of the portion of the sample.

[0041] In some embodiments, if the analysis in step (d) determines that the result measured in step (b) is not credible, the result is discarded.

[0042] In some embodiments, multiple assay devices are used to perform the assay, wherein the assay has a step of using image analysis to check whether the assay result is credible, and wherein if the first assay device is found to be not credible, the second assay device is used until the assay result is found to be credible.

[0043] In some embodiments, the sample is a biological or chemical sample.

[0044] In some embodiments, in step (d), the analysis uses machine learning with a training set to determine whether the result is credible, where the training set uses operating variables of the analyte in the sample.

[0045] In some embodiments, in step (d), the analysis uses a look-up table to determine whether the result is credible, where the look-up table contains operating variables of the analyte in the sample.

[0046] In some embodiments, in step (d), the analysis uses a neural network to determine whether the result is credible, where the neural network is trained using operating variables of the analyte in the sample.

[0047] In some embodiments, in step (d), the analysis uses a threshold of the operating variables to determine whether the result is credible.

[0048] In some embodiments, in step (d), the analysis uses machine learning, a look-up table, or a neural network to determine whether the result is credible, where the operating variables include the condition of bubbles and / or dust in the image of the portion of the sample.

[0049] In some embodiments, in step (d), the analysis uses machine learning, which determines whether the result is credible and uses machine learning, a look-up table, or a neural network to determine the operating variables of bubbles and / or dust in the image of the portion of the sample.

[0050] In some embodiments, in step (b) of measuring the analyte, the measurement uses imaging.

[0051] In some embodiments, in step (b) of measuring the analyte, the measurement uses imaging, and the same image used for analyte measurement is used for the determination of credibility in step (d).

[0052] In some embodiments, in step (b) of measuring the analyte, the measurement uses imaging, and the same imager used for analyte measurement is used for the determination of credibility in step (d).

[0053] In some embodiments, the apparatus used in any of the preceding apparatus claims further comprises a monitoring marker.

[0054] In some embodiments, the monitoring marker is used as a parameter in an algorithm together with an imaging processing method, which (i) adjusts the image, (ii) processes the image of the sample, (iii) determines properties related to micro-features, or (iv) any combination of the above.

[0055] In some embodiments, the monitoring marker is used as a parameter together with step (b).

[0056] In some embodiments, the spacer is the monitoring marker, wherein the spacer has a substantially uniform height equal to or less than 200 microns, and a fixed spacer distance (ISD).

[0057] In some embodiments, the monitoring marker is used to estimate the TLD (true lateral dimension) and true volume estimation.

[0058] In some embodiments, step (b) further includes image segmentation for image-based determination.

[0059] In some embodiments, step (b) further includes focus checking in image-based determination.

[0060] In some embodiments, step (b) further includes the uniformity of the analyte distribution in the sample.

[0061] In some embodiments, step (b) further includes analyzing and detecting aggregated analytes in the sample.

[0062] In some embodiments, step (b) further includes analyzing the dry texture in the image of the sample in the sample.

[0063] In some embodiments, step (b) further includes analyzing defects in the sample.

[0064] In some embodiments, step (b) further includes correction of camera parameters and conditions, such as distortion removal, temperature correction, brightness correction, contrast correction.

[0065] In some embodiments, step (b) further includes methods and operations having histogram-based operations, mathematics-based operations, convolution-based operations, smoothing operations, derivative-based operations, morphology-based operations.

[0066] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the sample into a sample holding device (e.g., a QMAX device) having a gap proportional to the size of the analyte to be analyzed, or the analyte forms a single layer between the gaps; (b) using an imager to capture an image of the sample in the sample holding device over an area of interest (AoI) for determination; (c) segmenting the image of the sample captured by the imager in (b) into equally sized and non-overlapping sub-image patches (e.g., 8x8 equally sized small image patches); (d) performing machine learning-based inference using a trained machine learning model to perform analyte detection and segmentation on each image patch - to determine, but not limited to, analyte count and its concentration; (e) sorting the analyte concentrations of the constructed sub-image patches in ascending order and determining its 25th percentile Q1 and 75th percentile Q3; (f) determining the uniformity of the analyte in the sample image using a confidence measure based on the interquartile range: confidence-IQR = (Q3 - Q1) / (Q3 + Q1); and (g) if the confidence-IQR from (f) exceeds a specific threshold (e.g., 30%), raising a flag and the determination result is not credible, wherein the threshold is obtained from training / evaluation data or from physical rules governing the analyte distribution.

[0067] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the determination into a sample holding device (e.g., a QMAX device) having a gap proportional to the size of the analyte to be analyzed, or the analyte forms a single layer between the gaps; (b) using an imager to capture an image of the sample in the sample holding device over an area of interest (AoI) for determination; (c) performing machine learning-based inference using a trained machine learning model to perform dry texture detection and segmentation - to detect the dry texture area and determine the dry texture area in the AoI associated with a segmentation contour mask covering the dry texture area in the AoI of those sample images; (d) determining the area ratio between the dry texture area in the AoI (area-dry-texture-in-AoI) and the area of the AoI (area-of-the-AoI): ratio-dry-texture-area-in-AoI = dry texture area in the AoI / area of the AoI; and (e) if the ratio of the dry texture area in the AoI from (d) exceeds a specific threshold (e.g., 10%), raising a flag and the determination result is not credible, wherein the threshold is obtained from training / evaluation data or from physical rules governing the analyte distribution.

[0068] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the assay into a sample holding device (e.g., a QMAX device) having a gap proportional to the size of the analyte to be analyzed, or the analyte forms a monolayer between the gaps; (b) using an imager to take an image of the sample in the sample holding device over an area of interest (AoI) for the assay; (c) performing machine learning-based inference using a trained machine learning model for aggregated analyte detection and segmentation - to detect clustered analytes and determine the area associated with a segmentation contour mask covering them in the AoI (area of aggregated analytes in the AoI), the segmentation contour mask covering them in its AoI; (d) determining the area ratio between the area of aggregated analytes in the AoI and the area of the AoI: Area ratio of aggregated analytes in the AoI = Area of aggregated analytes in the AoI / Area of the AoI; and (e) if the area ratio of aggregated analytes in the AoI from (d) exceeds a specific threshold (e.g., 40%), raising a flag and the assay result is not credible, wherein the threshold is derived from training / evaluation data or from physical rules governing analyte distribution.

[0069] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the assay into a sample holding device (e.g., a QMAX device) having a gap proportional to the size of the analyte to be analyzed, or the analyte forms a monolayer between the gaps; (b) using an imager to take an image of the sample in the sample holding device over an area of interest (AoI) for the assay; (c) performing machine learning-based inference using a trained machine learning model to detect and segment defects in the sample image, where the defects include but are not limited to dust, oil, etc., to detect the defects and determine the area associated with a segmentation contour mask covering them in the AoI (defect area in the AoI); (d) determining the ratio between the defect area in the AoI and the area of the AoI: Defect area ratio in the AoI = Defect area in the AoI / Area of the AoI; and (e) if the defect area ratio in the AoI from (d) exceeds a specific threshold (e.g., 15%), raising a flag and the assay result is not credible, wherein the threshold is derived from training / evaluation data or from physical rules governing analyte distribution.

[0070] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the assay into a sample holding device (e.g., a QMAX device) having a gap proportional to the size of the analyte to be analyzed, or the analyte forms a single layer between the gaps; (b) using an imager to capture an image of the sample in the sample holding device over an area of interest (AoI) for the assay; (c) performing machine learning-based inference using a trained machine learning model for bubble and air gap detection and segmentation - to detect bubbles and air gaps and determine the bubble gap area in the AoI associated with the segmentation contour mask covering them; (d) determining the area ratio between the bubble gap area in the AoI and the area of the AoI: bubble gap area ratio in AoI = bubble gap area in AoI / area of AoI; and (e) if the bubble gap area ratio in the AoI from (d) exceeds a specific threshold (e.g., 10%), raising a flag and the assay result is not credible, wherein the threshold is derived from training / evaluation data or from physical rules governing the analyte distribution.

[0071] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the assay into a sample holding device (e.g., a QMAX device), wherein the sample holding device has a gap proportional to the size of the analyte to be analyzed, or the analyte forms a single layer between the gaps, and there are monitoring markers (e.g., pillars) located in the device and not submerged, which can be imaged with the sample by an imager on the sample holding device; (b) using an imager to capture an image of the sample in the sample holding device over an area of interest (AoI) for the assay; (c) performing machine learning-based inference using a trained machine learning model to detect and segment the monitoring markers (pillars) with analyte on top, to determine the area associated with the detected monitoring markers (area of analyte on pillar in AoI) based on their segmentation contour mask in the AoI; (d) determining the area ratio between the area of analyte on pillar in the AoI and the area of the AoI: area of analyte on pillar in AoI = area of analyte on pillar in AoI / area of AoI; and (e) if the area ratio of analyte on pillar in the AoI from (d) exceeds a specific threshold (e.g., 10%) raising a flag and the assay result is not credible, wherein the threshold is derived from training / evaluation data or from physical rules governing the analyte distribution.

[0072] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the assay into a sample holding device (e.g., a QMAX device); (b) using an imager to capture an image of the sample in the sample holding device over an area of interest (AoI) for the assay; (c) performing a machine learning-based focus check to detect whether the image of the sample captured by the imager is in focus on the sample, wherein the machine learning model for detecting the focus of the imager is built from a plurality of images of the imager with in-focus and out-of-focus conditions; and (d) if the image of the sample captured by the imager is detected as out-of-focus from (c), raising a flag and the assay result based on the image is not credible.

[0073] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the assay into a sample holding device (e.g., a QMAX device); (b) using an imager to capture an image of the sample in the sample holding device over an area of interest (AoI) for the assay; (c) performing a machine learning-based analyte detection; and (d) if the analyte count is extremely low outside a preset acceptable range, raising a flag and the result is not credible, wherein the acceptable range is specified based on the physical or biological conditions of the assay.

[0074] In some embodiments, the present invention provides a system for analyzing a sample under one or more operating conditions, comprising: (a) loading the assay into a sample holding device (e.g., a QMAX device); (b) using an imager to capture an image of the sample in the sample holding device over an area of interest (AoI) for the assay; (c) partitioning the image of the sample into non-overlapping, equal-sized sub-image patches; (d) performing a machine learning-based analyte detection on each of the sub-image patches; and (e) if for some of the sub-image patches, the count of the detected analyte is unrealistically low (e.g., in a complete blood count, the number of red blood cells in the sample is below the human-acceptable range), raising a flag and the result is not credible for insufficient sample or non-uniform distribution of the sample in the assay.

[0075] In some embodiments, detecting and segmenting anomalies in the image of the sample captured by the imager in an image-based assay is based on image processing, machine learning, or a combination of image processing and machine learning.

[0076] In some embodiments, the estimation of the area covered by the segmentation contour mask in the area of interest (AoI) of the image of the sample compensates for distortion in microscopy imaging using an estimate of the true lateral dimension (or field of view (FoV)) per image or per sub-image patch, including but not limited to spherical distortion from lenses, micro-level defects, misalignment in focusing, etc.

[0077] In some embodiments, the present invention further includes monitoring a plurality of markers (such as pillars) built into the sample holding device (such as a QMAX card); and the monitored markers (such as pillars) are applied as detectable anchors so that the estimation of the true lateral dimension (or field of view (FoV)) is accurate in the face of distortion in microscopy imaging.

[0078] In some embodiments, the monitored markers (such as pillars) of the sample holding device have some configurations with a prescribed periodic distribution in the sample holding device (such as a QMAX card) to make the detection and positioning of the monitoring markers as anchors in the true lateral dimension (TLD) (or field of view (FoV)) estimation reliable and robust.

[0079] In some embodiments, the detection and characterization of the outliers in the image-based assay are based on the non-overlapping sub-image patches of the input image of the sample described herein, and the determination of the outliers during the assay process can be based on non-parametric methods, parametric methods, and a combination of both.

[0080] In some embodiments, the present invention provides a method comprising: (a) detecting an analyte in a sample containing or suspected of containing the analyte, the detecting comprising: (i) depositing the sample into a detection instrument, and (ii) measuring the sample using the detection instrument to detect the analyte, thereby generating a detection result; (b) determining the reliability of the detection result, the determining comprising: (i) acquiring one or more images of a portion of the sample and / or a portion of the detection instrument adjacent to the portion of the sample, wherein the one or more images reflect one or more operating conditions under which the detection result is generated; and (ii) analyzing the one or more images using a computing device having an algorithm to determine the reliability of the detection result in step (a); and (c) reporting the detection result and the reliability of the detection result; wherein the one or more operating conditions are unpredictable and / or random.

[0081] In some embodiments, when assaying a sample in a limited resource setting (LRS), the results from the assay may be unreliable. However, traditionally, the reliability of a particular result has not been checked during or after a particular test of a given sample.

[0082] In some embodiments, in an LRS measurement (or even in a laboratory test environment), one or more unpredictable random operating conditions may occur and affect the measurement results. When this happens, even with the same sample, the results can be substantially different from one specific measurement to the next. However, rather than taking the measurement results as they are, the reliability of the specific results in a specific test of a specific sample can be evaluated by analyzing one or more factors related to the measurement operating conditions in the specific measurement.

[0083] In some embodiments, in an LRS measurement with one or more unpredictable random operating conditions, the overall accuracy of the measurement can be substantially improved by using an analysis of the reliability of each specific measurement and by rejecting untrustworthy measurement results.

[0084] In some embodiments, the measurement is performed not only by measuring the analyte in a specific test, but also by analyzing the operating conditions of the specific test to check the credibility of the measurement results.

[0085] In some embodiments, the credibility check of the measurement results in the measurement is modeled in a machine learning framework, and machine learning algorithms and models are designed and applied to handle the unpredictable random operating conditions that occur and affect the measurement results.

[0086] In some embodiments, the innovative use of machine learning has the following advantages: the process of automatically determining the credibility of the measurement results in the face of unpredictable random operating conditions in the measurement directly from the data without making explicit assumptions about the unpredictable conditions that may be complex, difficult to predict, and prone to errors.

[0087] In some embodiments, the machine learning framework in the present invention involves a process comprising the following steps: (a) collecting training data for the task; (b) preparing labeled data; (c) selecting a machine learning model; (d) training the selected machine learning model with the training data; (e) adjusting hyperparameters and model structure with training and evaluation data until the model achieves satisfactory performance on the evaluation and test data; and (f) performing inference on the test data using the trained machine learning model from (e). BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Those skilled in the art will understand that the drawings described below are for illustrative purposes only. The drawings are not intended to limit the scope of the present teachings in any way. The drawings are not drawn to scale. In the figures presenting experimental data points, the lines connecting the data points are for guiding the observation of the data only and have no other meaning.

[0089] FIG. 1 shows a block diagram of an embodiment for improving measurement accuracy by measuring the reliability of a measurement device and / or a measurement operation. In some embodiments, it checks the reliability of a sample.

[0090] FIG. 2 shows a block diagram of an embodiment for improving measurement accuracy by measuring the reliability of the measurement device and / or the measurement operation, where a monitoring structure is implemented inside the sample. In some embodiments, it checks the reliability of the sample.

[0091] FIG. 3 shows a block diagram of an embodiment for improving measurement accuracy by measuring the reliability of the measurement device and / or the measurement operation, where a decision is made on whether to report or not report a detection result.

[0092] FIG. 4 shows a block diagram of an embodiment for improving measurement accuracy by using multiple interactive measurements to measure the reliability of a measurement device and / or a measurement operation. In some embodiments, it checks the reliability of the sample.

[0093] FIGS. 5A and 5B show side views of a device for imaging-based assays. FIG. 5A shows a solid-phase surface with protruding monitoring markers. FIG. 5B shows a solid-phase surface with grooved monitoring markers. In an algorithm, characteristics corresponding to the monitoring markers (e.g., spacing and distance) can be used to determine the nature of an analyte in a sample.

[0094] FIGS. 6A and 6B show side views of a device for imaging-based assays. FIG. 6A shows how a monitoring marker (e.g., protruding type) can be a structure separated from a spacer. FIG. 6B shows how a monitoring marker can be a structure identical to a spacer. In an algorithm, characteristics corresponding to the monitoring markers can be used to determine the nature of an analyte in a sample.

[0095] FIG. 6C shows a side view of a device for imaging-based assays. FIG. 6C shows how a monitoring marker (e.g., grooved type) can be a structure separated from a spacer. In an algorithm, characteristics corresponding to the monitoring markers can be used to determine the nature of an analyte in a sample.

[0096] FIGS. 6D and 6E show side views of a device for imaging-based assays. FIGS. 6D and 6E show how a monitoring marker can be a structure separated from a spacer and disposed on two sample contact areas of the device. In an algorithm, characteristics corresponding to the monitoring markers can be used to determine the nature of an analyte in a sample.

[0097] FIG. 7 shows a block diagram of an embodiment of sample preparation and imaging of a sample holding device using a QMAX card in an image-based assay

[0098] FIG. 8 shows an image of red blood cells taken on a sample holding device of a QMAX card for an assay

[0099] Figure 9 shows an image of a sample for measurement having bubbles in the sample for measurement.

[0100] Figure 10 shows sample defects detected and segmented in the red blood cell image of the sample of Figure 9 having a covering mask in an embodiment of the described method.

[0101] Figure 11 shows another red blood cell image of a sample for measurement containing multiple defects.

[0102] Figure 12 shows defects detected and segmented in the image of the sample for measurement from Figure 12.

[0103] Figure 13 shows a defect in a sample holding device for measurement.

[0104] Figure 14 is a three-way diagram of a sample holding device QMAX card having a monitoring mark and imaging of the sample holding device in an image-based measurement.

[0105] Detailed Description of Exemplary Embodiments

[0106] The following detailed description shows some embodiments of the present invention by way of example and not limitation. The chapter headings and any subtitles used herein are for organizational purposes only and should not be construed as limiting the subject matter described in any way. The content under the chapter headings and / or subtitles is not limited to the chapter headings and / or subtitles, but applies to the entire description of the present invention.

[0107] Any reference to a publication is due to its being publicly available earlier than the filing date and should not be construed as an admission that the claims of the present invention are not entitled to antedate such publications by virtue of prior invention. Further, the provided publication dates may differ from the actual publication dates, which may need to be independently confirmed.

[0108] Definitions

[0109] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. Although any methods and materials similar or equivalent to those described herein can also be used in the practice or testing of this teaching, some exemplary methods and materials are now described.

[0110] As used herein, the term "sample" refers to a material or mixture of materials that contains one or more analytes or entities of interest. In certain embodiments, the sample is a bodily fluid sample from a subject. In some cases, a solid or semi-solid sample may be provided. The sample may include tissue and / or cells collected from a subject. The sample may be a biological sample. Examples of biological samples may include but are not limited to blood, serum, plasma, nasal swab, nasopharyngeal wash, saliva, urine, gastric fluid, spinal fluid, tears, feces, mucus, sweat, earwax, sebum, glandular secretions, cerebrospinal fluid, tissue, semen, vaginal secretions, interstitial fluid derived from tumor tissue, ocular fluid, spinal fluid, throat swab, exhaled condensate (e.g., breath), hair, nails, skin, biopsy, placental fluid, amniotic fluid, cord blood, lymph fluid, cavity fluid, sputum, pus, microbiota, meconium, breast milk, and / or other excretions. The sample may include a nasopharyngeal wash. Nasal swabs, throat swabs, fecal samples, hair, nails, earwax, breath, and other solid, semi-solid, or gaseous samples may be processed (e.g., fixed or for variable amounts of time) in an extraction buffer prior to their analysis. If desired, the extraction buffer or an aliquot thereof may then be processed similarly to other fluid samples. Examples of tissue samples from a subject may include but are not limited to connective tissue, muscle tissue, nerve tissue, epithelial tissue, cartilage, cancer samples, or bone. In specific embodiments, a sample may be obtained from a subject such as a human and may be processed prior to use in a subject assay. For example, prior to analysis, proteins / nucleic acids may be extracted from a tissue sample using methods known in the art. In certain embodiments, the sample may be a clinical sample, e.g., a sample collected from a patient. The sample may also be a sample of food, environment, etc.

[0111] The term "analyte" refers to any substance suitable for testing in the present invention. Analytes include but are not limited to atoms, molecules (e.g., proteins, peptides, DNA, RNA, nucleic acids, or other molecules), cells, tissues, viruses, bacteria, and nanoparticles of different shapes. Biomarkers are analytes.

[0112] As used herein, the terms "determine", "measure", "evaluate", and "assay" are used interchangeably and include both quantitative and qualitative determinations.

[0113] As used herein, the term "luminescent label" refers to a label that is capable of emitting light when externally excited. This may be luminescent. Fluorescent labels (which include dye molecules or quantum dots) and luminescent labels (e.g., electrochemiluminescent or chemiluminescent labels) are types of luminescent labels. The external excitation is light (photons) for fluorescence, current for electrochemiluminescence, and chemical reaction for chemiluminescence. The external excitation may be a combination of the above.

[0114] The phrase "labeled analyte" refers to an analyte that is detectably labeled with a luminescent tag such that the analyte can be detected by assessing the presence of the tag. The labeled analyte can be directly labeled (i.e., the analyte itself can be directly conjugated to the tag, e.g., by a strong bond such as a covalent or non-covalent bond), or the labeled analyte can be indirectly labeled (i.e., the analyte is bound by a second capture agent that is directly labeled).

[0115] The terms "spacer", "optical calibration marker", and "optical calibration marker" and "column" are interchangeable.

[0116] The term "assay" refers to a research (analytical) procedure in but not limited to laboratories, medicine, pharmacology, environmental biology, healthcare, and molecular biology - for but not limited to qualitatively evaluating or quantitatively measuring the presence, amount, concentration, or functional activity of a target entity (i.e., analyte). The analyte can be a drug, biochemical substance, or a cell in an organism or organic sample (such as human blood).

[0117] The term "image-based assay" refers to an assay procedure that utilizes an image of a sample taken by an imager, where the sample can be but not limited to medical, biological, and chemical samples.

[0118] The term "imager" refers to any device capable of taking an image of an object. It includes but not limited to cameras in microscopes, smart phones, or special devices that can take images at various wavelengths.

[0119] The term "sample feature" refers to some property of a sample that represents a condition of potential interest. In certain embodiments, the sample feature is a feature that appears in an image of the sample and can be segmented and classified by a machine learning model or some other algorithm. Examples of sample features include but not limited to the type of analyte in the sample, such as red blood cells, white blood cells, and tumor cells, and it includes analyte shape, count, size, volume, concentration, etc.

[0120] The term "microfeature in a sample" refers to an analyte, microstructure, and / or microchange of a substance in the sample. Analytes include particles, cells, macromolecules such as proteins, nucleic acids, and other moieties. Microstructures can refer to microscale differences in different materials. Microchanges refer to microscale changes in the local properties of the sample. Examples of microchanges are changes in local optical index and / or local mass. Examples of cells are blood cells, such as white blood cells, red blood cells, and platelets.

[0121] The term "defects in the sample" refers to foreign objects and artifacts that should not be present under ideal sample conditions or should not be considered in the sample. They can come from, but are not limited to, contaminants such as dust, bubbles, etc., and from peripheral objects, including structural objects in the sample, such as monitoring markers (e.g., columns) in the sample holding device. Defects in the sample can have significant dimensions and occupy a significant volume in the sample for determination, such as bubbles, where they can occur in different shapes, sizes, amounts, and concentrations in the sample, and they also have sample dependence depending on the sample.

[0122] The "morphological characteristics" of an analyte refer to the appearance (e.g., shape, color, size, etc.) and structure of the analyte.

[0123] The term "homography transformation" refers to a class of collinear transformations caused by the isomorphism of projective spaces. It is known in the field of image processing and is applied in camera models to characterize the image plane in the real world and the corresponding physical plane.

[0124] The term "machine learning" refers to algorithms, systems, and devices in the field of artificial intelligence that typically use statistical techniques and artificial neural networks to provide a computer with the ability to "learn" (i.e., gradually improve performance on a specific task) from data without being explicitly programmed.

[0125] The term "artificial neural network" refers to a hierarchically connected system inspired by biological networks that can "learn" by considering examples to perform tasks, usually without being programmed with any task-specific rules.

[0126] The term "convolutional neural network" refers to a class of multi-layer feedforward artificial neural networks most commonly applied to analyze visual images.

[0127] The term "deep learning" refers to a large class of machine learning methods in artificial intelligence (AI) that learn from data with a certain depth of network structure.

[0128] The term "machine learning model" refers to a trained computational model constructed from the training process in machine learning of data. The trained machine learning model is applied by a computer in the inference phase, which gives the computer the ability to perform specific tasks (e.g., detecting and classifying objects). Examples of machine learning models include ResNet, DenseNet, etc., which are also referred to as "deep learning models" due to the depth of their hierarchical network structure.

[0129] The term "image segmentation" refers to the image analysis process of dividing a digital image into multiple image patch segments (groups of pixels, usually with a set of bitmap masks covering the image segments enclosed by their segment boundary contours). Image segmentation can be achieved through image segmentation algorithms in image processing, such as watershed, grabcut, mean shift, etc., and it can also be achieved through dedicated machine learning algorithms, such as MaskRCNN, etc.

[0130] The term "unreliable" in the measurement result means that for measuring a given sample, the measurement result is not always accurate: sometimes the measured result is accurate, but sometimes the result is inaccurate, where the inaccurate result is substantially different from the accurate result. Such an inaccurate result is called an "error result". In some literature, an error result is also called an "outlier".

[0131] The term "accurate" in the measurement result means that within the allowed arrangement, the measurement result is consistent with the result of the same sample measured by a gold standard instrument operated by a trained professional in an ideal environment.

[0132] Traditionally, diagnostic measurements are usually carried out using complex (usually expensive) instruments and require highly trained personnel and complex infrastructure, which are not available in limited resource settings.

[0133] The term "limited resource setting" or "LRS" for measuring a sample refers to the setting during measurement where a simplified / low-cost measurement process or a simplified / low-cost instrument is used, carried out by an untrained person, in an adverse environment (such as an open and non-laboratory environment with dust), or any combination thereof.

[0134] The term "LRS measurement" refers to a measurement carried out under LRS.

[0135] The term "credible" in describing the reliability of a specific measurement result (or data) means that the reliability analysis of the specific measurement result determines that the result has a low probability of being inaccurate.

[0136] The term "non-credible" in describing the reliability of a specific measurement result (or data) means that the reliability analysis of the specific measurement result determines that the result has a high probability of being inaccurate.

[0137] The terms "worthiness" and "credible" are interchangeable.

[0138] The term "monitoring structure" means a structure used to monitor the operation of a measurement and / or the optical properties of the quality of a measurement device.

[0139] The "monitoring structure" includes monitoring marks, spacers, position marks, imaging marks, or scale marks.

[0140] The term "operating conditions" when making a measurement refers to the conditions under which the measurement is made. The operating conditions include, but are not limited to, at least three categories: (1) defects related to the sample, (2) defects related to the sample holder, and (3) defects related to the measurement process. The term "defect" means a deviation from the ideal conditions.

[0141] A1: Sample Defects in Image-Based Measurements

[0142] In image-based measurements, it involves sample preparation, a sample holding device for imaging, and an image of the sample taken by an imager on the sample holding device for measurement. During the measurement process, each operating action and component can produce defects that affect the accuracy of the measurement and the credibility of the measurement results.

[0143] For example, defects in sample preparation include, but are not limited to, air bubbles, dust, foreign objects (i.e., objects that should not be in the sample but have entered the sample), dry texture of the sample (where some parts of the sample have dried out in the sample holding device), insufficient amount of sample for measurement in the sample holding device, samples with incorrect matrices (e.g., blood, saliva), incorrect reaction of the reagent with the sample, non-uniform distribution of the sample, incorrect sample position in the sample holder (e.g., blood cells under the spacer), etc.

[0144] Examples of defects related to the sample holding device include, but are not limited to, missing spacers in the sample holder, the device not being properly closed, being placed in the incorrect position, being reopened after sealing, the surface being contaminated, incorrect spacer height, large surface roughness, incorrect transparency, incorrect absorptivity, missing monitoring marks, incorrect optical properties, incorrect electrical properties, incorrect geometric (size, thickness) properties, etc.

[0145] Further examples include defects from the measurement process, including but not limited to inappropriate light intensity, the sample not being in focus in the image for measurement, incorrect light color, leakage of ambient light, non-uniform luminous intensity, inappropriate lens conditions, filter conditions, optical component conditions, electrical component conditions, assembly conditions of the instrument, relative position of the sample holding device and the image plane of the imager, etc.

[0146] In addition, depending on the operating environment, analyte, reagents used, temperature, etc., many defects such as air bubbles and dust may occur unpredictably in image-based measurements, making it extremely difficult to completely eliminate them during the image-based measurement process. Therefore, a method and device are needed to detect, remove, and eliminate the negative impacts of defects in image-based measurements and make the measurement results accurate and credible.

[0147] Figure 7 is a block diagram of a process 100 for preparing a sample for measurement in a sample holding device having some intelligent monitoring structures for image-based measurement. In various implementations of process 100, actions may be removed, combined, or broken down into sub-actions. The process begins at action module 101, where a sample holding device is fabricated with high precision using nanoimprinting, where the device has intelligent monitoring structures (IMSs) for monitoring image-based measurement operations.

[0148] As shown in Figure 7, the IMSs in the sample holding device include columns fabricated on the substrate of the device. In some embodiments, these columns have uniform height and dimensions, and they are distributed in a pre-designed periodic pattern. In action module 102, the sample for measurement is dropped onto the sample contact area. The sample contact area may include reagents, antibodies, antigens, chemicals, etc. that interact with the sample and prepare the sample for measurement. In action module 103, the sample holding device of action module 102 is closed with a transparent cover, and the sample holding device containing the sample is sealed with surrounding walls and its two plates, where the sample for measurement is clamped between the two plates of the sample holding device. In action module 103, the sealed sample holding device is inserted into an adapter for imaging. The imager in action module 104 takes an image of the sample holding device (e.g., a QMAX card) in the adapter together with the intelligent monitoring structure on the sample holding contact area.

[0149] In action module 105, the image of the sample for measurement is saved for subsequent image-based measurement operations to analyze the properties of the analyte from the image of the sample. Figure 8 is an image of a complete blood count (CBC) of red blood cells taken by the imager in an image-based measurement.

[0150] A2: Defect Segmentation in Image-Based Measurement

[0151] In some embodiments for verifying the quality and credibility of measurement results, it is necessary to segment the object of interest from the image of the sample for measurement. Although machine learning-based image segmentation algorithms (e.g., Mask R-CNN) can be applied, they require precise contour marking of the shape of the object to be measured in the microscopic image of the sample in order to train the machine learning model, which is a bottleneck for many applications. For image-based measurement, this marking of the shape contour of the object can be expensive and difficult to achieve because the objects in the sample may be very small, their appearance is random, and moreover, there are significant variations in their shape, size, and color (e.g., dust, bubbles, etc.).

[0152] In some embodiments, a fine-grinned image segmentation algorithm is designed based on a combination of machine learning-based coarse bounding box segmentation and image processing-based fine grinding shape determination. It is applied to image segmentation in image-based assays, where each object only needs to be labeled in a coarse bounding box that does not depend on the shape and details of the shape contour. Thus, the need for fine-grinding labeling of the shape-related contours of the objects in the image of the sample is eliminated, which is difficult, complex, expensive, and difficult to be precise. The fine image segmentation algorithm includes:

[0153] a) Collecting a plurality of sample images taken by an imager for training, which include the objects to be detected in the image of the sample for the assay;

[0154] b) Labeling each object in the collected images with a coarse bounding box containing the object for model training;

[0155] c) Training a machine learning model (such as FRCNN) to detect the objects in the images of the samples with bounding boxes containing them;

[0156] d) In the inference phase, taking the image of the sample to be assayed as the input;

[0157] e) Applying the trained machine learning model to detect the objects and localize them using the bounding boxes in the image of the sample;

[0158] f) Transforming each image patch corresponding to the bounding box containing the detected object into gray, and then transforming it into binary using adaptive thresholding;

[0159] g) Performing morphological dilation (7x7) and erosion (3x3) from the background noise to enhance the contour of the shape;

[0160] h) Performing convex contour analysis on each of the image patches, and using the longest connected contour found in the patch as the contour of the object shape to determine the image mask of the object (for example, a binary bitmap covering the object in the image of the sample); and

[0161] i) Completing the image segmentation by collecting all the image masks from (h).

[0162] In some embodiments, in an image-based assay, the segmentation mask is applied to the detected object with an additional margin Δ to reduce the negative impact of the defect on the local adjacent region. Thus, the fine image segmentation described herein further includes:

[0163] 1) Using the margin Δ as a new mask to expand each detected contour in (h); and

[0164] 2) Image segmentation with a Δ margin is completed by collecting all magnified image masks from (1).

[0165] Figures 9-12 are application examples of applying the described fine image segmentation algorithm to image-based assays of red blood cells. As shown in the images, the described fine image segmentation method can handle objects with different sizes and shapes and produce a segmentation with a very tight mask covering the objects in the image of the sample, even when the general form of the shape and size of the defects in the image of the sample is unknown. This is completely different from the segmentation of analytes or cells in a sample (such as red blood cells in a complete blood count), where the general form of the shape or size of the analyte or cell is known or known within a certain degree of variation from its basic form.

[0166] Another aspect of the method for image segmentation of defects described herein can not only remove the defects detected in the image of the sample, but also remove the defects in the image of the sample by a controlled additional margin Δ. This ability is very important for assays because the defects in the sample used for assays can change the uniformity of the analyte distribution in the sample, change the local sample layer height, etc., resulting in large variations and inaccuracies in the assay results.

[0167] A3: Imaging-based assay using monitoring markers

[0168] In an image-based assay, the imager of the sample taken by the imager is used to determine at least one property of the analyte for the assay. And in some embodiments, an image of the sample is taken in a sample holding device with monitoring markers (such as the QMAX card described herein). Many factors can distort the image of the sample (i.e., different from the real sample or the image under perfect conditions). Image distortion can lead to inaccuracies and errors in determining the properties of the analyte in the assay. For example, one distortion is the change in the true lateral dimension (TLD) in the image of the sample taken by the imager.

[0169] The term "lateral dimension" refers to the linear dimension in the plane of the thin sample layer being imaged. In this document, the terms "true lateral dimension" and "field of view" are interchangeable.

[0170] Determining the TLD of the image of the sample is crucial in an image-based assay because it maps the size of the objects in the image of the sample described in pixels to their actual size in micrometers in the physical sample plane imaged by the imager. Once the TLD of the sample image is known, the actual size of the objects, such as actual length, area, etc., can be determined in an image-based assay.

[0171] A method of using a monitoring marker to estimate a TLD in an image-based assay is described in PCT / US19 / 46971, wherein structural features of the monitoring marker and machine learning are utilized to obtain a reliable TLD estimate under adverse conditions such as spherical and barrel lens distortion, strong light scattering and diffraction of particles in the sample, and focusing conditions of the imager.

[0172] In addition, PCT / US19 / 46971 also describes a method of using the structure of the monitoring marker and the properties of a sample holding device (such as a QMAX card) and machine learning to determine the remaining sample volume after removing certain objects from an image of a sample taken by an imager. These methods are applied in embodiments of the present invention to improve the accuracy of image-based assays and the credibility of assay results described herein.

[0173] A4: Removing defects in the sample in image-based assays

[0174] Defects in the sample can seriously affect the accuracy and credibility of assay results, where these defects can be any unwanted objects in the sample, including but not limited to dust, oil, etc., as shown in FIGS. 9 and 11. Once they enter the sample for assay, they are difficult to handle because their appearance and shape in the sample are random.

[0175] In some embodiments, a dedicated method for defect detection is designed and applied to image-based assays, where images of good samples without defects and samples with different degrees of defects in the sample are collected as training data, and the defect regions in the images are marked by bounding boxes. A machine learning model (such as Fast R-CNN) is selected and trained with the marked training images to detect defects in the sample images in the bounding boxes. Subsequently, the refined image segmentation method described in part A2 above is applied to determine the segmentation contour mask covering them. FIGS. 10 and 11 are images of segmented defects of samples from FIGS. 9 and 12 in a complete blood count of red blood cells using the fine grinding segmentation method described in part A2.

[0176] During the image-based assay process, defect detection and region determination are performed to verify the credibility of the assay results, including:

[0177] a) Taking the image of the sample from the imager as input;

[0178] b) Applying the trained defect detection machine learning model (such as Fast R-CNN) to the image of the sample for the assay to detect defects and locate them in the bounding boxes;

[0179] c) Using the refined image segmentation method described in part A2 to determine the segmentation contour mask of the detected defects;

[0180] d) Estimate the TLD of the image of the sample using the monitoring markers and the sample holding device described in part A3;

[0181] e) Using the estimated TLD from (d), determine the total area occupied by the defect (i.e., the defect area in the AoI) in the area of interest (AoI) for determination by summing all areas of the segmentation mask covering the detected defect;

[0182] f) Determine the area ratio between the defect area and the area of the AoI in the AoI for determination in the sample image:

[0183] Defect area ratio in AoI = Defect area in AoI / Area of the AoI for determination; and

[0184] g) If the defect area ratio in the AoI exceeds a specific threshold, raise a flag in the credibility of the determination result.

[0185] In some embodiments, the training of the machine learning model in (b) includes:

[0186] i. Collect multiple images of the measurement samples for model training;

[0187] ii. Mark the defects in the images of the training images of the samples from (i) with bounding boxes;

[0188] iii. Select a machine learning model in the form of a convolutional neural network, such as Fast RCNN;

[0189] iv. Adjust the parameters of the selected machine learning model for the loss function during training to classify the training samples into their correct identities by iterating through the multiple samples with bounding boxes marked with defects for training; and

[0190] v. When a certain stop condition is met, output the machine learning model from (iv).

[0191] In some embodiments, the threshold for determining the threshold based on the defect area ratio in the AoI in (g) is about 15%, where this threshold depends on the remaining sample volume after removing the sample volume corresponding to the area affected by the defect and where the sample volume estimation method of the image of the sample described in parts A2 and A3 and the constant height characteristic of the sample holding device can be used to obtain the sample volume after removing the defect.

[0192] In some embodiments, the object to be removed is the bubbles that appear in the sample for determination. In some embodiments, the object to be removed is dust, hair, etc., which enter the sample during the image-based measurement operation.

[0193] In some embodiments, the object to be removed and the associated volume can be the monitoring markers and volumes (e.g., columns) they assume in the sample holder, which are involved in the image of the sample for determination but should not be counted in the actual sample volume for determination. Otherwise, due to the incorrect amount of sample used, the determination results are inaccurate or unreliable.

[0194] A5: Removing aggregated analytes in the sample

[0195] In a determination, a troublesome situation is the aggregation of analytes in the sample for determination, which may affect the determination results and accuracy, such as counting, segmentation, etc.

[0196] In addition, analyte aggregation can be caused by operations in sample preparation, the type of analyte, the duration of the sample in the sample holding device, the freshness of the sample, the reagents used, etc. In addition, analyte aggregation is a moving process. Over time, it may get worse and should be checked step by step during the determination.

[0197] For example, in a complete blood count, red blood cells in certain parts of the sample for determination can aggregate, especially if they are exposed to open air for a certain period of time. The aggregated analyte clusters in the sample have some random sizes and shapes, which depend on how they aggregate together. If the portion of aggregated analytes exceeds a certain percentage in the sample, the sample should not be used for determination.

[0198] In some embodiments, a process using machine learning for analyte aggregation detection and removal is designed. It treats analyte aggregation / clustering as special defects in the sample for determination, which are dynamic and change over time. Similarly, it checks the image of the sample taken by the imager at each step of the determination to determine whether analytes are aggregating / clustering in the sample for determination.

[0199] In some embodiments, multiple sample images are collected to train a machine learning model to detect and segment analyte aggregation / clustering in the sample images. The collected training image set includes sample images without analyte clustering and sample images with different degrees of clustering. It follows a similar machine learning model training procedure described in Part A4, marking analyte aggregation / clustering in the sample images with bounding boxes. It trains the machine learning model based on the bounding box markings of the training data to detect analyte aggregation / clustering in the sample images during the inference process. After that, it applies the fine image segmentation method described in Part A2 to the detected bounding boxes containing analyte aggregation / clustering in the sample images.

[0200] During the determination, the detection and removal of analyte aggregation / clustering include:

[0201] a) Taking the image of the sample from the imager as input;

[0202] b) Apply the trained machine learning model to analyte aggregation / clustering detection to detect aggregated analytes in a sample image for determination in a bounding box;

[0203] c) Determine a fine segmentation contour mask of the detected analyte aggregation / clustering from (b) by applying the fine image segmentation method described in Part A2;

[0204] d) Determine the total area occupied by the aggregated analytes in the region of interest (AoI) in the image of the sample (aggregated analyte area in AoI) by summing all areas associated with the segmentation contour mask covering the region of interest (AoI) from (c);

[0205] e) Determine the area ratio between the aggregated analyte area in AoI and the area of AoI in the image of the sample:

[0206] Aggregated analyte area ratio in AoI = Aggregated analyte area in AoI / Area of AoI; and

[0207] f) If the aggregated analyte area proportion in AoI exceeds a specific threshold, raise a flag in the credibility of the determination result.

[0208] In some embodiments, the threshold is set to 10 - 20%, where the threshold depends on the impact of the aggregated analyte area on the final estimate, which can be estimated from the training and evaluation data.

[0209] A6: Remove bubbles in the sample for image - based determination

[0210] Bubbles in the sample for determination are a special type of defect, and they occur very frequently in the determination. Their appearance is random and can come from sample preparation, reactions between analytes and reagents in the sample, and other operating procedures.

[0211] In some embodiments of the present invention, it uses a dedicated process to handle bubbles (as special defects) to remove them in the determination. In some other embodiments, it adds bubbles as a new type of defect in the defect - removing procedure described in Part A4. Apply the fine image segmentation method described in Part A2 for fine segmentation and defect segmentation with an enlarged margin Δ to make the determination result accurate and credible.

[0212] In some embodiments, the process described in A4 is followed and the total area of the bubbles in the image of the sample is determined and the ratio of the bubble area in the region of interest for the determination is calculated: the bubble ratio in the AoI. If the bubble ratio in the AoI exceeds a certain threshold, a flag is raised in the credibility of the determination result. In some embodiments, the threshold for the bubbles is set to approximately 10%, where the threshold depends on the impact of the bubble area on the final estimate, which can be estimated based on training and evaluation data.

[0213] In some embodiments, a more stringent threshold for the bubbles is applied because a large area occupied by the bubbles is an indication of some chemical or biological reactions occurring between the components in the sample or some defects / problems in the sample holding device that require special attention for the determination.

[0214] A7: Removing the dry texture in the sample for image-based determination

[0215] In image-based determination, the dry texture in the sample image affects the accuracy and credibility of the determination result. This occurs when the amount of the sample for the determination is below the required amount or when some parts of the sample in the image holding device become dry due to some unpredictable factors.

[0216] In some embodiments, a machine learning-based process is designed and applied to detect the area of the dry texture in the image of the sample taken by the imager in image-based determination. It treats the dry texture in the sample for image-based determination as a special type of defect in the sample and follows the procedure described in section A4 for its detection and removal. Multiple samples are collected to train a machine learning model for detecting the dry texture in the sample for the determination of images of samples containing different degrees of dry texture area.

[0217] During image-based determination, the dry texture of the sample image is checked according to the procedure described in section A4. The machine learning-based dry texture detection is applied to check the image of the sample for the determination. The dry texture detected using the pre-trained machine learning model is located in a bounding box. The fine image segmentation method described in section A2 is applied to generate a fine segmentation of the detected dry texture. The total area of the detected dry texture is estimated using the estimated TLD (true lateral dimension) described in section A3 and summing all the masked areas of the detected dry texture.

[0218] In a certain embodiment, the ratio between the dry texture area in the AoI for the determination and the area of the AoI is calculated: dry texture area in AoI = dry texture area in AoI / area of AoI. And if the dry texture area ratio in the AoI exceeds a certain threshold, a flag is raised in the credibility of the determination result.

[0219] In some embodiments, the threshold is set to about 10%, where the threshold depends on the impact of the dry texture area on the final estimation result, which can be estimated from the training and evaluation data.

[0220] A8: Detecting an object on a column in an image-based assay

[0221] As described above, in some embodiments, a monitoring marker in the form of a column is used in the sample holding device to equalize the gap between two plates in an image-based assay. This uniformity of the gap in the sample holding device is crucial for controlling the sample volume used in the assay. However, if some analytes or foreign substances reach the top of the column and between the column and the top plate of the sample holding device, this may be disrupted. If this occurs, the gap of the sample holding device will not have a uniform height and will have an increased height around the column with analytes or foreign substances at the top.

[0222] In some embodiments, it detects the column in the sample image as in section A4, but it also uses a separately trained machine learning model to detect other defects and analytes in the image. In some embodiments, it constructs a large machine learning model as different object classes, which cover columns, analytes, bubbles, dust, etc. In the inference stage, it detects all objects belonging to each class. It applies the method of section A2 to segment the column from the image of the sample. Then it applies the method of section A2 to segment the objects of other detected analytes, dust, and bubbles. The method for detecting analytes, dust, and bubbles at the top of the column also includes:

[0223] a) Crossing the segmentation masks of the detected analytes, dust, and bubbles with the segmentation mask of the column in the sample image;

[0224] b) Identifying the affected columns whose segmentation masks have a non-empty intersection with the segmentation masks of analytes, dust, and bubbles;

[0225] c) If the number of affected columns exceeds a preset threshold, discard the image of the sample used for the assay;

[0226] d) If there are only a few affected columns in the image of the sample, remove the affected columns with a margin Δ in the image of the sample used for the assay according to the method described in section A4;

[0227] e) Update the sample volume by excluding from the sample image the sample volume corresponding to the removed affected columns with a margin Δ; and

[0228] f) Update the sample volume according to (d) and (e) and perform an image-based assay based on the updated assay sample.

[0229] A9: Detecting missing columns in the sample holding device

[0230] In some cases, the sample holding device may have its own defects. In particular, during the measurement operation, the monitoring markers (e.g., columns in a sample holding device such as a QMAX card) may be lost or broken. When this happens, the gap or sample thickness in the sample holding device can be changed, and the actual sample volume used for the measurement will be different.

[0231] In some embodiments of the present invention, it regards the monitoring markers (e.g., columns) as a special type of defect in the sample for the measurement. It uses the method described in Part A4 to detect the columns in the image of the sample, and determines whether some columns are missing in the sample holding device - an indication that the device itself has a defect. In addition, it uses the fine segmentation of Part A2 to identify the shape of the detected columns, and based on this, it detects the broken columns in the sample holding device, which also results from a defect in the sample holding device.

[0232] In some embodiments, according to the method and process described in Part A4, it combines the detection of objects on the opposing columns, the missing columns in the sample holding device, and the broken columns in the sample holding device in one process. In some embodiments, this becomes an important verification process to determine whether the sample holding device is properly closed in the measurement (e.g., there is no object on top of the columns), and whether the sample holding device has a defect (e.g., no missing columns or broken columns).

[0233] A10: Uniformity of analyte distribution in the sample

[0234] One factor affecting the credibility of the measurement result is the uniformity of analyte distribution in the sample, which is difficult to detect by visual inspection even for an experienced technician.

[0235] In some embodiments, a method for determining the uniformity of analyte distribution in the sample for the measurement based on the image of the sample for the measurement is designed and applied. It collects a plurality of sample images taken by the imager during the measurement for training an analyte detection machine learning model such as F - RCNN.

[0236] During the measurement process, the method for determining the uniformity of the analyte in the sample for the measurement includes:

[0237] a) Taking the image of the sample for the measurement as the input;

[0238] b) Dividing the image of the sample from (a) into equally sized and non - overlapping image patches (e.g., small image patches of 8x8 size);

[0239] c) Applying the trained machine learning model for analyte detection to each divided image patch from (b);

[0240] d) Determine the analyte concentration in each patch;

[0241] e) Classify the analyte concentrations of the image patches in ascending order based on (d), and determine the 25% quantile Q1 and its 75% quantile Q3 in the classified concentration sequence of the classified image patches in (b);

[0242] f) Apply an interquartile range (IQR)-based confidence measure to the concentration sequence from (e), and calculate the IQR-based confidence measure

[0243] Confidence - IQR = (Q3 - Q1) / (Q3 + Q1); and

[0244] If the confidence - IQR exceeds a certain preset threshold, a flag that the analyte is unevenly distributed in the image of the sample used for the determination is raised, and the determination result may be unreliable. In some embodiments, the threshold is set to 30%, where the threshold depends on the tolerance of the determination to the uneven distribution of the analyte in the sample. This tolerance of the determination to the uneven distribution of the analyte can be estimated from the training and evaluation of the determination data.

[0245] A12: Detect operational failures in image - based determinations

[0246] Operational failures are another source of inaccurate and unreliable determination results. One such failure is insufficient sample in the sample - holding device or sample spillage in the sample - holding device used for the determination.

[0247] In some embodiments, a dedicated sample - holding device (e.g., a QMAX card) is used to hold the sample for the determination, where the sample for the determination is clamped in the gap between two parallel plates of the sample - holding device, and the gap between the two plates in the sample - holding device is known and related to the analyte to be determined.

[0248] In some embodiments, the contour of the sample for the determination is detected in the image of the sample, and the total area enclosed by the sample contour in the image is estimated, where the total sample volume can be determined by the actual sample area in the image of the sample and the uniform height of the gap of the sample - holding device characterized by monitoring markers such as columns in the QMAX card. The actual area of the sample contour can be determined by the TLD (true lateral dimension) of the sample image using the method described in part A3. If the actual sample volume calculated from the image of the sample by removing the monitoring mask, columns, bubbles, dust, etc. is lower than the required volume for the determination, the determination process will be interrupted with an error message indicating insufficient sample for the determination.

[0249] Another operational malfunction that causes measurement inaccuracy is reopening and reclosing a closed sample holding device such as a QMAX card. One observation is that when the closed sample holding QMAX card is closed, the blood sample is reopened and then closed again, it will leave a recognizable blood contour mark on the bottom plate of the QMAX card, where when the top and bottom plates of the sample holding device are separated and closed again, a large amount of the sample blood tends to redistribute on the two sheets of the QMAX card.

[0250] When the sample holding device is closed again, part of the blood sample will redistribute and cover a slightly different area, leaving a visually recognizable contour mark. In addition, the newly generated sample area will not be exactly the same as the original covered area.

[0251] In some embodiments, it detects double boundaries in the image of the sample to determine whether the sample holding device has been reopened and then reclosed after the first closure.

[0252] In addition, sample spillage in the sample holding device is another type of operational malfunction in image-based assays. In some embodiments, it uses sample boundary detection in the image of the sample used for the assay to determine sample spillage in the sample holding device. Using the method described in part A2 that treats the entire image of the sample as a special large bounding box, sample boundary detection in the image of the sample is determined by image contour analysis.

[0253] In image-based assays, one type of malfunction that occurs during the assay operation is using an incorrect sample holding device with incorrect specifications for analysis. In some embodiments, it places a special code on the upper side of the plate of the sample holding device (such as a QMAX card) to specify the nature of the device and the intended assay use of the card, which can be imaged with the sample in the sample holding device by an imager. The special code can be recognized from the image of the sample to determine whether the correct sample holding device is used for the assay, and if the wrong card is used during the assay process, the assay process will stop.

[0254] In some embodiments, it uses the shape and distribution of monitoring marks (such as columns) in the sample holding device to encode the nature of the sample holding device (such as a QMAX card). In an embodiment, the method in part A4 is used to detect the columns in the sample image, determine their sizes and distribution patterns, thereby determining the nature of the sample holding device and its intended assay application, to eliminate the operational malfunction of using the wrong sample holding device for the assay.

[0255] A12: Focus Check in Image-Based Assays

[0256] In image-based assays, the image of the sample taken by the imager needs to be focused on the sample by the imager for the assay, and the out-of-focus in the image of the sample taken by the imager blurs the analyte in the image of the sample. As a result, the assay results become unreliable. However, there are many factors that can cause the image of the sample to be partially or even completely out of focus, including but not limited to vibrations and handshakes during image capture, misalignment of the sample holding device to the image sensor plane, etc. In addition, the prior art mainly relies on specific edge-content-based measurements, such as Tenengrad measurement, etc., with some preset content-dependent thresholds, which may be unreliable, fragile, and insufficient to meet the requirements of image-based assays.

[0257] In some embodiments, a machine learning-based verification process is designed and applied to determine whether the image of the sample taken by the imager is in focus or out of focus in an image-based assay. It collects multiple images of the sample for the assay taken by the imager under in-focus and out-of-focus conditions, and uses them as training data. The images in the collected training data are labeled based on their known focus conditions. A machine learning model such as DenseNet is selected and trained with the labeled training data. During the assay process, the trained machine learning model for classifying the focus condition in the sample image is applied, and it is detected whether the sample image taken by the imager is in focus during its inference process. Thus, it can be determined whether the image of the sample should be retaken to make the assay results more reliable.

[0258] In some embodiments, pre-designed monitoring marks on the sample holding device (such as a QMAX card) are utilized as references or anchors to assist in determining the focus quality in an image-based assay.

[0259] B1: Embodiments for improving the credibility of image-based assays

[0260] In an imaging-based assay for determining an analyte in a sample, an imager is used to create an image of the sample in a sample holder, and the image is used to determine the nature of the analyte.

[0261] However, many factors can distort the image (i.e., be different from the image of the real sample or under perfect conditions). Image distortion can lead to inaccurate determination of the analyte nature. For example, the fact that the focus is poor because biological samples themselves do not have sharp edges that are preferred to be in focus. When the focus is poor, the object size will be different from the real object, and other objects (such as blood cells) may become unrecognizable. Another example is that the lens may be perfect, resulting in different positions of the sample with different distortions. Another example is that the sample holder is not in the same plane as the optical imaging system, resulting in good focus in one area and poor focus in another area.

[0262] The present invention relates to devices and methods for obtaining a "true" image from a distorted image, thus improving the accuracy of measurements.

[0263] One aspect of the present invention is a device and method using a monitoring marker having an optically observable flat surface parallel to an adjacent surface.

[0264] Another aspect of the present invention is a device and method using a QMAX card to form a uniform layer of at least a portion of a sample and using a monitoring marker on the card to improve measurement accuracy.

[0265] Another aspect of the present invention is a device and method using a monitoring marker in conjunction with computational imaging, artificial intelligence, and / or machine learning.

[0266] The term "lateral dimension" refers to the linear dimension in the plane of a thin sample layer being imaged.

[0267] The terms "true lateral dimension (TLD)" and "field of view (FoV)" are interchangeable.

[0268] The term "microfeature in a sample" may refer to an analyte, microstructure, and / or microscale change in a substance in the sample. An analyte refers to particles, cells, macromolecules such as proteins, nucleic acids, and other moieties. A microstructure may refer to microscale differences in different materials. A microscale change refers to a microscale change in the local properties of the sample. Examples of microscale changes are changes in local optical index and / or local mass. Examples of cells are blood cells such as white blood cells, red blood cells, and platelets.

[0269] A. Monitoring Markers on a Solid Phase Surface

[0270] A1-1. A device for determining microfeatures in a sample using an imager, the device comprising:

[0271] (a) A solid phase surface comprising a sample contact area for contacting a sample containing microfeatures; and

[0272] (b) One or more monitoring markers, wherein the monitoring markers:

[0273] i. Are made of a material different from the sample;

[0274] ii. Are located inside the sample during determination of the microstructure, wherein the sample forms a thin layer having a thickness of less than 200 um on the sample contact area;

[0275] iii. Have a lateral linear dimension of about 1 um (micrometer) or greater, and

[0276] iv. Have at least one lateral linear dimension of 300 um or less; and

[0277] During the measurement process, the imager images at least one monitoring marker.

[0278] It is used during the determination of the analyte; and the geometric parameters (such as shape and size) of the monitoring marker and / or the spacing between the monitoring markers are (a) pre-determined and known before the analyte determination, and (b) used as parameters in an algorithm for determining properties related to the microfeatures.

[0279] A1-2. An apparatus for using an imager to determine microfeatures in a sample, the apparatus comprising:

[0280] A solid-phase surface comprising a sample contact area for contacting a sample containing microfeatures; and

[0281] One or more monitoring markers, each monitoring marker comprising a protrusion or groove from the solid-phase surface, wherein:

[0282] v. The protrusion or groove includes a flat surface substantially parallel to an adjacent surface, which is the part of the solid-phase surface adjacent to the protrusion or groove;

[0283] vi. The distance between the flat surface and the adjacent surface is about 200 micrometers (μm) or less;

[0284] vii. The flat surface has (a) a linear dimension of at least about 1 μm or greater, and (b) a region with at least one linear dimension of 150 μm or less;

[0285] viii. The flat surface of at least one monitoring marker is imaged by the imager used in the determination of the microfeatures; and

[0286] ix. The shape of the flat surface, the size of the flat surface, the distance between the flat surface and the adjacent surface, and / or the spacing between the monitoring markers are (a) predetermined and known before the determination of the microfeatures, and (b) used as parameters in an algorithm for determining properties related to the microfeatures.

[0287] B. Monitoring markers on the QMAX card

[0288] A2-1. An apparatus for using an imager to determine microfeatures in a sample, the apparatus comprising:

[0289] A first plate, a second plate, a spacer, and one or more monitoring markers, wherein:

[0290] i. The first plate and the second plate are movable relative to each other into different configurations;

[0291] ii. Each of the first plate and the second plate comprises an inner surface, and the inner surface comprises a sample contact area for contacting a sample containing microfeatures;

[0292] iii. One or both of the first plate and the second plate include spacers permanently fixed to the inner surface of the respective plate,

[0293] iv. The spacers have a substantially uniform height and a fixed spacing distance (ISD) equal to or less than 200 micrometers;

[0294] v. The monitoring markers are made of a material different from that of the sample;

[0295] vi. During the determination of the microstructure, the monitoring markers are located inside the sample, where the sample forms a thin layer with a thickness less than 200 um on the sample contact area; and

[0296] vii. The monitoring markers have a lateral linear dimension of about 1 um (micrometer) or greater and at least one lateral linear dimension of 300 um or less;

[0297] wherein during the determination process, the imager images at least one monitoring marker

[0298] which is used during the determination of the microfeatures; and the shape, size, distance between the flat surface and the adjacent surface, and / or the spacing between the monitoring markers (a) are predetermined and known before the determination of the microfeatures, and (b) are used as parameters in the algorithm for determining the characteristics related to the microfeatures;

[0299] wherein one configuration is an open configuration, in which: the two plates are partially or completely separated, the spacing between the plates is not adjusted by the spacers, and the sample is deposited on one or both of the plates;

[0300] wherein the other configuration is a closed configuration, which is configured after the sample is deposited in the open configuration and the plates are forced to reach the closed configuration by applying an imprecise pressing force on the force application area; and in the closed configuration: at least a part of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, wherein the uniform thickness of the layer is defined by the sample contact areas of the two plates and is adjusted by the plates and the spacers; and

[0301] wherein the monitoring markers are (i) a structure different from the spacers, or (ii) the same structure as the spacers used.

[0302] A2-2. An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0303] A first plate, a second plate, spacers, and one or more monitoring markers, wherein:

[0304] viii. The first plate and the second plate are movable relative to each other into different configurations;

[0305] ix. Each of the first and second plates includes an inner surface that includes a sample contact region for contacting a sample containing microfeatures;

[0306] x. One or both of the first and second plates include spacers permanently fixed to the inner surface of the respective plate,

[0307] xi. The spacers have a substantially uniform height and a fixed spacing distance (ISD) that is equal to or less than 200 micrometers;

[0308] xii. Each monitoring mark includes a protrusion or a groove on one or both of the sample contact regions;

[0309] xiii. The protrusion or groove includes a flat surface that is substantially parallel to an adjacent surface, which is the portion of the solid phase surface adjacent to the protrusion or groove;

[0310] xiv. The distance between the flat surface and the adjacent surface is about 200 micrometers (um) or less;

[0311] xv. The flat surface has (a) a linear dimension of at least about 1 um or greater, and (b) a region with at least one linear dimension of 150 um or less;

[0312] xvi. The flat surface of at least one monitoring mark is imaged by an imager used in determining the microfeatures; and

[0313] xvii. The shape of the flat surface, the size of the flat surface, the distance between the flat surface and the adjacent surface, and / or the spacing between the monitoring marks (a) are predetermined and known prior to the determination of the microfeatures, and (b) are used as parameters in an algorithm for determining a property associated with the microfeatures.

[0314] One of the configurations is an open configuration, in which: the two plates are partially or fully separated, the spacing between the plates is not adjusted by the spacers, and the sample is deposited on one or both of the plates;

[0315] The other of the configurations is a closed configuration, which is configured after the sample is deposited in the open configuration and the plates are forced to the closed configuration by applying an imprecise pressing force on the force application area; and in the closed configuration: at least a portion of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, where the uniform thickness of the layer is defined by the sample contact regions of the two plates and is adjusted by the plates and the spacers; and

[0316] where the monitoring marks are (i) a structure different from the spacers, or (ii) the same structure as the spacers used.

[0317] A3. An apparatus for imaging-based assays, comprising:

[0318] An apparatus as in any of the preceding apparatus embodiments, wherein the apparatus has at least five monitoring markers, and at least three of the monitoring markers are not aligned on a straight line.

[0319] A4. An apparatus for determining an analyte in a sample using an imager, the system comprising:

[0320] (a) An apparatus as in any of the preceding apparatus embodiments; and

[0321] (b) An imager for determining a sample containing the analyte.

[0322] A5. A system for performing imaging-based assays, the system comprising:

[0323] (a) An apparatus as in any of the preceding apparatus embodiments;

[0324] (b) An imager for determining a sample containing the analyte; and

[0325] (c) An algorithm for determining characteristics related to the analyte using the monitoring markers of the apparatus.

[0326] In some embodiments, the thickness of the thin layer is configured such that for a given analyte concentration, there is a monolayer of the analyte in the thin layer. The term "monolayer" means that in a thin sample layer, there is substantially no overlap between two adjacent analytes in a direction perpendicular to the plane of the sample layer.

[0327] C. Monitoring markers with computational imaging artificial intelligence and / or machine learning

[0328] Another aspect of the present invention is the combination of monitoring markers with computational imaging, artificial intelligence, and / or machine learning. It utilizes the process of forming an image based on measurements, uses algorithms to process the image, and maps the objects in the image to their physical dimensions in the real world. Machine learning (ML) is applied in the present invention to learn the significant features of the objects in the sample that are embedded in the ML model and constructed and trained from the images of the sample taken by the imager. During the inference process of the present invention, intelligent decision logic is constructed and applied to detect and classify the target objects in the sample based on the knowledge embedded in the ML model. Computational imaging is a process of indirectly forming an image from measurements using algorithms that rely on a large amount of computation.

[0329] A6. A system for determining an analyte in a sample using an imager, the system comprising:

[0330] (a) An apparatus as in any of the preceding apparatus embodiments;

[0331] (b) An imager for assaying a sample containing an analyte; and

[0332] (c) An algorithm that uses the monitoring markers of the device to assay a characteristic related to the analyte, wherein the algorithm uses machine learning.

[0333] A7. A method for assaying an analyte in a sample using an imager, comprising:

[0334] (a) Obtaining a device, apparatus, or system as in any of the foregoing embodiments;

[0335] (b) Obtaining a sample and depositing the sample on a sample contact area in the device, apparatus, or system of (a), wherein the sample contains an analyte; and

[0336] (c) Assaying the analyte.

[0337] 8. A method for assaying an analyte in a sample using an imager, comprising:

[0338] (a) Obtaining a device, apparatus, or system as in any of the foregoing embodiments;

[0339] (b) Obtaining a sample and depositing the sample on a sample contact area in the device, apparatus, or system of (a), wherein the sample contains an analyte;

[0340] (c) Assaying the analyte, wherein the assay comprises a step of using machine learning.

[0341] A key idea of the present invention is to use columns as detectable anchors for calibration in a sample holding device (such as a QMAX device) and improve the accuracy of image-based assays. In a QMAX device, the columns are monitoring markers to keep the gap between two plates that hold the sample in the sample holding device uniform. However, since the columns are penetrated and surrounded by the analyte within the sample holding device, it is a challenge to accurately detect the columns as anchors for calibration in the sample holding device and improve the accuracy of the assay. Moreover, in microscopy imaging, due to spherical (barrel) distortion of the lens, light diffraction from the microscopic object, defects at the microscopic level, misalignment in focusing, noise in the sample image, etc., their images are distorted and blurred. And it becomes even more difficult if the imaging is performed by a consumer device (e.g., a camera from a smartphone), because these cameras are not calibrated by dedicated hardware once they leave the factory.

[0342] In the present invention, column detection is formulated as a machine learning framework - to detect columns in a sample holding device (such as a QMAX device) - with an accuracy suitable for calibration and an improvement in accuracy in image-based assays. Since the distribution and physical configuration of the columns are a priori known and controlled by fine nano-scale manufacturing (such as in a QMAX device), this innovative approach of using detectable monitoring markers (such as columns) as anchors in image-based assays is not only feasible but also effective.

[0343] In some embodiments, the algorithms of any of the foregoing embodiments include algorithms for computational imaging, artificial intelligence, and / or machine learning.

[0344] In some embodiments, the algorithms of any of the foregoing embodiments include machine learning algorithms.

[0345] In some embodiments, the algorithms of any of the foregoing embodiments include artificial intelligence and / or machine learning algorithms.

[0346] In some embodiments, the algorithms of any of the foregoing embodiments include computational imaging and / or machine learning algorithms.

[0347] Embodiments of the present invention include:

[0348] (1) Using a sample loading device (such as a QMAX device) in an image-based assay, where there are monitoring markers with known structures located in the device, the monitoring markers are not submerged in the sample and can be imaged from the top by an imager in the image-based assay;

[0349] (2) Taking an image of the sample in the sample loading device including the analyte and the monitoring markers;

[0350] (3) Constructing and training a machine learning (ML) model for detecting monitoring markers in the sample holding device from the images taken by the imager;

[0351] (4) Using the ML detection model from (3) to detect and locate the monitoring markers in the sample image taken by the imager in the sample loading device;

[0352] (5) Generating a marker grid based on the monitoring markers detected in (4);

[0353] (6) Calculating a homography transformation based on the generated monitoring marker grid; and

[0354] (7) Estimating the TLD and determining the area, size, and concentration of the imaged analyte in the image-based assay.

[0355] The present invention can be further improved for region-based TLD estimation and calibration, thereby improving the accuracy of image-based assays. Embodiments of such a method include:

[0356] (1) Using a sample loading device (e.g., a QMAX device) in an image-based assay, where there is a monitoring marker - not submerged in the sample and residing in a device that can be imaged from the top by an imager in the image-based assay;

[0357] (2) Taking an image of the sample in a sample holding device that includes the analyte and the monitoring marker;

[0358] (3) Constructing and training a machine learning (ML) model for detecting the monitoring marker in the sample holding device from the image taken by the imager;

[0359] (4) Dividing the sample image taken by the imager into non-overlapping regions;

[0360] (5) Using the ML model of (3) to detect and locate the monitoring marker from the sample image taken by the imager;

[0361] (6) Generating a region-based marker grid for each region in which more than 5 non-collinear monitoring markers are detected in the local region;

[0362] (7) Generating a marker grid for all regions not in (6) based on the monitoring markers detected from the image of the sample taken by the imager;

[0363] (8) Calculating a region-specific homography transformation for each region in (6) based on its own region-based marker grid generated in (6);

[0364] (9) Calculating a homography transformation for all other regions based on the marker grid generated in (7);

[0365] (10) Estimating the region-based TLD for each region in (6) based on the region-based homography transformation generated in (8);

[0366] (11) Estimating the TLD for other regions based on the homography transformation from (9); and

[0367] (12) Applying the estimated TLDs from (10) and (12) to determine the area and concentration of the analyte imaged in each partition in the image-based assay.

[0368] In some embodiments, the monitoring marker has sharp edges and a flat surface.

[0369] In some embodiments, the monitoring markers are used to determine local properties of the image and / or local operating conditions (e.g., gap size, plate quality).

[0370] In some embodiments, the monitoring markers have the same shape as the spacer.

[0371] Monitoring measurements using the monitoring markers

[0372] One aspect of the present invention is for performing measurements using a QMAX card having two movable plates. Monitoring markers placed within a thin sample can be used to monitor the operating conditions of the QMAX card. The operating conditions can include whether the sample is loaded correctly, whether the two plates are closed correctly, and whether the gap between the two plates is the same as or approximately the same as a predetermined value.

[0373] In some embodiments, for a QMAX card that includes two movable plates and has a predetermined gap between the two plates in a closed configuration, the operating conditions of the QMAX measurement are monitored by taking an image of the monitoring markers in the closed configuration. For example, if the two plates are not closed correctly, the monitoring markers will appear differently in the image compared to if the two plates are closed correctly. Monitoring markers surrounded by the sample will have a different appearance than those not surrounded by the sample. Thus, it can provide information about the sample loading conditions.

[0374] Z-1.1 An apparatus for monitoring the operating conditions of a device using monitoring markers, the apparatus comprising:

[0375] A first plate, a second plate, a spacer, and one or more monitoring markers, wherein:

[0376] i. The first plate and the second plate are movable relative to each other into different configurations;

[0377] ii. Each of the first plate and the second plate includes an inner surface that includes a sample contact area for contacting the sample to be analyzed;

[0378] iii. One or both of the first plate and the second plate include a spacer permanently fixed to the inner surface of the respective plate,

[0379] iv. The monitoring markers have at least one of (a) a predetermined and known, and (b) a size observable by an imager;

[0380] v. The monitoring markers are microstructures having at least one lateral linear dimension of 300 um or less; and

[0381] vi. The monitoring markers are located inside the sample;

[0382] One of the configurations is an open configuration, in which: the two plates are partially or completely separated, the spacing between the plates is not adjusted by spacers, and the sample is deposited on one or both of the plates;

[0383] Another of the configurations is a closed configuration, which is configured after the sample is deposited in the open configuration and the plates are forced to the closed configuration by applying an imprecise pressing force on the force application area; and in the closed configuration: at least a portion of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, where the uniform thickness of the layer is defined by the sample contact areas of the two plates and is adjusted by the plates and spacers; and

[0384] After a force is used to bring the two plates to the closed configuration, the monitoring mark is imaged to determine (i) whether the two plates have reached the desired closed configuration, thereby adjusting the sample thickness to a substantially predetermined thickness, and / or (ii) determining whether the sample has been loaded as needed.

[0385] In some embodiments, an image of the monitoring mark is used to determine whether the two plates have reached the desired closed configuration, where the sample is adjusted to have a thickness of approximately the predetermined thickness.

[0386] In some embodiments, an image of the monitoring mark is used to determine whether the sample has been loaded as needed.

[0387] In some embodiments, the monitoring mark is imaged to determine whether the two plates have reached the desired closed configuration in which the sample thickness is adjusted to the predetermined thickness, and to determine whether the sample has been loaded as needed.

[0388] In some embodiments, the spacer serves as the monitoring mark.

[0389] In some embodiments, the system includes a device and a computing device and a non-transitory computer-readable medium having instructions that, when executed, perform the determination.

[0390] In some embodiments, a non-transitory computer-readable medium having instructions that, when executed, perform a method that includes using one or more images of the thin sample layer and the monitoring mark to determine (i) whether the two plates have reached the desired closed configuration, thereby adjusting the sample thickness to a substantially predetermined thickness, or (ii) whether the sample has been loaded as needed.

[0391] In some embodiments, the system includes a non-transitory computer-readable medium having instructions that, when executed, perform any method of the present disclosure.

[0392] W-1. A method for monitoring the operating conditions of a device using monitoring markers, the method comprising:

[0393] (a) obtaining a device as in any of the foregoing embodiments, wherein the device comprises two movable plates, spacers, and one or more monitoring markers, wherein the monitoring markers are in the sample contact area;

[0394] (b) obtaining an imager;

[0395] (c) depositing a sample in the sample contact area of the device of (a) and forcing the two plates into a closed configuration;

[0396] (d) using the imager to take one or more images of the thin sample layer and the monitoring markers; and

[0397] (e) using the images of the monitoring markers to determine (i) whether the two plates have reached an expected closed configuration, thereby adjusting the sample thickness to a substantially predetermined thickness, or (ii) whether the sample has been loaded as required.

[0398] In some embodiments, the images of the monitoring markers are used to determine whether the two plates have reached an expected closed configuration, wherein the sample is adjusted to have a thickness of approximately a predetermined thickness.

[0399] In some embodiments, the images of the monitoring markers are used to determine whether the sample has been loaded as required.

[0400] In some embodiments, the monitoring markers are imaged to determine whether the two plates have reached an expected closed configuration in which the sample thickness is adjusted to a predetermined thickness, and to determine whether the sample has been loaded as required.

[0401] In some embodiments, the system comprises a device and a computing device and a non-transitory computer-readable medium having instructions that, when executed, perform the determination.

[0402] Selecting a region of interest and / or removing defective image regions

[0403] In some embodiments, the sample has defects, and a method for removing the influence of the defects on the determination comprises: identifying the defects in the image, taking an image of the defects, or selecting a good region in the image that does not have an image caused by the defects.

[0404] In some embodiments, the region taken from the image to be removed is larger than the defective image region.

[0405] In some embodiments, the thickness of the sample is configured to be a thin thickness such that the objects of interest (e.g., cells) form a single layer (i.e., there is no significant overlap between the objects in the direction perpendicular to the sample layer).

[0406] A method for determining the manufacturing quality of a QMAX card using an imager, the method comprising:

[0407] (f) Obtaining a device as in any of the foregoing embodiments, wherein the device comprises two movable plates, spacers, and one or more monitoring marks, wherein the monitoring marks are in the sample contact area;

[0408] (g) Obtaining an imager;

[0409] (h) Depositing a sample in the sample contact area of the device of (a) and forcing the two plates into a closed configuration;

[0410] (i) Using the imager to take one or more images of the thin sample layer; and

[0411] (j) Determining the manufacturing quality of the QMAX card using the images of the monitoring marks.

[0412] A method for determining the manufacturing quality of a QMAX card using an imager, the method comprising:

[0413] (a) Obtaining a device as in any of the foregoing embodiments, wherein the device comprises two movable plates, spacers, and one or more monitoring marks, wherein the monitoring marks are in the sample contact area;

[0414] (b) Obtaining an imager;

[0415] (c) Depositing a sample in the sample contact area of the device of (a) and forcing the two plates into a closed configuration;

[0416] (d) Using the imager to take one or more images of the thin sample layer and the monitoring marks; and

[0417] (e) Determining the manufacturing quality of the QMAX card using the images of the monitoring marks.

[0418] A method as in any of the foregoing embodiments, wherein determining the manufacturing quality comprises measuring one or more characteristics of the monitoring marks (e.g., length, width, spacing, thick edge), and comparing the measured characteristics with a reference value to determine the manufacturing quality of the QMAX card.

[0419] A method as in any of the foregoing embodiments, wherein determining the manufacturing quality comprises measuring a first characteristic of one or more first monitoring marks (e.g., quantity, length, width, spacing, thick edge), and comparing the measured first characteristic with a second characteristic of one or more second monitoring marks (e.g., number, length, width, spacing, thick edge) to determine the manufacturing quality of the QMAX card..

[0420] The method of any of the foregoing embodiments, wherein the determination is performed during the analysis of a sample using the apparatus of any of the foregoing embodiments.

[0421] Another aspect of the invention is to have the monitoring markers in a periodic pattern in a sample holding device such as a QMAX device, such that they periodically appear at a certain spacing in the image of the sample taken by the imager. Based on this periodic characteristic, since all the monitoring markers can be identified and derived from only a few detected monitoring markers because they are periodically positioned in a predetermined configuration, and moreover, such a configuration can be precisely fabricated using nanofabrication techniques such as nanoimprinting, the monitoring marker detection can become very reliable. And thereby, due to the periodic pattern of the monitoring markers, both the sample image-based and image region-based TLD estimation can become more accurate and robust.

[0422] Some examples

[0423] Single board

[0424] AA-1.1 An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0425] (a) A solid-phase surface that includes a sample contact area for contacting a thin sample having a thickness of 200 um or less and that includes or is suspected of including microfeatures; and

[0426] (b) One or more markers, wherein the markers:

[0427] x. Have sharp edges that (i) have a predetermined and known shape and size and (ii) are observable by an imager for imaging the microfeatures;

[0428] xi. Are microstructures having at least one lateral linear dimension of 300 um or less; and

[0429] xii. Are located inside the sample;

[0430] Wherein at least one of the markers is imaged by the imager during the determination.

[0431] AA-1.2 An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0432] (a) A solid-phase surface that includes a sample contact area for contacting a thin sample having a thickness of 200 um or less and that includes or is suspected of including microfeatures; and

[0433] (b) One or more markers, wherein the markers:

[0434] i. Include protrusions or grooves from the solid-phase surface

[0435] ii. having a sharp edge, which (i) has a predetermined and known shape and size, and (ii) can be observed by an imager for imaging microfeatures;

[0436] iii. being a microstructure with at least one lateral linear dimension of 300 um or less; and

[0437] iv. being located inside the sample;

[0438] wherein at least one of the markers is imaged by the imager during the determination.

[0439] Two plates with a constant spacing

[0440] AA-2.1 An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0441] A first plate, a second plate, and one or more monitoring markers, wherein:

[0442] xviii. Each of the first plate and the second plate includes an inner surface that includes a sample contact area for contacting a sample containing or suspected of containing microfeatures;

[0443] xix. At least a portion of the sample is defined by the first and second plates as a thin layer having a substantially constant thickness of 200 um or less;

[0444] xx. The monitoring marker has a sharp edge, which (a) has a predetermined and known shape and size, and (b) can be observed by an imager for imaging microfeatures;

[0445] xxi. The monitoring marker is a microstructure with at least one lateral linear dimension of 300 um or less; and

[0446] xxii. The monitoring marker is located inside the sample;

[0447] wherein at least one of the markers is imaged by the imager during the determination.

[0448] Two movable plates

[0449] AA-3 An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0450] A first plate, a second plate, a spacer, and one or more monitoring markers, wherein:

[0451] vii. The first plate and the second plate are movable relative to each other into different configurations;

[0452] viii. Each of the first and second plates includes an inner surface, and the inner surface includes a sample contact area for contacting a sample containing or suspected of containing microfeatures;

[0453] ix. One or both of the first and second plates include spacers permanently fixed on the inner surfaces of the respective plates,

[0454] x. The monitoring mark has a sharp edge, which (a) has a predetermined and known shape and size, and (b) can be observed by an imager for imaging microfeatures;

[0455] xi. The monitoring mark is a microstructure with at least one lateral linear dimension of 300 μm or less; and

[0456] xii. The monitoring mark is located inside the sample;

[0457] wherein at least one of the marks is imaged by the imager during the assay.

[0458] One configuration is an open configuration, in which: the two plates are partially or fully separated, the spacing between the plates is not adjusted by the spacers, and the sample is deposited on one or both of the plates;

[0459] The other configuration is a closed configuration, which is configured after the sample is deposited in the open configuration and the plates are forced to reach the closed configuration by applying an imprecise pressing force on the force application area; and in the closed configuration: at least a part of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, wherein the uniform thickness of the layer is defined by the sample contact areas of the two plates and is adjusted by the plates and the spacers; and

[0460] wherein the monitoring mark is (i) a structure different from the spacer, or (ii) the same structure used as the spacer.

[0461] B. Improvement of image capture using a sample holder with micro marks

[0462] BB-1. An apparatus for improving image capture of microfeatures in a sample, the apparatus comprising:

[0463] (c) An apparatus as in any of the preceding apparatus embodiments; and

[0464] (d) An imager for assaying a sample containing or suspected of containing microfeatures;

[0465] wherein the imager captures an image, and at least one image includes a part of the sample and the monitoring mark both.

[0466] CB-2. A system for improving image capture of microfeatures in a sample, the system comprising:

[0467] (d) A device as in any of the preceding device embodiments;

[0468] (e) An imager for determining a sample containing or suspected of containing microfeatures; and

[0469] (f) An algorithm that uses the marker as a parameter to adjust the settings of the imager for the next image together with an imaging processing method.

[0470] C. Imaging analysis using a sample holder with micro - markers

[0471] CC - 1 An apparatus for improving the analysis of an image of microfeatures in a sample, the apparatus comprising:

[0472] (a) A device as in any of the preceding device embodiments; and

[0473] (b) A computing device for receiving an image of a marker and a sample containing or suspected of containing microfeatures;

[0474] wherein the computing device runs an algorithm that uses the marker as a parameter together with an imaging processing method to improve the image quality in the image.

[0475] CC - 2. A system for improving the analysis of an image of microfeatures in a sample, the system comprising:

[0476] (a) A device as in any of the preceding device embodiments;

[0477] (b) An imager for determining a sample containing or suspected of containing microfeatures by taking one or more images of the sample and the marker; and

[0478] (c) An algorithm that uses the marker as a parameter together with an imaging processing method to improve the image quality of at least one image taken in (c).

[0479] CC - 3 A computer program product for determining microfeatures in a sample, the program comprising computer program code means for applying to and adapted for at least one image:

[0480] (a) Receiving an image of the sample and a monitoring marker, wherein the sample is loaded into the device of any of the preceding device claims, and wherein the image is taken by an imager; and

[0481] (b) Processing and analyzing the image to calculate the quantity of the microfeatures, wherein the analysis uses a machine - learning - based detection model and information provided by the image of the monitoring marker.

[0482] CC-4 A computing device for analyzing micro-features in a sample, the computing device comprising a computing device that operates the algorithms in any of the embodiments of the present invention.

[0483] CC-5 A method, apparatus, computer program product, or system as in any of the preceding embodiments, wherein the improvement in image quality comprises at least one selected from the group consisting of: denoising, image normalization, image sharpening, image scaling, alignment (e.g., for face detection), super-resolution, deblurring, and any combination thereof.

[0484] CC-6 A method, apparatus, computer program product, or system as in any of the preceding embodiments, wherein the imaging processing method comprises at least one selected from the group consisting of: histogram-based operations, mathematics-based operations, convolution-based operations, smoothing operations, derivative-based operations, morphology-based operations, shadow correction, image enhancement and / or restoration, segmentation, feature extraction and / or matching, object detection and / or classification and / or localization, image understanding, and any combination thereof.

[0485] CC-6.1 A method, apparatus, computer program product, or system as in any of the preceding embodiments, wherein the histogram-based operations comprise at least one selected from the group consisting of: contrast stretching, equalization, minimum filtering, median filtering, maximum filtering, and any combination thereof.

[0486] CC-6.2 A method, apparatus, computer program product, or system as in any of the preceding embodiments, wherein the mathematics-based operations comprise at least one selected from the group consisting of: binary operations (e.g., NOT, OR, AND, XOR, and SUB), arithmetic-based operations (e.g., ADD, SUB, MUL, DIV, LOG, EXP, SQRT, TRIG, and INVERT), and any combination thereof.

[0487] CC-6.3 A method, apparatus, computer program product, or system as in any of the preceding embodiments, wherein the convolution-based operations comprise at least one selected from the group consisting of: operations in the spatial domain, Fourier transform, DCT, integer transform, operations in the frequency domain, and any combination thereof.

[0488] CC-6.4 A method, apparatus, computer program product, or system as in any of the preceding embodiments, wherein the smoothing operations comprise at least one selected from the group consisting of: linear filtering, uniform filtering, triangular filtering, Gaussian filtering, non-linear filtering, median filtering, kuwahara filtering, and any combination thereof.

[0489] CC-6.5 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the derivative-based operation is selected from at least one of the group consisting of: first derivative operation, gradient filtering, basic derivative filtering, Prewitt gradient filtering, Sobel gradient filtering, alternative gradient filtering, Gaussian gradient filtering, second derivative filtering, basic second derivative filtering, Laplacian in the frequency domain, Gaussian second derivative filtering, alternative Laplacian filtering, second derivative of gradient direction (SDGD) filtering, third derivative filtering, higher-order derivative filtering (e.g., filtering greater than the third derivative), and any combination thereof.

[0490] CC-6.6 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the morphology-based operation includes at least one selected from the group consisting of: dilation, erosion, Boolean convolution, opening and / or closing, hit-or-miss operation, contour, skeleton, propagation, gray value morphology, gray-level dilation, gray-level erosion, gray-level opening, gray-level closing, morphological smoothing, morphological gradient, morphological Laplacian, and any combination thereof.

[0491] CC-6.7 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the image enhancement and / or restoration includes at least one selected from the group consisting of: sharpening, de-sharpening, noise suppression, distortion suppression, and any combination thereof.

[0492] CC-6.8 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the segmentation includes at least one selected from the group consisting of: thresholding, fixed thresholding, histogram-derived thresholding, Isodata algorithm, background symmetry algorithm, triangle algorithm, edge finding, gradient-based process, zero-crossing-based process, PLUS-based process, binary mathematical morphology, salt-and-pepper filtering, separating objects with holes, filling holes in objects, removing boundary-touching objects, outer skeleton, touching objects, gray value mathematical morphology, Top-hat transform, thresholding, local contrast stretching, and any combination thereof.

[0493] CC-6.9 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the feature extraction and / or matching includes at least one selected from the group consisting of: independent component analysis, isometric mapping, kernel principal component analysis, latent semantic analysis, partial least squares, principal component analysis, multi-factor dimensionality reduction, non-linear dimensionality reduction, multilinear principal component analysis, multilinear subspace learning, semidefinite embedding, autoencoder, and any group thereof.

[0494] A. Sample holder with micro-markings

[0495] Single board

[0496] AA-1.1 An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0497] (a) A solid-phase surface that includes a sample contact area for contacting a thin sample having a thickness of 200 μm or less and that includes or is suspected of including microfeatures; and

[0498] (b) One or more markers, wherein the markers:

[0499] xiii. Have sharp edges that (i) have a predetermined and known shape and size and (ii) are observable by an imager that images the microfeatures;

[0500] xiv. Are microstructures having at least one lateral linear dimension of 300 μm or less; and

[0501] xv. Are located inside the sample;

[0502] wherein at least one of the markers is imaged by the imager during the determination.

[0503] AA-1.2 An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0504] (a) A solid-phase surface that includes a sample contact area for contacting a thin sample having a thickness of 200 μm or less and that includes or is suspected of including microfeatures; and

[0505] (b) One or more markers, wherein the markers:

[0506] v. Include protrusions or grooves from the solid-phase surface

[0507] vi. Have sharp edges that (i) have a predetermined and known shape and size and (ii) are observable by an imager that images the microfeatures;

[0508] vii. Are microstructures having at least one lateral linear dimension of 300 μm or less; and

[0509] viii. Are located inside the sample;

[0510] wherein at least one of the markers is imaged by the imager during the determination.

[0511] Two plates with a constant spacing

[0512] AA-2.1 An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0513] A first plate, a second plate, and one or more monitoring markers, wherein:

[0514] xxiii. Each of the first plate and the second plate includes an inner surface that includes a sample contact region for contacting a sample that includes or is suspected of including microfeatures;

[0515] xxiv. At least a portion of the sample is defined by the first and second plates as a thin layer having a substantially constant thickness of 200 μm or less;

[0516] xxv. The monitoring mark has a sharp edge that (a) has a predetermined and known shape and size, and (b) can be observed by an imager that images the microfeatures;

[0517] xxvi. The monitoring mark is a microstructure having at least one transverse linear dimension of 300 μm or less; and

[0518] xxvii. The monitoring mark is located inside the sample;

[0519] wherein at least one of the marks is imaged by the imager during the assay.

[0520] Two movable plates

[0521] AA-3 An apparatus for assaying microfeatures in a sample using an imager, the apparatus comprising:

[0522] A first plate, a second plate, a spacer, and one or more monitoring marks, wherein:

[0523] xiii. The first plate and the second plate are movable relative to each other into different configurations;

[0524] xiv. Each of the first plate and the second plate includes an inner surface that includes a sample contact region for contacting a sample that includes or is suspected of including microfeatures;

[0525] xv. One or both of the first plate and the second plate include a spacer permanently fixed to the inner surface of the respective plate,

[0526] xvi. The monitoring mark has a sharp edge that (a) has a predetermined and known shape and size, and (b) can be observed by an imager that images the microfeatures;

[0527] xvii. The monitoring mark is a microstructure having at least one transverse linear dimension of 300 μm or less; and

[0528] xviii. The monitoring mark is located inside the sample;

[0529] wherein at least one of the marks is imaged by the imager during the assay.

[0530] One of the configurations is an open configuration, in which: two plates are partially or fully separated, the spacing between the plates is not adjusted by spacers, and the sample is deposited on one or both of the plates;

[0531] Another of the configurations is a closed configuration, which is configured after the sample is deposited in the open configuration and the plates are forced to the closed configuration by applying an imprecise pressing force on the stress area; and in the closed configuration: at least a portion of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, wherein the uniform thickness of the layer is defined by the sample contact area of the two plates and is adjusted by the plates and the spacers; and

[0532] wherein the monitoring mark is (i) a structure different from the spacer, or (ii) the same structure used as the spacer.

[0533] B. Improvement of Imaging Using a Sample Holder with Micro-Marks

[0534] BB-1. An apparatus for improving the imaging of micro-features in a sample, the apparatus comprising:

[0535] (e) An apparatus as in any of the previous apparatus embodiments; and

[0536] (f) An imager for determining a sample containing or suspected of containing micro-features;

[0537] wherein the imager captures an image, wherein at least one image includes a portion of the sample and the monitoring mark both.

[0538] CB-2. A system for improving the imaging of micro-features in a sample, the system comprising:

[0539] (g) An apparatus as in any of the previous apparatus embodiments;

[0540] (h) An imager for determining a sample containing or suspected of containing micro-features; and

[0541] (i) A non-transitory computer-readable medium having instructions that, when executed, adjust the settings of the imager for the next image using the mark as a parameter together with an imaging processing method.

[0542] C. Imaging Analysis Using a Sample Holder with Micro-Marks

[0543] CC-1 An apparatus for improving the analysis of an image of micro-features in a sample, the apparatus comprising:

[0544] (c) An apparatus as in any of the previous apparatus embodiments; and

[0545] (d) A computing device configured to receive images of a sample and a tag that contains or is suspected of containing microfeatures;

[0546] wherein the computing device runs an algorithm that uses the tag, together with an imaging processing method, as a parameter to improve the image quality in the image.

[0547] CC-2. A system for improving the analysis of an image of microfeatures in a sample, the system comprising:

[0548] (d) A device as in any of the preceding device embodiments;

[0549] (e) An imager configured to determine a sample that contains or is suspected of containing microfeatures by taking one or more images of the sample and the tag; and

[0550] (f) A non-transitory computer-readable medium having instructions that, when executed, use the tag as a parameter together with an imaging processing method to improve the image quality of at least one image taken in (c).

[0551] CC-3 A computer program product for determining microfeatures in a sample, the program comprising computer program code means for applying to and adapted for at least one image:

[0552] (a) Receiving an image of the sample and a monitoring tag, wherein the sample is loaded into a device as claimed in any of the preceding device claims, and wherein the image is taken by an imager; and

[0553] (b) Processing and analyzing the image to calculate the amount of the microfeatures, wherein the analysis uses a machine learning-based detection model and information provided by the image of the monitoring tag.

[0554] CC-4 A computing device for analyzing microfeatures in a sample, the computing device comprising a computing device that operates the algorithm in any embodiment of the present invention.

[0555] CC-5 The method, device, computer program product or system as in any of the preceding embodiments, wherein the improvement of the image quality comprises at least one selected from the group consisting of: denoising, image normalization, image sharpening, image scaling, alignment (e.g., for face detection), super-resolution, deblurring, and any combination thereof.

[0556] CC-6 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the imaging processing method includes at least one selected from the group consisting of: histogram-based operations, mathematics-based operations, convolution-based operations, smoothing operations, derivative-based operations, morphology-based operations, shadow correction, image enhancement and / or restoration, segmentation, feature extraction and / or matching, object detection and / or classification and / or localization, image understanding, and any combination thereof.

[0557] CC-6.1 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the histogram-based operations include at least one selected from the group consisting of: contrast stretching, equalization, minimum filtering, median filtering, maximum filtering, and any combination thereof.

[0558] CC-6.2 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the mathematics-based operations include at least one selected from the group consisting of: binary operations (e.g., NOT, OR, AND, XOR, and SUB), arithmetic-based operations (e.g., ADD, SUB, MUL, DIV, LOG, EXP, SQRT, TRIG, and INVERT), and any combination thereof.

[0559] CC-6.3 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the convolution-based operations include at least one selected from the group consisting of: operations in the spatial domain, Fourier transform, DCT, integer transform, operations in the frequency domain, and any combination thereof.

[0560] CC-6.4 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the smoothing operations include at least one selected from the group consisting of: linear filtering, uniform filtering, triangular filtering, Gaussian filtering, non-linear filtering, median filtering, kuwahara filtering, and any combination thereof.

[0561] CC-6.5 is a method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the derivative-based operations are selected from the group consisting of: first-order derivative operations, gradient filtering, basic derivative filtering, Prewitt gradient filtering, Sobel gradient filtering, alternative gradient filtering, Gaussian gradient filtering, second-order derivative filtering, basic second-order derivative filtering, frequency-domain Laplacian, Gaussian second-order derivative filtering, alternative Laplacian filtering, gradient direction second-order derivative (SDGD) filtering, third-order derivative filtering, higher-order derivative filtering (e.g., filtering greater than third-order derivative), and any combination thereof.

[0562] CC-6.6 A method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the morphology-based operations include at least one selected from the group consisting of dilation, erosion, Boolean convolution, opening and / or closing, hit-or-miss operations, contour, skeleton, propagation, gray-value morphology processing, gray-level dilation, gray-level erosion, gray-level opening, gray-level closing, morphological smoothing, morphological gradient, morphological Laplacian, and any combination thereof.

[0563] CC-6.7 A method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the image enhancement and / or restoration includes at least one selected from the group consisting of sharpening, de-sharpening, noise suppression, distortion suppression, and any combination thereof.

[0564] CC-6.8 A method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the segmentation includes at least one selected from the group consisting of thresholding, fixed thresholding, histogram-derived thresholding, Isodata algorithm, background symmetry algorithm, triangle algorithm, edge finding, gradient-based process, zero-crossing-based process, PLUS-based process, binary mathematical morphology, salt-and-pepper filtering, separating objects with holes, filling holes in objects, removing boundary-touching objects, outer skeleton, touching objects, gray-value mathematical morphology, Top-hat transform, thresholding, local contrast stretching, and any combination thereof.

[0565] CC-6.9 A method, apparatus, computer program product, or system as in any of the foregoing embodiments, wherein the feature extraction and / or matching includes at least one selected from the group consisting of independent component analysis, isometric mapping, kernel principal component analysis, latent semantic analysis, partial least squares, principal component analysis, multi-factor dimensionality reduction, non-linear dimensionality reduction, multilinear principal component analysis, multilinear subspace learning, semidefinite embedding, autoencoder, and any combination thereof.

[0566] T1. A method for determining a true lateral dimension (TLD) of a sample on a sample holder from a distorted image, the method comprising:

[0567] (a) Obtaining a device as in any of the foregoing embodiments, wherein the device includes one or more monitoring marks in the sample contact area;

[0568] (b) Obtaining an imager, computing hardware, and a non-transitory computer-readable medium containing an algorithm;

[0569] (c) Depositing a thin sample layer containing micro-features in the sample contact area of the device of (a);

[0570] (d) Taking one or more images of the thin sample layer using an imager and monitoring the markers, wherein the imager is located above the thin sample layer; and

[0571] (e) Using an algorithm to determine the true lateral dimensions of the sample;

[0572] wherein

[0573] (i) The algorithm is computer code executed on a computer system; and

[0574] (ii) The algorithm uses the images of the monitoring markers as parameters.

[0575] T2. The method, apparatus, computer program product or system of any of the preceding embodiments, wherein each monitoring marker comprises a protrusion or groove from the solid phase surface.

[0576] T3. The method, apparatus, computer program product or system of any of the preceding embodiments, wherein the microstructure does not have sharp edges.

[0577] T4. The method, apparatus, computer program product or system of any of the preceding embodiments, wherein the sample comprises a group selected from: biological samples, chemical samples, and samples without sharp edges.

[0578] T5. The method, apparatus, computer program product or system of any of the preceding embodiments, wherein the monitoring markers are used as parameters in the algorithm together with an imaging processing method, which (i) adjusts the image, (ii) processes the image of the sample, (iii) determines properties related to microfeatures, or (iv) any combination of the above.

[0579] T6. The method, apparatus, computer program product or system of any of the preceding embodiments, wherein the spacer has a substantially uniform height equal to or less than 200 microns and a fixed spacer distance (ISD);

[0580] AA. A method and apparatus for improving the imaging of a thin sample layer.

[0581] AA1. A method for improving the imaging of a thin sample layer, the method comprising:

[0582] (f) Obtaining a marker sample holder, wherein the sample comprises one or more monitoring markers in the sample contact area;

[0583] (g) Obtaining an imager, computing hardware, and a non - transitory computer - readable medium comprising an algorithm;

[0584] (h) Depositing a thin sample layer comprising microfeatures in the sample contact area of the apparatus of (a);

[0585] (i) Taking one or more images of a thin sample layer using an imager and monitoring markers, wherein the imager is located above the thin sample layer; and

[0586] (j) Using an algorithm to determine the true lateral dimension of the sample;

[0587] wherein

[0588] (i) The algorithm is computer code executed on a computer system; and

[0589] (ii) The algorithm uses the images of the monitoring markers as parameters.

[0590] A-1

[0591] T1. A method for determining the true lateral dimension (TLD) of a sample on a sample holder from a distorted image, the method comprising:

[0592] (k) Obtaining a device as in any of the foregoing embodiments, wherein the device comprises one or more monitoring markers in the sample contact area;

[0593] (l) Obtaining an imager, computing hardware, and a non-transitory computer-readable medium comprising an algorithm;

[0594] (m) Depositing a thin sample layer comprising microfeatures in the sample contact area of the device of (a);

[0595] (n) Taking one or more images of the thin sample layer and the monitoring markers using an imager, wherein the imager is located above the thin sample layer; and

[0596] (o) Using an algorithm to determine the true lateral dimension of the sample;

[0597] wherein

[0598] (i) The algorithm is computer code executed on a computer system; and

[0599] (ii) The algorithm uses the images of the monitoring markers as parameters.

[0600] 14. A device for determining microfeatures in a sample using an imager, the device comprising:

[0601] (a) A solid-phase surface comprising a sample contact area for contacting a sample comprising microfeatures; and

[0602] (b) One or more monitoring markers, wherein the monitoring markers:

[0603] ix. Are made of a material different from the sample;

[0604] x. Located inside the sample during the determination of the microstructure, where the sample forms a thin layer with a thickness less than 200 um on the sample contact area;

[0605] xi. Having a lateral linear dimension of about 1 um (micrometer) or greater, and

[0606] xii. Having at least one lateral linear dimension of 300 um or less; and

[0607] wherein during the determination process, the imager images at least one monitoring marker

[0608] wherein used during the determination of the analyte; and the geometric parameters (such as shape and size) of the monitoring marker and / or the spacing between the monitoring markers are (a) predetermined and known before the analyte determination, and (b) used as parameters in an algorithm for determining properties related to the microfeatures.

[0609] N8. A device for determining microfeatures in a sample using an imager, the device comprising:

[0610] A solid-phase surface comprising a sample contact area for contacting a sample containing microfeatures; and

[0611] One or more monitoring markers, where each monitoring marker comprises a protrusion or groove from the solid-phase surface, wherein:

[0612] xiii. The protrusion or groove includes a flat surface substantially parallel to an adjacent surface, which is the part of the solid-phase surface adjacent to the protrusion or groove;

[0613] xiv. The distance between the flat surface and the adjacent surface is about 200 micrometers (um) or less;

[0614] xv. The flat surface has (a) a linear dimension of at least about 1 um or greater, and (b) a region with at least one linear dimension of 150 um or less;

[0615] xvi. The flat surface of at least one monitoring marker is imaged by the imager used in the determination of the microfeatures; and

[0616] xvii. The shape of the flat surface, the size of the flat surface, the distance between the flat surface and the adjacent surface, and / or the spacing between the monitoring markers (a) are predetermined and known before the determination of the microfeatures, and (b) are used as parameters in an algorithm for determining properties related to the microfeatures.

[0617] N9. A device for determining microfeatures in a sample using an imager, the device comprising:

[0618] A first plate, a second plate, a spacer, and one or more monitoring markers, wherein:

[0619] xix. The first plate and the second plate are movable relative to each other into different configurations;

[0620] xx. Each of the first plate and the second plate includes an inner surface that includes a sample contact area for contacting a sample that includes microfeatures;

[0621] xxi. One or both of the first plate and the second plate include a spacer permanently fixed to the inner surface of the respective plate,

[0622] xxii. The spacer has a substantially uniform height and a fixed spacer distance (ISD) that is equal to or less than 200 micrometers;

[0623] xxiii. The monitoring marker is made of a material different from that of the sample;

[0624] xxiv. During the determination of the microstructure, the monitoring marker is located inside the sample, where the sample forms a thin layer with a thickness less than 200 um on the sample contact area; and

[0625] xxv. The monitoring marker has a transverse linear dimension of about 1 um (micrometer) or greater and has at least one transverse linear dimension of 300 um or less;

[0626] wherein during the determination process, the imager images at least one monitoring marker

[0627] wherein it is used during the determination of the microfeatures; and the shape, size, distance between the flat surface and the adjacent surface, and / or the spacing between the monitoring markers (a) are predetermined and known before the determination of the microfeatures, and (b) are used as parameters in the algorithm for determining the characteristics related to the microfeatures;

[0628] wherein one configuration is an open configuration, in which: the two plates are partially or completely separated, the spacing between the plates is not adjusted by the spacer, and the sample is deposited on one or both of the plates;

[0629] wherein the other of the configurations is a closed configuration, which is configured after the sample is deposited in the open configuration and the plates are forced to the closed configuration by applying an imprecise pressing force on the force application area; and in the closed configuration: at least a portion of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, where the uniform thickness of the layer is defined by the sample contact areas of the two plates and is adjusted by the plates and the spacer; and

[0630] wherein the monitoring marker is (i) a structure different from the spacer, or (ii) the same structure as the spacer used.

[0631] N10. An apparatus for determining microfeatures in a sample using an imager, the apparatus comprising:

[0632] A first plate, a second plate, a spacer, and one or more monitoring marks, wherein:

[0633] xxvi. The first plate and the second plate are movable relative to each other into different configurations;

[0634] xxvii. Each of the first plate and the second plate includes an inner surface that includes a sample contact area for contacting a sample containing microfeatures;

[0635] xxviii. One or both of the first plate and the second plate include a spacer permanently fixed to the inner surface of the respective plate,

[0636] xxix. The spacer has a substantially uniform height equal to or less than 200 microns and a fixed spacer distance (ISD);

[0637] xxx. Each monitoring mark includes a protrusion or a groove on one or both of the sample contact areas;

[0638] xxxi. The protrusion or the groove includes a flat surface substantially parallel to an adjacent surface, which is a portion of the solid phase surface adjacent to the protrusion or the groove;

[0639] xxxii. The distance between the flat surface and the adjacent surface is about 200 microns (um) or less;

[0640] xxxiii. The flat surface has (a) a linear dimension of at least about 1 um or greater, and (b) a region with at least one linear dimension of 150 um or less;

[0641] xxxiv. The flat surface of at least one monitoring mark is imaged by an imager used in determining the microfeatures; and

[0642] xxxv. The shape of the flat surface, the dimensions of the flat surface, the distance between the flat surface and the adjacent surface, and / or the spacing between the monitoring marks are (a) predetermined and known prior to the determination of the microfeatures, and (b) used as parameters in an algorithm for determining a property associated with the microfeatures.

[0643] One of the configurations is an open configuration, in which: the two plates are partially or fully separated, the spacing between the plates is not adjusted by the spacer, and the sample is deposited on one or both of the plates;

[0644] Another one of the configurations is a closed configuration, which is configured after the sample is deposited in the open configuration and the plate is forced to reach the closed configuration by applying an imprecise pressing force on the stressed area; and in the closed configuration: at least a part of the sample is compressed by the two plates into a layer with a very uniform thickness and is substantially stationary relative to the plates, where the uniform thickness of the layer is defined by the sample contact areas of the two plates and is adjusted by the plates and the spacers; and

[0645] where the monitoring mark is (i) a structure different from the spacer, or (ii) the same structure used as the spacer.

[0646] N12. An apparatus for imaging-based assays, comprising:

[0647] An apparatus as in any of the preceding apparatus embodiments, wherein the apparatus has at least five monitoring marks, and at least three of the monitoring marks are not aligned on a straight line.

[0648] N12. An apparatus for determining microfeatures in a sample using an imager, the system comprising:

[0649] (e) An apparatus as in any of the preceding apparatus embodiments; and

[0650] (f) An imager for determining a sample comprising microfeatures.

[0651] N14. A system for performing imaging-based assays, the system comprising:

[0652] (d) An apparatus as in any of the preceding apparatus embodiments;

[0653] (e) An imager for determining a sample comprising microfeatures; and

[0654] (f) A non-transitory computer-readable medium comprising instructions that, when executed, utilize the monitoring marks of the apparatus to determine characteristics related to the microfeatures.

[0655] NN8. A system for determining microfeatures in a sample using an imager, the system comprising:

[0656] (j) An apparatus as in any of the preceding apparatus embodiments;

[0657] (k) An imager for determining a sample comprising microfeatures; and

[0658] (l) A non-transitory computer-readable medium comprising instructions that, when executed, utilize the monitoring marks of the apparatus to determine characteristics related to the microfeatures, wherein the instructions comprise machine learning.

[0659] NN9. A method for determining microfeatures in a sample using an imager, comprising:

[0660] (d) obtaining a device, apparatus or system as in any of the foregoing embodiments;

[0661] (e) obtaining a sample and depositing the sample on a sample contact area in the device, apparatus or system of (a), wherein the sample contains microfeatures; and

[0662] (f) measuring the microfeatures.

[0663] 140. A method for measuring microfeatures in a sample using an imager, comprising:

[0664] (d) obtaining a device, apparatus or system as in any of the foregoing embodiments;

[0665] (e) obtaining a sample and depositing the sample on a sample contact area in the device, apparatus or system of (a), wherein the sample contains microfeatures;

[0666] (f) measuring the microfeatures, wherein the measurement comprises a step of using machine learning.

[0667] T1. A method for determining the true lateral dimension (TLD) of a sample on a sample holder from a distorted image, the method comprising:

[0668] (p) obtaining a device as in any of the foregoing embodiments, wherein the device includes one or more monitoring marks in the sample contact area;

[0669] (q) obtaining an imager, computing hardware and a non-transitory computer-readable medium containing an algorithm;

[0670] (r) depositing a thin sample layer containing microfeatures in the sample contact area of the device of (a);

[0671] (s) using the imager to take one or more images of the thin sample layer and the monitoring marks, wherein the imager is located above the thin sample layer; and

[0672] (t) using the algorithm to determine the true lateral dimension of the sample;

[0673] wherein

[0674] (i) the algorithm is computer code executed on a computer system; and

[0675] (ii) the algorithm uses the images of the monitoring marks as parameters.

[0676] T2. A method for determining the true lateral dimension (TLD) of a sample on a sample holder from a distorted image, the method comprising:

[0677] (a) Obtain a device as in any of the foregoing embodiments, wherein the device comprises one or more monitoring markers in the sample contact area;

[0678] (b) Obtain an imager, computing hardware, and a non-transitory computer-readable medium comprising an algorithm;

[0679] (c) Deposit a thin sample layer comprising microfeatures in the sample contact area of the device of (a);

[0680] (d) Use the imager to take one or more images of the thin sample layer and the monitoring markers, wherein the imager is located above the thin sample layer; and

[0681] (e) Use the algorithm to determine the true lateral dimensions and coordinates of the imaged sample in the real world by a physical metric (e.g., microns);

[0682] wherein

[0683] (i) the algorithm is computer code executed on a computer system; and

[0684] (ii) the algorithm uses the images of the monitoring markers as parameters.

[0685] T3. A device, method, or system as in any of the foregoing embodiments, wherein the microfeatures and monitoring markers from the sample are disposed within a sample holding device.

[0686] T4. A device, method, or system as in any of the foregoing embodiments, wherein the determination comprises detecting and locating the monitoring markers in the image of the sample taken by the imager.

[0687] T5. A device, method, or system as in any of the foregoing embodiments, wherein the determination comprises generating a monitoring marker grid based on the monitoring markers detected from the image of the sample taken by the imager.

[0688] T6. A device, method, or system as in any of the foregoing embodiments, wherein the determination comprises calculating a homography transformation from the generated monitoring marker grid.

[0689] T7. A device, method, or system as in any of the foregoing embodiments, wherein the determination comprises estimating the TLD from the homography transformation and determining the area, dimensions, and concentration of the microfeatures detected in the image-based determination.

[0690] T8. A method, device, or system as in any of the foregoing embodiments, wherein the TLD estimation is based on regions in the sample image taken by the imager and comprises:

[0691] (a) Obtain a sample;

[0692] (b) Load the sample into a sample holding device (such as a QMAX device) where there is a monitoring marker, where the monitoring marker is not immersed in the sample and is located in the device, and it can be imaged from the top by an imager in the image-based determination;

[0693] (c) Take an image of the sample in the sample loading device including the microfeatures and the monitoring marker;

[0694] (d) Detect the monitoring marker in the sample image taken by the imager;

[0695] (e) Divide the sample image into non-overlapping regions;

[0696] (f) Generate a region-based marker grid for each non-overlapping region where more than 5 non-collinear monitoring markers are detected in the local region;

[0697] (g) Generate a marker grid for all other regions not in (f) based on the monitoring markers detected from the sample image taken by the imager;

[0698] (h) Calculate the region-specific homography transformation for each region in (f) based on its own region-based marker grid generated in (f);

[0699] (i) Calculate the homography transformation for all other regions not in (f) based on the marker grid generated in (g);

[0700] (j) Estimate the region-based TLD for each region in (f) based on the region-based homography transformation in (g);

[0701] (k) Estimate the TLD for other regions not included in (f) based on the homography transformation in (i); and

[0702] (l) Apply the estimated TLDs from (j) and (k) to determine the area and concentration of the microfeatures imaged in each image partition in the image-based determination.

[0703] T9. The method, apparatus or system of any of the foregoing embodiments, wherein the monitoring markers in the sample holding device are distributed according to a periodic pattern having a defined pitch period.

[0704] T10. The method, apparatus or system of any of the foregoing embodiments, wherein the monitoring marker is detected and used as a detectable anchor for calibrating and improving the measurement accuracy in the image-based determination.

[0705] T12. The method, apparatus or system of any of the foregoing embodiments, wherein the detection of the monitoring markers in the sample image captured by the imager utilizes the periodicity of the distribution of the monitoring markers in the sample holding device for error correction and / or detection reliability.

[0706] T12. The method, apparatus or system of any of the foregoing embodiments, wherein the detection, recognition, area and / or shape profile estimation of the monitoring markers in the image-based determination are performed by machine learning (ML), which has an ML-based monitoring marker detection model and a device constructed or trained from the images captured by the imager on the device in the image-based determination.

[0707] T14. The method, apparatus or system of any of the foregoing embodiments, wherein the detection, recognition, area and / or shape profile estimation of the monitoring markers in the image-based determination are performed by image processing or image processing combined with machine learning.

[0708] T14. The method, apparatus or system of any of the foregoing embodiments, wherein the detected monitoring markers are applied to the TLD estimation in the image-based determination to calibrate the system and / or improve the measurement accuracy in the imaging-based determination.

[0709] T15. The method, apparatus or system of any of the foregoing embodiments, wherein the detected monitoring markers are applied in the image-based determination not limited to the microfeature size, volume and / or concentration estimation to calibrate the system and / or improve the measurement accuracy.

[0710] T16. The method, apparatus or system of any of the foregoing embodiments, wherein the detection of the monitoring markers and / or the TLD estimation are applied to the fault detection in the image-based determination, including but not limited to detecting defects in the sample holding device, misalignment of the sample holding device in the imager and / or focusing faults of the imager.

[0711] T17. The method, apparatus or system of any of the foregoing embodiments, wherein the monitoring markers are detected as anchors for application to the system to estimate the area of an object in the image-based determination, including:

[0712] i. Loading the sample into a sample holding device having monitoring markers located in the device in the image-based determination;

[0713] ii. Capturing an image of the sample in the sample holding device including the microfeatures and the monitoring markers; and

[0714] iii. Detect the monitoring marks in the image of the sample captured by the imager on the sample holding device, determine the TLD and calculate the area estimate in the image-based determination to determine the size of the imaged object from the pixels in the image to its physical size in microns in the real world.

[0715] T18. A method, apparatus, or system as in any of the preceding embodiments, wherein the system comprises:

[0716] i. Detect the monitoring marks in the digital image;

[0717] ii. Generate a monitoring mark grid;

[0718] iii. Calculate the image transformation based on the monitoring mark grid; and

[0719] iv. Estimate the area of the object in the image of the sample and its physical size in the real world in the image-based determination.

[0720] T19. A method, apparatus, or system as in any of the preceding embodiments, wherein a homography transformation is calculated using the monitoring mark grid generated from the detected monitoring marks to estimate the TLD, the area of the object in the image of the sample captured by the imager, and the physical size of the object in the real world.

[0721] T20. A method, apparatus, or system as in any of the preceding embodiments, wherein the method comprises:

[0722] i. Divide the image of the sample captured by the imager in the image-based determination into non-overlapping regions;

[0723] ii. Detect the local monitoring marks in the image;

[0724] iii. If more than 5 non-collinear monitoring marks are detected in the region, generate a region-based mark grid for the region;

[0725] iv. Generate a mark grid for all other regions based on the monitoring marks detected in the image of the sample captured by the imager;

[0726] v. Calculate the region-based homography transformation from the region-based mark grid generated for each region in (iii);

[0727] vi. Calculate the homography transformation for all other regions not in (iii) based on the mark grid generated in (iv); and

[0728] vii. Estimate the TLD for each region based on the homography transformations resulting from (v) and (vi), and then determine the area of the object in the image of the sample for each region and their sizes in the real world in the image-based determination.

[0729] T21. A method, apparatus, or system as in any of the foregoing embodiments, wherein the determination is a medical, diagnostic, chemical, or biological test.

[0730] T22. A method, apparatus, or system as in any of the foregoing embodiments, wherein the microfeature is a cell.

[0731] T23. A method, apparatus, or system as in any of the foregoing embodiments, wherein the microfeature is a blood cell.

[0732] T24. A method, apparatus, or system as in any of the foregoing embodiments, wherein the microfeature is a protein, peptide, DNA, RNA, nucleic acid, small molecule, cell, or nanoparticle.

[0733] T25. A method, apparatus, or system as in any of the foregoing embodiments, wherein the microcomponent comprises a tag.

[0734] T26. A method, apparatus, or system as in any of the foregoing embodiments, wherein the algorithm comprises a computer program product, the computer program product comprising computer program code means adapted in at least one image to:

[0735] (a) Receive an image of a sample, wherein the sample is loaded into a QMAX device and the image is taken by an imager connected to the QMAX device, wherein the image includes both the sample and a monitoring marker; and

[0736] (b) Analyze the image using a detection model and generate a two-dimensional data array of the image, wherein the two-dimensional data array includes probability data of the microfeatures at each position in the image, the detection model being established through a training process, the training process comprising:

[0737] i. Feeding an annotated data set into a convolutional neural network, wherein the annotated data set is from a sample of the same type as the test sample and for the same microfeatures; and

[0738] ii. Training and establishing a detection model through convolution; and

[0739] (c) Analyze the two-dimensional data array to detect local signal peaks by:

[0740] i. A signal list process, or

[0741] ii. A local search process; and

[0742] (d) Calculate the amount of the microfeatures based on the local signal peak information.

[0743] T27. A method, apparatus, or system as in any of the foregoing embodiments, wherein the algorithm comprises a computer program product, the computer program product comprising computer program code means which is applied to and adapted to in at least one image:

[0744] (a) Represent an inferred pattern between an object in a sample and a pixel contour map of the object in an image of the sample taken by an imager on a sample holding device;

[0745] (b) Digitally reconstruct an image of at least one object detected from the inferred pattern in an image of the sample and generate a contour mask surrounding the object identified by the inference module, wherein the object is in focus;

[0746] (c) Identify at least a portion of the sample image for at least one object in a selected portion of the sample image; and

[0747] (d) Calculate at least one feature of the object from the at least one portion to identify the object in the selected portion of the image of the sample taken by the imager;

[0748] (e) Calculate the count and concentration of the detected objects in the selected portion from the image of the sample;

[0749] When the program runs on a computing device or in a computing cloud connected via a network.

[0750] T28. A method, apparatus, or system as in any of the foregoing embodiments, wherein the algorithm comprises a computer program product, the computer program product comprising computer program code means which is adapted to in at least one image:

[0751] (a) Receive an image of a sample, wherein the sample is loaded into a QMAX device and the image is taken by an imager connected to the QMAX device, wherein the image includes both the sample and a monitoring marker; and

[0752] (b) Analyze the image to calculate the amount of the microfeatures, wherein the analysis uses a machine learning-based detection model and information provided by the image of the monitoring marker.

[0753] A method, apparatus, or system as in any of the foregoing embodiments further comprises a computer-readable storage medium or memory storage unit which comprises a computer program as in any of the foregoing embodiments.

[0754] A method, apparatus, or system as in any of the foregoing embodiments further includes a computing device or mobile device which comprises a computing device as in any of the foregoing embodiments.

[0755] The method, apparatus, or system of any of the foregoing embodiments further includes a computing device or a mobile device that includes a computer program product of any of the foregoing embodiments.

[0756] The method, apparatus, or system of any of the foregoing embodiments further includes a computing device or a mobile device that includes a computer-readable storage medium or storage unit of any of the foregoing embodiments.

[0757] A system for analyzing a sample, comprising:

[0758] A first plate, a second plate, a surface amplification layer, and a capture agent, wherein

[0759] (a) The first plate and the second plate are movable relative to each other into different configurations and have sample contact regions on their respective inner surfaces for contacting a sample containing a target analyte;

[0760] (b) The surface amplification layer is on one of the sample contact regions,

[0761] (c) The capture agent is immobilized on the surface amplification layer, wherein the capture agent specifically binds to the target analyte,

[0762] wherein when the surface amplification layer is adjacent to the surface amplification layer, the surface amplification layer amplifies an optical signal from the target analyte or a label attached to the target analyte, and the optical signal is much stronger than the optical signal when the surface amplification layers are separated by micrometers or more;

[0763] wherein one configuration is an open configuration in which the average spacing between the inner surfaces of the two plates is at least 200 μm; and

[0764] wherein the other configuration is a closed configuration in which at least a portion of the sample is located between the two plates and the average spacing between the inner surfaces of the plates is less than 200 μm.

[0765] A system for analyzing a sample, comprising:

[0766] A first plate, a second plate, a surface amplification layer, and a capture agent, wherein

[0767] (d) The first plate and the second plate are movable relative to each other into different configurations and have sample contact regions on their respective inner surfaces for contacting a sample containing a target analyte;

[0768] (e) The surface amplification layer is on one of the sample contact regions,

[0769] (f) The capture agent is immobilized on the surface amplification layer, wherein the capture agent specifically binds to the target analyte,

[0770] When the surface amplification layer is adjacent to the surface amplification layer, the surface amplification layer amplifies the optical signal from the label attached to the target analyte, and the optical signal is much stronger than the optical signal when the surface amplification layer is separated by micrometers or more.

[0771] One configuration is an open configuration, where the average spacing between the inner surfaces of the two plates is at least 200 um;

[0772] Another configuration is a closed configuration, where at least a portion of the sample is located between the two plates and the average spacing between the inner surfaces of the plates is less than 200 um;

[0773] The thickness of the sample in the closed configuration, the concentration of the label dissolved in the sample in the closed configuration, and the amplification factor of the surface amplification layer are configured such that any label directly or indirectly bound to the capture agent in the closed configuration is visible without washing away the unbound labels.

[0774] An apparatus comprising the device of any of the preceding embodiments and a reader for reading the device.

[0775] A homogeneous assay using the device of any of the preceding embodiments, wherein the thickness of the sample in the closed configuration, the concentration of the label, and the amplification factor of the amplified surface are configured such that the label bound to the amplified surface is visible without washing away the unbound labels.

[0776] The method of any of the preceding embodiments, wherein the method is performed by:

[0777] Obtaining a device of any of the preceding embodiments;

[0778] Depositing the sample on one or both plates when the plates are in the open configuration;

[0779] Closing the plates to the closed configuration; and

[0780] Reading the sample contact area with a reading device to generate an image of the signal.

[0781] The device or method of any of the preceding embodiments, wherein the label bound to the amplified surface is visible in less than 60 seconds.

[0782] The device or method of any of the preceding embodiments, wherein the method is a homogeneous assay, wherein the signal is read without using a washing step to remove any biomaterial or label not bound to the amplified surface.

[0783] The device or method of any of the preceding embodiments, wherein the label bound to the amplified surface is read by a pixelated reading method.

[0784] The device or method of any of the foregoing embodiments, wherein the label bound to the amplification surface is read by a lumped reading method.

[0785] The device or method of any of the foregoing embodiments, wherein the assay has a detection sensitivity of 0.1 nM or less.

[0786] The device or method of any of the foregoing embodiments, wherein unbound biomaterial or label is removed by a sponge prior to reading.

[0787] The device or method of any of the foregoing embodiments, wherein the signal amplification layer comprises D2PA.

[0788] The device or method of any of the foregoing embodiments, wherein the signal amplification layer comprises a metallic material layer.

[0789] The device or method of any of the foregoing embodiments, wherein the signal amplification layer comprises a continuous metal film made of a material selected from the group consisting of gold, silver, copper, aluminum, alloys thereof, and combinations thereof.

[0790] The device or method of any of the foregoing embodiments, wherein different metal layers locally enhance or act as reflectors, or both, to enhance the optical signal.

[0791] The device or method of any of the foregoing embodiments, wherein the signal amplification layer comprises a metallic material layer and a dielectric material on top of the metallic material layer, wherein the capture agent is on the dielectric material.

[0792] The device or method of any of the foregoing embodiments, wherein the metallic material layer is a uniform metal layer, a nanostructured metal layer, or a combination.

[0793] The device or method of any of the foregoing embodiments, wherein the signal is amplified by plasmon enhancement.

[0794] The device or method of any of the foregoing embodiments, wherein the assay comprises detecting the label by Raman scattering.

[0795] The device or method of any of the foregoing embodiments, wherein the capture agent is an antibody.

[0796] The device or method of any of the foregoing embodiments, wherein the capture agent is a polynucleotide.

[0797] The device or method of any of the foregoing embodiments, wherein the device further comprises a spacer fixed to one plate, wherein the spacer adjusts the spacing between the first and second plates in a closed configuration.

[0798] The device or method of any of the foregoing embodiments, wherein the amplification factor of the surface amplification layer is adjusted such that the optical signal from a single label bound directly or indirectly to the capture agent is visible.

[0799] The device or method of any of the foregoing embodiments, wherein the amplification factor of the surface amplification layer is adjusted such that the optical signal from a single tag directly or indirectly bound to the capture agent is visible, and the visible single tags bound to the capture agent are counted individually.

[0800] The device or method of any of the foregoing embodiments, wherein the spacing between the first plate and the second plate in the closed configuration is configured such that the saturation binding time of the target analyte to the capture agent is 300 seconds or less.

[0801] The device or method of any of the foregoing embodiments, wherein the spacing between the first plate and the second plate in the closed configuration is configured such that the saturation binding time of the target analyte to the capture agent is 60 seconds or less.

[0802] The device or method of any of the foregoing embodiments, wherein the amplification factor of the surface amplification layer is adjusted such that the light signal from a single tag is visible.

[0803] The device or method of any of the foregoing embodiments, wherein the capture agent is a nucleic acid.

[0804] The device or method of any of the foregoing embodiments, wherein the capture agent is a protein.

[0805] The device or method of any of the foregoing embodiments, wherein the capture agent is an antibody.

[0806] The device or method of any of the foregoing embodiments, wherein the sample contact area of the second plate has a reagent storage site, and in the closed configuration the storage site is approximately above the binding site on the first plate.

[0807] The device or method of any of the foregoing embodiments, wherein the reagent storage site contains a detection agent that binds to the target analyte.

[0808] The device or method of any of the foregoing embodiments, wherein the detection agent contains a tag.

[0809] The device or method of any of the foregoing embodiments, wherein both the capture agent and the detection agent bind to the target analyte to form a sandwich containing a tag.

[0810] The device or method of any of the foregoing embodiments, wherein the signal amplification layer comprises a metal material layer.

[0811] The device or method of any of the foregoing embodiments, wherein the signal amplification layer comprises a metal material layer and a dielectric material on top of the metal material layer, and the capture agent is on the dielectric material.

[0812] The device or method of any of the foregoing embodiments, wherein the metal material layer is a uniform metal layer, a nanostructured metal layer, or a combination.

[0813] The apparatus or method of any of the foregoing embodiments, wherein the amplification layer comprises a layer of metallic material and a dielectric material on top of the metallic material layer, wherein the capture agent is on the dielectric material, and the dielectric material layer has a thickness of 0.5 nm, 1 nm, 5 nm, 10 nm, 20 nm, 50 nm, 100 nm, 200 nm, 500 nm, 1000 nm, 2 µm, 3 µm, 5 µm, 10 µm, 20 µm, 30 µm, 50 µm, 100 µm, 200 µm, 500 µm, or within a range between any two values.

[0814] The apparatus or method of any of the foregoing embodiments, wherein the method further comprises quantifying the signal in a region of the image to provide an estimate of the amount of one or more analytes in the sample.

[0815] The apparatus or method of any of the foregoing embodiments, wherein the method comprises identifying and counting individual binding events between analytes and capture agents in the image region, thereby providing an estimate of the amount of one or more analytes in the sample.

[0816] The apparatus or method of any of the foregoing embodiments, wherein the identifying and counting step comprises: (1) determining the local intensity of the background signal; (2) determining the local signal intensity of one label, two labels, three labels, four or more labels; (3) determining the total number of labels within the imaging region.

[0817] The apparatus or method of any of the foregoing embodiments, wherein the identifying and counting step comprises: (1) determining the local spectrum of the background signal; (2) determining the local signal spectrum of one label, two labels, three labels, four or more labels; (3) determining the total number of labels within the imaging region.

[0818] The apparatus or method of any of the foregoing embodiments, wherein the identifying and counting step comprises: (1) determining the local Raman signature of the background signal; (2) determining the local signal Raman signature of one label, two labels, three labels, four or more labels; (3) determining the total number of labels within the imaging region.

[0819] The apparatus or method of any of the foregoing embodiments, wherein the identifying and counting step comprises determining one or more of the local intensity, spectrum, and Raman signature.

[0820] The apparatus or method of any of the foregoing embodiments, wherein the method comprises quantifying the lumped signal in a region of the image, thereby providing an estimate of the amount of one or more analytes in the sample.

[0821] The apparatus or method of any of the foregoing embodiments, wherein the sample contact region of the second plate has a reagent storage site, and in the closed configuration the storage site is approximately above the binding site on the first plate.

[0822] An apparatus or method as in any of the preceding embodiments, wherein the method further comprises the step of labeling the target analyte with a detection agent.

[0823] An apparatus or method as in any of the preceding embodiments, wherein the detection agent comprises a label.

[0824] An apparatus or method as in any of the preceding embodiments, wherein both the capture agent and the detection agent bind to the target analyte to form a sandwich.

[0825] An apparatus or method as in any of the preceding embodiments, wherein the method further comprises measuring the volume of the sample in the area imaged by the reading device.

[0826] An apparatus or method as in any of the preceding embodiments, wherein the target analyte is a protein, peptide, DNA, RNA, nucleic acid, small molecule, cell, or nanoparticle.

[0827] An apparatus or method as in any of the preceding embodiments, wherein the image shows the position, local intensity, and local spectrum of the signal.

[0828] An apparatus or method as in any of the preceding embodiments, wherein the signal is a luminescence signal selected from the group consisting of fluorescence, electroluminescence, chemiluminescence, and electrochemiluminescence signals.

[0829] An apparatus or method as in any of the preceding embodiments, wherein the signal is a Raman scattering signal.

[0830] An apparatus or method as in any of the preceding embodiments, wherein the signal is a force resulting from local electrical interaction, local mechanical interaction, local biological interaction, or local optical interaction between the plate and the reading device.

[0831] A method or apparatus as in any of the preceding embodiments, wherein the spacer has a columnar shape and an almost uniform cross-section.

[0832] A method or apparatus as in any of the preceding embodiments, wherein the spacing distance (SD) is equal to or less than about 120 μm (micrometers).

[0833] A method or apparatus as in any of the preceding embodiments, wherein the spacing distance (SD) is equal to or less than about 100 μm (micrometers).

[0834] A method or apparatus as in any of the preceding embodiments, wherein the fourth power of the spacing distance (ISD) divided by the thickness (h) and Young's modulus (E) of the flexible plate (ISD 4 / (hE)) is 5 × 10 6 μm 3 / GPa or less.

[0835] A method or apparatus as in any of the foregoing embodiments, wherein the fourth power of the spacing distance (ISD) divided by the thickness (h) and Young's modulus (E) of the flexible plate (ISD 4 / (hE)) is 5×10 5 um 3 / GPa or less.

[0836] A method or apparatus as in any of the foregoing embodiments, wherein the spacer has a columnar shape, a substantially flat top surface, a predetermined substantially uniform height, and a predetermined constant spacing distance, the spacing distance being at least about 2 times greater than the size of the analyte, wherein the Young's modulus of the spacer multiplied by the filling factor of the spacer is equal to or greater than 2 MPa, wherein the filling factor is the ratio of the spacer contact area to the total plate area, and wherein for each spacer, the ratio of the lateral dimension of the spacer to its height is at least 1 (1).

[0837] A method or apparatus as in any of the foregoing embodiments, wherein the spacer has a columnar shape, a substantially flat top surface, a predetermined substantially uniform height, and a predetermined constant spacing distance, the spacing distance being at least about 2 times greater than the size of the analyte, wherein the Young's modulus of the spacer multiplied by the filling factor of the spacer is equal to or greater than 2 MPa, wherein the filling factor is the ratio of the spacer contact area to the total plate area, and wherein for each spacer, the ratio of the lateral dimension of the spacer to its height is at least 1 (1), wherein the fourth power of the spacing distance (ISD) divided by the thickness (h) and Young's modulus (E) of the flexible plate (ISD 4 / (hE)) is 5x10 6 um 3 / GPa or less.

[0838] A method or apparatus as in any of the foregoing embodiments, wherein the ratio of the spacing distance of the spacer to the average width of the spacer is 2 or greater, and the filling factor of the spacer multiplied by the Young's modulus of the spacer is 2 MPa or greater.

[0839] A method or apparatus as in any of the foregoing embodiments, wherein the analyte is a protein, peptide, nucleic acid, synthetic compound, or inorganic compound.

[0840] A method or apparatus as in any of the foregoing embodiments, wherein the sample is a biological sample, the biological sample being selected from amniotic fluid, aqueous humor, vitreous humor, blood (e.g., whole blood, fractionated blood, plasma, or serum), breast milk, cerebrospinal fluid (CSF), earwax (cerumen), chyle, chyme, endolymph, perilymph, feces, breath, gastric acid, gastric juice, lymph fluid, mucus (including nasal drainage and sputum), pericardial fluid, peritoneal fluid, pleural fluid, pus, rheumatic fluid, saliva, exhaled condensate, sebum, semen, sputum, sweat, synovial fluid, tears, vomit, and urine.

[0841] A method or apparatus as in any of the foregoing embodiments, wherein the spacer has a columnar shape and the ratio of the width to the height of the column is equal to or greater than 1.

[0842] A method or apparatus as in any of the foregoing embodiments, wherein the sample deposited on one or both of the plates has an unknown volume.

[0843] A method or apparatus as in any of the foregoing embodiments, wherein the spacer has a columnar shape and the column has a substantially uniform cross-section.

[0844] A method or apparatus as in any of the foregoing embodiments, wherein the sample is used for detecting, purifying, and quantifying compounds or biomolecules related to the stages of certain diseases.

[0845] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to infectious and parasitic diseases, injuries, cardiovascular diseases, cancers, mental disorders, neuropsychiatric disorders, lung diseases, kidney diseases, and other organic diseases.

[0846] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to the detection, purification, and quantification of microorganisms.

[0847] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to viruses, fungi, and bacteria from the environment (e.g., water, soil, or biological samples).

[0848] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to the detection and quantification of compounds or biological samples (e.g., toxic waste, anthrax) that pose a hazard to food safety or national security.

[0849] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to the quantification of vital parameters in medical or physiological monitoring.

[0850] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to glucose, blood, oxygen levels, total blood cell count.

[0851] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to the detection and quantification of specific DNA or RNA from biological samples.

[0852] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to the sequencing and comparison of genetic sequences of DNA in chromosomes and mitochondria for genomic analysis.

[0853] A method or apparatus as in any of the foregoing embodiments, wherein the sample is related to detecting reaction products, for example, during drug synthesis or purification.

[0854] The method or device of any of the foregoing embodiments, wherein the sample is a cell, tissue, body fluid, and feces.

[0855] The method or device of any of the foregoing embodiments, wherein the sample is a sample in the fields of human, veterinary, agricultural, food, environmental, and drug testing.

[0856] The method or device of any of the foregoing embodiments, wherein the sample is a biological sample selected from hair, nails, ear wax, breath, connective tissue, muscle tissue, nerve tissue, epithelial tissue, cartilage, cancer sample, or bone.

[0857] The method or device of any of the foregoing embodiments, wherein the spacing distance is in the range of 5 um to 120 um.

[0858] The method or device of any of the foregoing embodiments, wherein the spacing distance is in the range of 120 um to 200 um.

[0859] The method or device of any of the foregoing embodiments, wherein the thickness of the flexible plate is in the range of 20 um to 250 um and the Young's modulus is in the range of 0.1 to 5 GPa.

[0860] The method or device of any of the foregoing embodiments, wherein for the flexible plate, the thickness of the flexible plate multiplied by the Young's modulus of the flexible plate is in the range of 60 to 750 GPa-um.

[0861] The method or device of any of the foregoing embodiments, wherein the uniformly thick sample layer is uniform over a lateral area of at least 1 mm 2 thereof.

[0862] The method or device of any of the foregoing embodiments, wherein the uniformly thick sample layer is uniform over a lateral area of at least 3 mm 2 thereof.

[0863] The method or device of any of the foregoing embodiments, wherein the uniformly thick sample layer is uniform over a lateral area of at least 5 mm 2 thereof.

[0864] The method or device of any of the foregoing embodiments, wherein the uniformly thick sample layer is uniform over a lateral area of at least 10 mm 2 thereof.

[0865] The method or device of any of the foregoing embodiments, wherein the uniformly thick sample layer is uniform over a lateral area of at least 20 mm 2 thereof.

[0866] The method or device of any of the foregoing embodiments, wherein the uniformly thick sample layer is uniform over a lateral area in the range of 20 mm 2 to 100 mm 2 thereof.

[0867] A method or apparatus as in any of the foregoing embodiments, wherein the sample layer of uniform thickness has a thickness uniformity of up to + / - 5% or better.

[0868] A method or apparatus as in any of the foregoing embodiments, wherein the sample layer of uniform thickness has a thickness uniformity of up to + / - 10% or better.

[0869] A method or apparatus as in any of the foregoing embodiments, wherein the sample layer of uniform thickness has a thickness uniformity of up to + / - 20% or better.

[0870] A method or apparatus as in any of the foregoing embodiments, wherein the sample layer of uniform thickness has a thickness uniformity of up to + / - 30% or better.

[0871] A method, apparatus, computer program product or system as in any of the foregoing embodiments having five or more monitoring markers, wherein at least three of the monitoring markers are not straight.

[0872] A method, apparatus, computer program product or system as in any of the foregoing embodiments, wherein each of the plates includes a force-receiving area on its respective outer surface for applying an imprecise pressing force that forces the plates together;

[0873] A method, apparatus, computer program product or system as in any of the foregoing embodiments, wherein one or two of the plates are flexible;

[0874] A method, apparatus, computer program product or system as in any of the foregoing embodiments, wherein the fourth power of the spacing distance (IDS) divided by the thickness (h) and Young's modulus (E) of the flexible plate (ISD4 / (hE)) is 5 x 106 um3 / GPa or less.

[0875] A method, apparatus, computer program product or system as in any of the foregoing embodiments, wherein at least one of the spacers is within the sample contact area;

[0876] A method, apparatus, computer program product or system as in any of the foregoing embodiments, wherein the fat-defined analyte

[0877] A method, apparatus, computer program product or system as in any of the foregoing embodiments, wherein the fat-defined algorithm

[0878] A method, apparatus, computer program product or system as in any of the foregoing embodiments, wherein the fat-defined imprinting force and hand pressing force.

[0879] An apparatus, system, or method as in any of the foregoing embodiments, wherein the algorithm is stored on a non-transitory computer-readable medium, and wherein the algorithm comprises instructions that, when executed, perform a method of determining a property corresponding to the analyte using the monitoring markers of the apparatus.

[0880] Some examples of markers

[0881] In the present invention, in some embodiments, the markers have the same shape as the spacers.

[0882] In some embodiments, the markers are periodic or aperiodic.

[0883] In some embodiments, the distance between two markers is predetermined and known, but the absolute coordinates on the plate are unknown.

[0884] In some embodiments, the markers have a predetermined and known shape.

[0885] In some embodiments, the markers are configured to have a distribution in the plate such that, regardless of the position of the plate, there are always markers in the field of view of the imaging optics.

[0886] In some embodiments, the markers are configured to have a distribution in the plate such that, regardless of the position of the plate, there are always markers in the field of view of the imaging optics, and the number of markers is sufficient for local optical information.

[0887] In some embodiments, the markers are used to control the optical properties of a local region of the sample, and the region size is 1 um^2, 5 um^2, 10 um^2, 20 um^2, 50 um^2, 100 um^2, 200 um^2, 500 um^2, 1000 um^2, 2000 um^2, 5000 um^2, 10000 um^2, 100000 um^2, 500000 um^2, or within a range between any two values.

[0888] Use of "finite imaging optics"

[0889] In the present invention, in some embodiments, the optical system for imaging the assay has "finite imaging optics". Some embodiments of finite imaging optics include, but are not limited to:

[0890] 1. A finite imaging optical system comprising:

[0891] An imaging lens;

[0892] An imaging sensor;

[0893] Wherein the imaging sensor is part of a camera of a smartphone;

[0894] At least one of the imaging lenses is part of a camera of a smartphone;

[0895] 2. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the physical optical resolution is less than 1um, 2um, 3um, 5um, 10um, 50um, or within a range between any two values.

[0896] 3. The finite imaging optical system according to any one of the foregoing embodiments, wherein: each physical optical resolution is less than 1um, 2um, 3um, 5um, 10um, 50um, or within a range between any two values.

[0897] 4. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the preferred each physical optical resolution is between about 1um and 3um;

[0898] 5. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the numerical aperture is less than 0.1, 0.15, 0.2, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5, or within a range between any two values.

[0899] 6. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the preferred numerical aperture is between 0.2 and 0.25.

[0900] 7. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the working distance is 0.2mm, 0.5mm, 1mm, 2mm, 5mm, 10mm, 20mm, or within a range between any two values.

[0901] 8. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the working distance is 0.2mm, 0.5mm, 1mm, 2mm, 5mm, 10mm, 20mm, or within a range between any two values.

[0902] 9. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the preferred working distance is between 0.5mm and 1mm.

[0903] 10. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the depth of focus is 100nm, 500nm, 1um, 2um, 10um, 100um, 1mm, or within a range between any two values.

[0904] 11. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the depth of focus is 100nm, 500nm, 1um, 2um, 10um, 100um, 1mm, or within a range between any two values.

[0905] 12. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the image sensor is part of a smartphone camera module.

[0906] 13. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the diagonal length of the image sensor is less than 1 inch, 1 / 2 inch, 1 / 3 inch, 1 / 4 inch, or within a range between any two values;

[0907] 14. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the imaging lens includes at least two lenses, and one lens is part of the camera module of the smartphone.

[0908] 15. The finite imaging optical system according to any one of the foregoing embodiments, wherein: at least one outer lens is paired with an inner lens of the smartphone.

[0909] 16. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the optical axis of the outer lens is aligned with the inner lens of the smartphone, and the alignment tolerance is less than 0.1 mm, 0.2 mm, 0.5 mm, 1 mm, or within a range between any two values.

[0910] 17. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the height of the outer lens is less than 2 mm, 5 mm, 10 mm, 15 mm, 20 mm, or within a range between any two values.

[0911] 18. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the preferred height of the outer lens is between 3 mm and 8 mm.

[0912] 19. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the preferred height of the outer lens is between 3 mm and 8 mm.

[0913] 20. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the diameter of the outer lens is less than 2 mm, 4 mm, 8 mm, 10 mm, 15 mm, 20 mm, or within a range between any two values.

[0914] 21. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the per physical optical magnification is less than 0.1X, 0.5X, 1X, 2X, 4X, 5X, 10X, or within a range between any two values.

[0915] 22. The finite imaging optical system according to any one of the foregoing embodiments, wherein: the preferred per physical optical magnification is less than 0.1X, 0.5X, 1X, 2X, 4X, 5X, 10X, or within a range between any two values.

[0916] The term "image-based determination" refers to a determination procedure that utilizes an image of a sample taken by an imager, where the sample can be, but is not limited to, medical, biological, and chemical samples.

[0917] The term "imager" refers to any device capable of taking an image of an object. It includes, but is not limited to, a camera in a microscope, a smartphone, or a special device that can take images at various wavelengths.

[0918] The term "sample feature" refers to some properties of a sample that represent conditions of potential interest. In some embodiments, the sample features are features that appear in an image of the sample and can be segmented and classified by a machine learning model. Examples of sample features include, but are not limited to, the type of analyte in the sample, such as red blood cells, white blood cells, and tumor cells, and it includes analyte count, size, volume, concentration, etc.

[0919] The term "machine learning" refers to algorithms, systems, and devices in the field of artificial intelligence that typically use statistical techniques and artificial neural networks to provide a computer with the ability to "learn" (i.e., gradually improve performance on a specific task) from data without being explicitly programmed.

[0920] The term "artificial neural network" refers to a hierarchically connected system inspired by biological networks that can "learn" by considering examples to perform tasks, usually without being programmed with any task-specific rules.

[0921] The term "convolutional neural network" refers to a class of multi-layer feedforward artificial neural networks most commonly applied to analyze visual images.

[0922] The term "deep learning" refers to a large class of machine learning methods in artificial intelligence (AI) that learn from data with a certain depth network structure.

[0923] The term "machine learning model" refers to a trained computational model constructed from a training process in machine learning of data. The trained machine learning model is applied by a computer in an inference phase, which gives the computer the ability to perform specific tasks (e.g., detecting and classifying objects). Examples of machine learning models include ResNet, DenseNet, etc., which are also referred to as "deep learning models" due to the hierarchical depth in their network structures.

[0924] The term "image segmentation" refers to an image analysis process of dividing a digital image into multiple segments (groups of pixels, usually with a set of bitmap masks covering the image segments surrounded by the contour of their segment boundaries). Image segmentation can be achieved by image segmentation algorithms in image processing, such as watershed, grabcut, mean shift, etc., and by machine learning algorithms, such as MaskRCNN, etc.

[0925] The term "defects in the sample" refers to artifacts that should not be present under ideal sample conditions or should not be considered in the characteristics of the sample. They can come from, but are not limited to, contaminants such as dust, bubbles, etc., and from peripheral objects in the sample, such as monitoring markers (e.g., columns) in the sample holding device. Defects can have significant dimensions and occupy a significant amount of volume in the sample, such as bubbles. In addition to their distribution and quantity in the sample, they can have different shapes - which depend on the sample.

[0926] The term "threshold" in this article refers to any number used, for example, as a cut-off value for classifying sample characteristics as a specific type of analyte, or the ratio of abnormal cells to normal cells in the sample. Thresholds can be identified empirically or analytically.

[0927] Traditionally, high-quality (including but not limited to: high resolution, good illumination, low noise level, low distortion level, good focus, good sharpness, etc.) microscopic images need to be collected by complex and very advanced imaging optics, which include high numerical aperture and high magnification compound lenses. And we have invented a method for reconstructing images with high quality using low-cost limited optics in microscopic imaging with the aid of machine learning. As shown in Figure C-1, during the training process, we use high-quality images (captured with high-quality optics) and low-quality images (captured with limited optics and / or generated from high-quality images (referred to as virtual images) by using image processing and / or machine learning) as training pairs for the same field of view (region) of interest on the sample. Then we use the high-quality and low-quality image pairs to train the machine to learn what the low-quality image in high-quality form should be. Using the trained model, for new samples of the same kind of objects, the machine can predict the high-quality form of the image based on the low-quality input.

[0928] In some embodiments, monitoring markers are used in machine learning to achieve high-resolution images using a lower-resolution optical system.

[0929] Use of "limited sample operation"

[0930] In the present invention, in some embodiments, the sample positioning system for imaging an assay has "limited sample operation". Some embodiments of limited sample operation include but are not limited to:

[0931] Description of the limited sample operating system:

[0932] 1. A limited sample operating system, comprising:

[0933] A sample holder;

[0934] Wherein the sample holder has a socket for receiving a sample card.

[0935] 2. The limited sample operating system according to any of the foregoing embodiments, wherein: the accuracy of positioning the sample in the direction along the optical axis is worse than 0.1um, 1um, 10um, 100um, 1mm, or within the range between any two values.

[0936] 3. The limited sample operating system according to any of the foregoing embodiments, wherein: the preferred accuracy of positioning the sample in the direction along the optical axis is between 50um and 200um.

[0937] 4. The limited sample operating system according to any of the foregoing embodiments, wherein: the accuracy of positioning the sample in a plane perpendicular to the optical axis is worse than 0.01um, 0.1um, 1um, 10um, 100um, 1mm, or within the range between any two values.

[0938] 5. The limited sample operating system according to any of the foregoing embodiments, wherein: the preferred accuracy of positioning the sample in a plane perpendicular to the optical axis is between 100um and 1mm.

[0939] 6. The limited sample operating system according to any of the foregoing embodiments, wherein the horizontal error of positioning the sample card is worse than 0.01 degree, 0.1 degree, 0.5 degree, 1 degree, 10 degrees, or within the range between any two values.

[0940] 7. The limited sample operating system according to any of the foregoing embodiments, wherein the preferred horizontal error of positioning the sample card is between 0.5 degrees and 10 degrees.

[0941] 8. The limited sample operating system according to any of the foregoing embodiments, wherein the preferred horizontal error of positioning the sample card is between 0.5 degrees and 10 degrees.

[0942] Monitoring the sharp edge of the marker

[0943] The term "sharp optical edge" refers to the edge or boundary of an object in an image captured by a specific optical system having the following properties: at the edge or boundary of the object in the image, the light intensity (including but not limited to R, G, B, gray, hue, brightness) changes sharply with respect to position. For example, quantitatively, as shown in Figure 1, at the boundary, the distance (X 90％ ) where the normalized intensity decreases from 90% to 10% should be less than 5% of the length of the object in the image by such an optical system. In other words, the boundary gradient = 90% / X 90％ / 1800% / object length. An example of a sharp edge is the edge of a columnar spacer having a flat top and sidewalls approaching 90 degrees. An example of an object without a sharp edge is a sphere.

[0944] Examples of using the TLD and volume estimation of the monitoring marker

[0945] Figure 5 shows an example of a sample holding device, the QMAX device, and its monitoring markers and columns used in some embodiments of the present invention. The column in the QMAX device makes the gap between two parallel plates of the sample holding device uniform. The gap is narrow and related to the size of the analyte, where the analyte forms a monolayer in the gap. In addition, the monitoring markers in the QMAX device are a special form of columns, so they are not submerged by the sample and can be imaged together with the sample by an imager in an image-based assay.

[0946] Example of estimating the true lateral dimension (TLD) using monitoring markers

[0947] For TLD and true volume estimation in some embodiments of the present invention. The monitoring markers (columns) are used as detectable anchors. However, it is difficult to detect the monitoring markers in an image-based assay with an accuracy suitable for TLD estimation. This is because these monitoring markers are penetrated and surrounded by the analyte within the sample holding device, and they are distorted and blurred in the image due to lens distortion, light diffraction of micro-objects, defects at the micro-level, misalignment of focus, noise in the sample image, etc. It becomes even more difficult if the imager is a camera from a commercial device (e.g., a camera from a smartphone), because these cameras are not calibrated by dedicated hardware once they leave the manufacturing process.

[0948] In the present invention, the detection and localization of the monitoring markers, which are used as detectable anchors for TLD estimation, are formulated in a machine learning framework, and a dedicated machine learning model is constructed / trained to detect them in microscopy imaging. In addition, in some embodiments of the present invention, the distribution of the monitoring markers is intentionally set to be periodic and distributed in a predetermined pattern. This makes the method of the present invention more robust and reliable.

[0949] Specifically, embodiments of the present invention include:

[0950] (8) Loading a sample into a sample holding device (e.g., a QMAX device), where there are monitoring markers with a known structure in the device that are not submerged in the sample and can be imaged by an imager;

[0951] (9) Taking an image of the sample in the sample holding device that includes the analyte and the monitoring markers;

[0952] (10) Constructing and training a machine learning (ML) model to detect the monitoring markers in the image of the sample;

[0953] (11) Using the ML detection model from (3) to detect and locate the monitoring markers in the sample holding device from the image of the sample;

[0954] (12) Generate a marker grid based on the monitoring markers detected in (4);

[0955] (13) Calculate the homography transformation based on the generated monitoring marker grid;

[0956] (14) Estimate and save the true lateral dimension of the sample image according to the homography transformation from (6); and

[0957] (15) Apply the estimated TLD from (7) in subsequent image-based determinations to determine the area, size, volume, and concentration of the analyte.

[0958] In some embodiments of the present invention, region-based TLD estimation and calibration are employed in image-based determinations. It includes:

[0959] (13) Load the sample into a sample holding device (such as a QMAX device), where there are monitoring markers in the device that are not submerged in the sample and can be imaged by an imager in the image-based determination;

[0960] (14) Take an image of the sample in the sample holding device including the analyte and the monitoring markers;

[0961] (15) Construct and train a machine learning (ML) model for detecting monitoring markers from the image of the sample taken by the imager;

[0962] (16) Divide the sample image taken by the imager into non-overlapping regions;

[0963] (17) Use the ML model of (3) to detect and locate the monitoring markers from the image of the sample taken by the imager;

[0964] (18) Generate a region-based marker grid for each region where more than 5 non-collinear monitoring markers are detected in the local region;

[0965] (19) Generate a marker grid for all regions not in (6) based on the monitoring markers detected from the image of the sample taken by the imager;

[0966] (20) Calculate a region-specific homography transformation for each region in (6) based on its own region-based marker grid generated in (6);

[0967] (21) Calculate the homography transformation for all other regions based on the marker grid generated in (7);

[0968] (22) Estimate the region-based TLD of each region in (6) based on the region-based homography transformation generated in (8);

[0969] (23)Estimate the TLD of other regions based on the homography transformation from (9); and

[0970] (24)In subsequent image-based determinations, save and apply the estimated TLD from (10) and (12) on the partition

[0971] When the monitoring marks are distributed in a predefined periodic pattern, such as in the QMAX device, they appear and are distributed periodically at a certain interval. As a result, in the above process, the detection of the monitoring marks becomes more robust and reliable. This is because, if the detected positions and configurations do not follow the predefined periodic pattern, by using periodicity, all the monitoring marks can be identified and determined from only a small number of detected monitoring marks, and detection errors can be corrected and eliminated.

[0972] Utilize the volume estimation of the monitoring marks

[0973] Figure 3 is a block diagram of the present invention for volume estimation in image-based determination. To estimate the true volume of the sample in the determination, it is necessary to remove the volume from the defects and surrounding objects in the sample. Embodiments of the present invention include:

[0974] (1) Load the sample onto a sample holding device with multiple monitoring marks (such as a QMAX device), and capture an image of the sample through an imager;

[0975] (2) Estimate the true lateral dimension (TLD) of the image by detecting the monitoring marks in the sample image as anchors using the above methods and devices;

[0976] (3) Use a machine learning (ML) detection model constructed and trained from the image of the sample to detect and locate defects in the image of the sample, such as bubbles, dust, monitoring marks, etc.;

[0977] (4) Use a machine learning (ML) segmentation model constructed and trained from the image of the sample to determine the coverage mask of the detected defects in the image of the sample;

[0978] (5) Determine the margin distance Δ (e.g., the maximum analyte diameter) based on the dimensions of the analyte / object in the sample and the sample holding device;

[0979] (6) Determine the Δ+ mask for all detected defects, that is, a mask with an additional margin Δ extending from (4) to its coverage mask;

[0980] (7) Remove the detected defects in the sample image based on the Δ+ mask in (6);

[0981] (8) In subsequent image-based measurements, the images from (7) are saved, and the corresponding volumes are saved after removing the volumes with the corresponding Δ+ masks of (7); and

[0982] (9) If the area of the removed Δ+ mask or the remaining true volume of the sample exceeds a certain preset threshold, the sample is rejected.

[0983] In the present invention, an additional margin Δ+ mask is used to remove defects from the image of the sample. This is important because defects can affect their surrounding environment. For example, some defects can change the height of the gap in the sample holding device, and the local volume or concentration distribution around the defect can become different.

[0984] The terms "monitoring mark", "monitor mark", and "mark" are interchangeable in the description of the present invention.

[0985] The terms "imager" and "camera" are interchangeable in the description of the present invention.

[0986] The term "denoising" refers to the process of removing noise from a received signal. An example is removing noise from the image of a sample, because the image from the imager / camera can pick up noise from various sources, including but not limited to white noise, salt and pepper noise, Gaussian noise, etc. Denoising methods include but are not limited to: linear and non-linear filtering, wavelet transform, statistical methods, deep learning, etc.

[0987] The term "image normalization" refers to algorithms, methods, and devices for changing the range of pixel intensity values in a processed image. For example, it includes but is not limited to increasing the contrast by histogram stretching, subtracting the average pixel value from each image, etc.

[0988] The term "image sharpening" refers to the process of enhancing the edge contrast and edge content of an image.

[0989] The term "image scaling" refers to the process of resizing an image. For example, if the object in the image of the sample is too small, image scaling can be applied to enlarge the image to assist in detection. In some embodiments of the present invention, the image needs to be resized to a specified size before being input into a deep learning model for training or inference purposes.

[0990] The term "alignment" refers to transforming different data sets into a common coordinate system for comparing and combining them. For example, in image processing, different data sets come from but are not limited to images from multiple imager sensors, and images from the same sensor but at different times, focus depths, etc.

[0991] The term "super-resolution" refers to the process of obtaining a higher-resolution image from one or more low-resolution images.

[0992] The term "deblurring" refers to the process of removing blurring artifacts from an image, such as blurring caused by defocus, jitter, motion, etc. during the imaging process.

[0993] In some embodiments of the present invention, methods and algorithms are designed to utilize monitoring markers in a sample holding device (e.g., a QMAX device). This includes, but is not limited to, the estimation and adjustment of the following parameters in an imaging device:

[0994] 1. Shutter speed,

[0995] 2. ISO,

[0996] 3. Focus (lens position),

[0997] 4. Exposure compensation,

[0998] 5. White balance: temperature, color, and

[0999] 6. Zoom (scale factor).

[1000] Examples of image processing / analysis algorithms used with markers

[1001] In some embodiments of the present invention, the monitoring markers in the present invention are used to apply and enhance image processing / analysis. They include, but are not limited to, the following image processing algorithms and methods:

[1002] 1. Histogram-based operations include, but are not limited to:

[1003] a. Contrast stretching;

[1004] b. Equalization;

[1005] c. Minimum filtering;

[1006] d. Median filtering; and

[1007] e. Maximum filtering.

[1008] 2. Math-based operations include, but are not limited to:

[1009] a. Binary operations: NOT, OR, AND, XOR, SUB, etc., and

[1010] b. Arithmetic-based operations: ADD, SUB, MUL, DIV, LOG, EXP, SQRT, TRIG, INVERT, etc.

[1011] 3. Convolution-based operations in the spatial and frequency domains include, but are not limited to, Fourier transform, DCT, integer transform, wavelet transform, etc.

[1012] 4. Smoothing operations include, but are not limited to:

[1013] a. Linear filtering: uniform filtering, triangular filtering, Gaussian filtering, etc., and

[1014] b. Nonlinear filtering: median filtering, kuwahara filtering, etc.

[1015] 5. Derivative-based operations include, but are not limited to:

[1016] a. First-order derivatives: gradient filtering, basic derivative filtering, Prewitt gradient filtering, Sobel gradient filtering, alternative gradient filtering, Gaussian gradient filtering, etc.;

[1017] b. Second-order derivatives: basic second-order derivative filtering, Laplacian in the frequency domain, Gaussian second-order derivative filtering, alternative Laplacian filtering, second-order derivative of gradient direction (SDGD) filtering, etc., and

[1018] c. Other filtering with higher-order derivatives, etc.

[1019] 6. Morphology-based operations include, but are not limited to:

[1020] a. Dilation and erosion;

[1021] b. Boolean convolution;

[1022] c. Opening and closing;

[1023] d. Hit and miss operations;

[1024] e. Segmentation and contour;

[1025] f. Skeleton;

[1026] g. Propagation;

[1027] h. Gray-scale value morphology processing: gray-scale dilation, gray-scale erosion, gray-scale opening, gray-scale closing, etc.; and

[1028] i. Morphological smoothing, morphological gradient, morphological Laplacian, etc.

[1029] Other examples of image processing / analysis techniques

[1030] In some embodiments of the present invention, the image processing / analysis algorithms are used in conjunction with and enhanced by the monitoring markers described in this disclosure. They include, but are not limited to, the following:

[1031] 1. Image enhancement and restoration include, but are not limited to

[1032] a. Sharpening and unsharpening,

[1033] b. Noise suppression, and

[1034] c. Distortion suppression.

[1035] 2. Image segmentation includes but is not limited to:

[1036] a. Thresholding - fixed threshold, histogram-derived threshold, Isodata algorithm, background symmetry algorithm, triangle algorithm, etc.;

[1037] b. Edge finding - gradient-based process, zero-crossing-based process, PLUS-based process, etc.;

[1038] c. Binary mathematical morphology - salt-and-pepper filtering, separating objects with holes, filling holes in objects, removing boundary-touching objects, outer skeleton, touching objects, etc.; and

[1039] d. Gray-scale value mathematical morphology - top-hat transform, adaptive thresholding, local contrast stretching, etc.

[1040] 3. Feature extraction and matching includes but is not limited to:

[1041] a. Independent component analysis;

[1042] b. Isometric mapping;

[1043] c. Principal component analysis and kernel principal component analysis;

[1044] d. Latent semantic analysis;

[1045] e. Least squares and partial least squares;

[1046] f. Multi-factor dimensionality reduction and non-linear dimensionality reduction;

[1047] g. Multilinear principal component analysis;

[1048] h. Multilinear subspace learning;

[1049] i. Semidefinite embedding; and

[1050] j. Autoencoder / decoder.

[1051] 4. Object detection, classification, and localization

[1052] 5. Image understanding

[1053] Example E1. Improvement of microscopic imaging using a monitoring marker

[1054] Improving focus in microscopic imaging using the monitoring marker in the present invention:

[1055] Markers with sharp edges will provide detectable (visible features) for the focus evaluation algorithm to analyze the focus conditions of a particular focusing setting, especially in low-light environments and microscopy imaging. In some embodiments of the present invention, the focus evaluation algorithm is located in the core part of the autofocus implementation as shown in FIG. E1-1.

[1056] For some diagnostic applications (e.g., colorimetry, absorption-based hemoglobin tests, and CBC for samples with very low cell concentrations), the detectable features provided by the analytes in the image of the sample are usually not sufficient for the focus evaluation algorithm to operate accurately and smoothly. Markers with sharp edges, such as the monitoring markers in the QMAX device, provide additional detectable features for the focus evaluation procedure to achieve the accuracy and reliability required in image-based assays.

[1057] For some diagnostic applications, the analytes in the sample are unevenly distributed. Relying solely on the features provided by the analytes often results in some unfair focusing settings, which give high weights to the focusing of some locally high-concentration regions while the low analyte concentration regions deviate from the target. In some embodiments of the present invention, this effect is controlled by focus adjustment according to the information of the monitoring markers with strong edges and evenly distributed in a precisely processed periodic pattern.

[1058] Generate a higher-resolution image using super-resolution of a single image.

[1059] Each imager has an imaging resolution that is partially limited by the number of pixels in its sensor, which varies from one million to several million pixels. For some microscopy imaging applications, the analytes have small or tiny sizes in the sample, e.g., the size of platelets in human blood is about 1.4 um. When the target detection procedure requires a certain number of pixels, in addition to the available size of the FOV, the limited resolution in the image sensor poses a significant constraint on the device's ability in image-based assays.

[1060] Single-image super-resolution (SISR) is a technique that uses image processing and / or machine learning techniques to upsample the original source image to a higher resolution and remove the blur caused by interpolation as much as possible, so that the object detection procedure can also run on the newly generated image. This will significantly reduce the above constraints and enable some other impossible applications. Markers with known shapes and structures (such as the monitoring markers in QMAX) can be used as local references to evaluate the SISR algorithm to avoid the oversharpening effect produced by most existing technology algorithms.

[1061] In some embodiments of the present invention, image fusion is performed to break the physical SNR (signal-to-noise ratio) limit in image-based assays.

[1062] Signal-to-noise ratio (SNR) measurements are made on the quality of the images of a sample taken by an imager in microscopy. Due to cost, technology, manufacturing, etc., there are practical limitations in imaging devices. In some cases, such as in mobile healthcare, the application requires a higher SNR than what the imaging device can provide. In some embodiments of the present invention, multiple images (with the same and / or different imaging settings, e.g., embodiments of 3D fusion for combining multiple images focused at different depths of focus into one super-focused image) are taken and processed to generate an output image with a higher SNR, thus enabling these applications.

[1063] However, images taken by one imager or multiple imagers tend to have some drawbacks and defects caused by physical limitations and implementation constraints. In image-based assays, this situation becomes severe in the microscopy of samples because the analytes in the sample have tiny sizes and often no distinct edge features. In some embodiments of the present invention, monitoring marks in an image holding device (such as a QMAX device) are used for enhanced solutions.

[1064] One such embodiment is to process the distortion in the image of a sample taken by an imager. Figure E1-2 is a diagram of an improvement to the general camera model. When the distortion parameters are known, the situation is relatively simple (most are manufactured to give curves / tables for their lenses to describe scale distortion, and other distortions can be measured in well-defined experiments). However, when the distortion parameters are unknown (in our case, it varies with the focus position and even the sample), using monitoring marks, a new algorithm can iteratively estimate the distortion parameters using regularly or even periodically placed monitoring marks in a sample holding device (such as a QMAX device), without the need for a single coordinate reference as shown in Figure E1-3.

[1065] Figure E1-3 is a diagram of distortion removal and camera refinement when unknown

[1066] In the present invention, in some embodiments, the sample holding device has a flat surface that has some special monitoring marks for analyzing micro-features in image-based assays.

[1067] Some exemplary embodiments are listed below:

[1068] A1: Estimation of the True Lateral Dimension (TLD) of a microscopic image of a sample in an image-based assay. The True Lateral Dimension (TLD) determines the physical size of the imaged analyte in the real world and also determines the image coordinates of the sample in the real world related to concentration estimation in the image-based assay. Monitoring markers can be used as detectable anchors to determine the TLD and improve the accuracy of the image-based assay. In an embodiment of the present invention, a machine learning model for the monitoring marker is used to detect the monitoring marker and derive the TLD of the sample image from it. Additionally, if the monitoring marker has a periodic distribution pattern on the flat surface of the sample holding device, the detection of the monitoring marker and the TLD estimation based on each sample can become more reliable and robust in the image-based assay.

[1069] A2: Analyzing the analyte using the response of the measured analyte compound at a specific light wavelength or multiple light wavelengths to predict the analyte concentration. Monitoring markers not immersed in the sample can be used to determine the light absorption corresponding to the background without the analyte compound - to determine the analyte concentration by light absorption, such as in the HgB test in a whole blood test. Additionally, each monitoring marker can serve as an independent detector of background absorption to make the concentration estimation robust and reliable.

[1070] A3: Focusing on the microscopic image of the image-based assay. Uniformly distributed monitoring markers can be used to improve the focusing accuracy. (a) It can be used to provide a minimum amount of visual features for a sample that does not have / has fewer features than required for reliable focusing, and this can be performed under low light due to the edge content of the monitoring marker. (b) It can be used to provide visual features when the feature distribution in the sample is uneven to make the focusing determination more fair. (c) It can provide a reference for local illumination conditions that have no / less / different impact on the content of the sample to adjust the weights in the focus evaluation algorithm.

[1071] A4: Monitoring markers can be used as a reference for detecting and / or correcting image defects, which are caused by but not limited to: unevenly distributed illumination, various types of image distortion, noise, and defective image preprocessing operations. (a) For example, as shown in Figure E1-4, when a straight line in the 3D world is mapped to a curve in the image, the position of the marker can be used to detect and / or correct the scale distortion. The scale distribution parameters of the entire image can be estimated based on the position change of the marker. And by performing a linear test on the horizontal / vertical lines in the reproduced image through distortion elimination based on the assumed scale distortion parameters, the value of the scale distortion parameters can be iteratively estimated.

[1072] Examples of Machine Learning (ML) Calculations

[1073] B1: One way to use machine learning is to use a trained machine learning model during the inference process of processing to perform the analysis of analytes in an image of a sample and calculate the bounding boxes covering their locations. Another way to use machine learning methods to detect and locate analytes in a sample image is to construct and train a detection and segmentation model, which involves annotating analytes in the sample image at the pixel level. In this method, in an image-based assay, a tight binary pixel mask covering the analytes in the sample image can be used to detect and locate analytes in the sample image.

[1074] B2: When measuring hemoglobin in human blood, an image is taken at a given narrowband wavelength, and then the average energy passing through the analyte region and the reference region is analyzed. Based on the known absorption rate of the analyte at the given wavelength and the height of the analyte sample region, the concentration can be estimated. However, this measurement has noise. To eliminate the noise, multiple images can be taken using light of different wavelengths and machine learning regression can be used to achieve a more accurate and robust estimate. Machine learning-based inference takes multiple input images of the sample, acquired at different wavelengths, and outputs a single concentration value.

[1075] E. Examples of Identifying Error Risks to Improve Measurement Reliability

[1076] In some embodiments, methods are provided for improving the reliability of an assay, the methods comprising:

[1077] (a) Imaging the sample on a QMAX card;

[1078] (b) Analyzing error risk factors; and

[1079] (c) If the error risk factor is higher than a threshold, rejecting the measurement result of the card and reporting the card;

[1080] where the error risk factor is one or any combination of the following factors. These factors are, but not limited to, (1) the edge of the blood, (2) air bubbles in the blood, (3) too small or too large a blood volume, (4) blood cells under the spacer, (5) aggregated blood cells, (6) lysed blood cells, (7) overexposed images of the sample, (8) underexposed images of the sample, (8) poor focus of the sample, (9) optical system errors such as incorrect lever position, (10) the card not being closed, (12) an incorrect card for a card without a spacer (12) dust in the card, (14) oil in the card, (14) dirt outside the focus plane of the card, (15) the card not being in the correct position inside the reader, (16) the card being empty, (17) manufacturing errors in the card, (18) an incorrect card for other applications, (19) dried blood, (20) an expired card, (21) large variations in the distribution of blood cells, (22) no blood sample or no target blood sample, and others.

[1081] In some embodiments, the error risk analyzer is capable of detecting, differentiating, classifying, modifying, and / or correcting the following in biological and chemical applications in the device: (1) at the sample edge, (2) air bubbles in the sample, (3) too small or too large a sample volume, (4) the sample under the spacer, (5) aggregated samples, (6) lysed samples, (7) overexposed images of the sample, (8) underexposed images of the sample, (8) poor sample focus, (9) optical system errors such as misaligned levers, (10) the card not being closed, (12) the wrong card for a spacerless card (12) dust in the card, (14) oil in the card, (14) dirt outside the focus plane of the card, (15) the card not being in the correct position inside the reader, (16) the card being empty, (17) manufacturing errors in the card, (18) the wrong card for other applications, (19) dry samples, (20) expired cards, (21) large variations in blood cell distribution, (22) wrong samples, and others.

[1082] Wherein the threshold is determined from a group of tests.

[1083] Wherein the threshold is determined from machine learning.

[1084] Wherein the monitoring marker is used as a comparison to identify the error risk factor.

[1085] Wherein the monitoring marker is used as a comparison to evaluate the threshold of the error risk factor.

[1086] Other embodiments

[1087] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. Although any methods and materials similar or equivalent to those described herein can also be used to practice or test the present teachings, some exemplary methods and materials are now described.

[1088] The "QMAX" (Q: Quantification; M: Amplification; A: Reagent addition; X: Acceleration; also known as Self-Calibrated Compressed Open Flow (SCOF)) device, assay, method, kit, and

[1089] The system is described in U.S. Provisional Patent Application No. 62 / 202,989, filed on August 10, 2015; U.S. Provisional Patent Application No. 62 / 218,455, filed on September 14, 2015; U.S. Provisional Patent Application No. 62 / 293,188, filed on February 9, 2016; U.S. Provisional Patent Application No. 62 / 305,123, filed on March 8, 2016; and U.S. Provisional Patent Application No. 62 / 369,181, filed on July 31, 2016; U.S. Provisional Patent Application No. 62 / 394,753, filed on September 15, 2016; PCT Application (designating the United States) No. PCT / US2016 / 045437, filed on August 10, 2016; PCT Application (designating the United States) No. PCT / US2016 / 051775, filed on September 14, 2016; PCT Application (designating the United States) No. PCT / US2016 / 051794, filed on September 15, 2016; and PCT Application (designating the United States) No. PCT / US2016 / 054025, filed on September 27, 2016. All of these disclosures are hereby incorporated by reference in their entirety and for all purposes into this text.

[1090] As used herein, the term "sample" refers to a material or mixture of materials that contains one or more analytes or entities of interest. In some embodiments, a sample can be obtained from a biological sample, such as cells, tissues, body fluids, and feces. Body fluids of interest include, but are not limited to, amniotic fluid, aqueous humor, vitreous humor, blood (e.g., whole blood, fractionated blood, plasma, serum, etc.), breast milk, cerebrospinal fluid (CSF), cerumen (ear wax), chyle, chyme, endolymph, perilymph, feces, gastric acid, gastric juice, lymph fluid, mucus (including nasal drainage and sputum), pericardial fluid, peritoneal fluid, pleural fluid, pus, inflammatory secretions, saliva, sebum (skin oil), semen, sputum, sweat, synovial fluid, tears, vomit, urine, and exhaled condensate. In specific embodiments, a sample can be obtained from a subject, such as a human, and can be processed before being used for subject assays. For example, proteins / nucleic acids can be extracted from tissue samples before analysis, and the methods are known. In certain embodiments, a sample can be a clinical sample, e.g., a sample collected from a patient.

[1091] The term "analyte" refers to molecules (such as proteins, peptides, DNA, RNA, nucleic acids, or other molecules), cells, tissues, viruses, and nanoparticles having different shapes. In some embodiments, an "analyte" as used herein is any substance suitable for testing in the present method.

[1092] As used herein, "diagnostic sample" refers to any biological sample obtained from a subject that is a bodily byproduct such as a body fluid. A diagnostic sample can be obtained directly from the subject in liquid form or can be obtained from the subject by first placing the bodily byproduct in a solution such as a buffer. Exemplary diagnostic samples include, but are not limited to, saliva, serum, blood, sputum, urine, sweat, tears, semen, feces, breath, biopsy, mucus, and the like.

[1093] As used herein, "environmental sample" refers to any sample obtained from the environment. Environmental samples can include liquid samples from rivers, lakes, ponds, oceans, glaciers, icebergs, rain, snow, sewage, reservoirs, tap water, drinking water, etc.; solid samples from soil, compost, sand, rock, concrete, wood, brick, sewage, etc.; and gaseous samples from air, underwater thermal vents, industrial exhaust, vehicle exhaust, etc. Generally, samples in non-liquid form are converted to liquid form before being analyzed by the methods of the present invention.

[1094] As used herein, "food sample" refers to any sample suitable for animal consumption (e.g., human consumption). Food samples can include raw materials, cooked foods, plant and animal sources of food, pre-treated foods, and partially or fully processed foods, etc. Generally, samples in non-liquid form are converted to liquid form before being analyzed by the methods of the present invention.

[1095] As used herein, the term "diagnosis" refers to a method or analyte used to identify, predict the outcome of a disease or condition of interest, and / or predict the therapeutic response to a disease or condition of interest. Diagnosis can include predicting the likelihood or propensity of having a disease or condition, estimating the severity of a disease or condition, determining the risk of disease or condition progression, assessing the clinical response to treatment, and / or predicting the response to treatment.

[1096] As used herein, a "biomarker" is any molecule or compound found in a sample of interest and known to be diagnostic or associated with the presence or propensity of a disease or condition of interest in the subject from which the sample is derived. Biomarkers include, but are not limited to, polypeptides or their complexes (e.g., antigens, antibodies) known to be associated with a disease or condition of interest, nucleic acids (e.g., DNA, miRNA, mRNA), drug metabolites, lipids, carbohydrates, hormones, vitamins, and the like.

[1097] As used herein, "condition" with respect to diagnosing a health condition refers to a mental or physical physiological state that is distinct from other physiological states. In some cases, a health condition may not be diagnosed as a disease. Exemplary health conditions of interest include, but are not limited to, nutritional health; aging; exposure to environmental toxins, pesticides, herbicides, synthetic hormone analogs; pregnancy; menopause; climacteric; sleep; stress; prediabetes; exercise; fatigue; chemical balance; etc.

[1098] Other notes

[1099] It should be noted that, as used in this specification and the appended claims, the singular forms "a", "an", and "the" include plural referents unless the context clearly dictates otherwise, such as when the term "single" is used. For example, reference to "analyte" includes a single analyte and plural analytes, reference to "capture agent" includes a single capture agent and plural capture agents, reference to "detection agent" includes a single detection agent and plural detection agents, and reference to "reagent" includes a single reagent and plural reagents.

[1100] As used herein, the phrases "at least one" and "one or more" with respect to a list of more than one entity refer to any one or more of the entities in the list of entities and are not limited to at least one of each and every entity specifically listed in the list of entities. For example, "at least one of A and B" (or equivalently, "at least one of A or B", or equivalently, "at least one of A and / or B") can refer to A alone, B alone, or a combination of A and B.

[1101] As used herein, the term "and / or" placed between a first entity and a second entity refers to (1) the first entity, (2) the second entity, and (3) one of the first entity and the second entity. Multiple entities listed using "and / or" shall be interpreted in the same manner, i.e., "one or more" of the entities so combined. Optionally, there may be other entities in addition to those specifically identified by the "and / or" clause, whether or not they are related to those specifically identified.

[1102] When a numerical range is recited herein, the present invention includes embodiments that include the endpoints, embodiments that exclude both endpoints, and embodiments that include one endpoint and exclude the other. It should be assumed that both endpoints are included unless otherwise stated. Further, unless otherwise stated or apparent to one of ordinary skill in the art from the context and understanding.

[1103] In the event that any patent, patent application, or other reference is incorporated herein by reference and (1) defines a term in a manner inconsistent with the unincorporated portion of this disclosure or other incorporated references and / or (2) is otherwise inconsistent with the unincorporated portion of this disclosure or other incorporated references, the unincorporated portion of this disclosure shall govern, and the term or the incorporated disclosure therein shall be applicable only to the reference in which the term is defined and / or incorporated initially.

Claims

1. A method for improving the accuracy of determining an analyte in or suspected to be in a test sample using a measurement device, wherein the measurement device or the operation of the measurement device each has one or more randomly varying parameters, characterized in that: The measurement device has a sample holder; The method comprises: (a) Determining the analyte using the measurement device, comprising: (i) Placing the sample into the measurement device; (ii) Measuring the analyte in the sample using the measurement device to generate a test result; (b) Determining the credibility of the test result in step (a), comprising: (i) Imaging, using an imager to capture one or more images of at least a portion of the sample and / or at least a portion of the measurement device; The image represents the conditions for measuring at least a portion of the sample when generating the test result in step (a); (ii) Determining the credibility of the test result in step (a) and generating a credibility score by analyzing the image and one or more parameters using an algorithm, The one or more parameters include dust, bubbles, non-sample materials, or any combination thereof; The algorithm refers to a combined design fine image segmentation algorithm based on machine learning's rough bounding box segmentation and image processing's fine grinding shape determination; The fine image segmentation algorithm comprises: a) Collecting a plurality of sample images for training captured by the imager, which include objects to be detected in the images of the samples for determination; b) Marking each object in the collected images with a rough bounding box containing the object for model training; c) Training a machine learning model to detect the objects in the images of the samples having bounding boxes containing them; d) In the inference phase, taking the image of the sample to be determined as the input; e) Applying the trained machine learning model to detect the objects and locating them using the bounding boxes in the image of the sample; f) Transforming each image patch corresponding to the bounding box containing the detected object into gray, and then transforming it into binary using adaptive thresholding; g) Performing morphological dilation and erosion from the background noise to enhance the contour of the shape; h) Performing convex contour analysis on each of the image patches and using the longest connected contour found in the patch as the contour of the object shape to determine the image mask of the object; And i) Completing image segmentation by collecting all the image masks from (h); j) Using the margin Δ as a new mask to expand each detected contour in (h); The Δ is an additional margin for the detected object applied to the segmentation mask in image detection, used to reduce the negative impact of defects on adjacent local regions; and k) Completing image segmentation with the Δ margin by collecting all the enlarged image masks from j).

2. The method according to claim 1, wherein The measurement in step (a) includes analyzing one or more images of at least a portion of the sample and / or at least a portion of the measurement device.

3. The method according to claim 1, wherein (a) the measurement device comprises a sample holder having a sample contact area for contacting the sample; (b) The sample is placed in the sample contact area; (c) wherein the at least one sample contact area comprises one or more monitoring structures; (d) imaging at least a portion of the sample and at least a portion of the monitoring structure for confidence determination; and (e) the monitoring structure comprises a structure for monitoring the optical properties for monitoring the operation of the assay and / or the quality of the assay device.

4. The method according to claim 1, further comprising, when the confidence score is not credible, discarding the detection result and repeating steps (a) and (b) using a second assay device.

5. The method according to claim 1, wherein the algorithm is machine learning.

6. The method according to claim 1, wherein the sample comprises at least one of the parameters having random variations.

7. The method according to claim 1, wherein In steps (b)(i)-(b)(ii), the algorithm uses a threshold of the operating variable to determine whether the result is credible.

8. The method according to claim 1, further comprising the step of maintaining or rejecting the detection result using the confidence score.

9. The method according to claim 1, wherein the assay is a device for detecting the analyte using a chemical reaction.

10. The method according to claim 1, wherein the assay is an immunoassay, a nucleic acid assay, a colorimetric assay, a luminescence assay, or any combination thereof.

11. The method according to claim 1, wherein the assay device comprises a sample holder having two plates facing each other with a gap of 250 μm or less, and at least a portion of the sample is within the gap.

12. The method according to claim 11, wherein the assay device comprises a sample holder comprising two plates movable relative to each other and a spacer for adjusting the spacing between the plates, and at least a portion of the sample is within the gap.

13. The method according to claim 3, wherein some of the monitoring structures are arranged periodically.

14. The method according to claim 1, wherein the sample is selected from cells, tissues, body fluids or feces.

15. The method according to claim 1, wherein the sample is amniotic fluid, aqueous humor, vitreous humor, blood, breast milk, cerebrospinal fluid (CSF), earwax, chyle, chyme, endolymph, perilymph, feces, gastric juice, lymph fluid, mucus, pericardial fluid, peritoneal fluid, pleural fluid, inflammatory secretions, saliva, sebum, semen, sputum, sweat, synovial fluid, tears, vomit, urine, and exhaled condensate; wherein the blood includes whole blood, fractionated blood, plasma, serum.

16. The method according to claim 1, wherein the analyte comprises molecules, cells, tissues, viruses and nanoparticles, wherein the molecules include proteins, peptides, DNA or RNA.

17. The method according to claim 1, wherein the sample is non-flowable but deformable.

18. The method according to claim 1, wherein the method further comprises the step of discarding the detection result generated in step (a) if the value determined in step (b) is below a threshold.

19. The method according to claim 1, wherein the algorithm is machine learning, artificial intelligence, statistical methods, or a combination thereof.

20. The method according to claim 1, wherein the measuring device comprises a sample holder having a first plate, a second plate, and a spacer, wherein the first plate and the second plate are movable relative to each other into different configurations, including an open configuration and a closed configuration; Among them, the open configuration, wherein the two plates are partially or fully separated, the spacing between the plates is not adjusted by the spacer, and the sample is deposited on one or both of the plates, and wherein the closed configuration is configured after the sample is deposited in the open configuration; in the closed configuration, at least a portion of the sample is compressed by the two plates into a layer of uniform thickness and is stationary relative to the plates, wherein the uniform thickness of the layer is defined by the sample contact areas of the two plates and is adjusted by the two plates and the spacer.

21. The method according to claim 1, wherein in steps (b)(i)-(b)(ii), the algorithm uses machine learning with a training set to determine whether the result is credible, wherein the training set uses operating variables of the analyte in the sample.

22. The method according to claim 1, wherein in steps (b)(i)-(b)(ii), the algorithm uses a look-up table to determine whether the result is credible, wherein the look-up table contains operating variables of the analyte in the sample.

23. The method according to claim 1, wherein in steps (b)(i)-(b)(ii), the algorithm uses a neural network to determine whether the result is credible, wherein the neural network is trained using operating variables of the analyte in the sample.

24. The method according to claim 1, wherein in steps (b)(i)-(b)(ii), the algorithm, look-up table or neural network is used to determine whether the result is credible, wherein the algorithm, look-up table or neural network contains operating variables, including the condition of bubbles and / or dust in the image of the portion of the sample.

25. The method according to claim 1, wherein in steps (b)(i)-(b)(ii), the algorithm uses machine learning, which determines whether the result is credible, using an algorithm, look-up table or neural network to determine the operating variables of bubbles and / or dust in the image of the portion of the sample.

26. The method according to claim 20, wherein (i) the device further comprises a monitoring marker; (ii) the device further includes a monitoring structure, which is used as a parameter in the algorithm together with the imaging processing method, which (1) adjusts the imaging, (2) processes the image of the sample, (3) determines the properties related to the micro-features, or (4) any combination of the above; (iii) the device further includes a monitoring structure, which is used as a parameter together with step (b); (iv) the device further includes a monitoring structure, wherein the spacer is the monitoring structure, wherein the spacer has a uniform height equal to or less than 200 microns, and a fixed spacer spacing (ISD); or (v) the device further includes a monitoring structure for estimating the TLD (true lateral dimension) and true volume estimation.

27. The method according to claim 1, wherein step (b) further comprises image segmentation for image-based determination.

28. The method according to claim 1, wherein step (b) further comprises focus inspection in image-based determination.

29. The method according to claim 1, wherein step (b) further comprises the uniformity of the analyte distribution in the sample.

30. The method according to claim 1, wherein step (b) further comprises analyzing and detecting aggregated analytes in the sample.

31. The method according to claim 1, wherein step (b) further comprises analyzing the dry texture in the image of the sample in the sample.

32. The method according to claim 1, wherein step (b) further comprises analyzing defects in the sample.

33. The method according to claim 1, wherein step (b) further comprises the correction of camera parameters and conditions, including distortion removal, temperature correction, brightness correction, and contrast correction.

34. The method according to claim 1, wherein step (b) further comprises operations based on histograms, operations based on convolutions, smoothing operations, operations based on derivatives, and operations based on morphology.

35. The method according to claim 1, wherein the detection device has a sample holder, and the method comprises; a) using an imager to capture an image of the sample on the region of interest (AoI) on the sample holder for determination; b) segmenting the image of the sample captured by the imager in (b) into equally sized and non-overlapping sub-image patches; c) performing machine learning-based inference using a trained machine learning model to perform analyte detection and segmentation on each image patch - to determine the analyte count and its concentration; d) sorting the analyte concentrations of the constructed sub-image patches in ascending order and determining its 25th percentile Q1 and 75th percentile Q3; e) determining the uniformity of analytes in the sample image using a confidence measure based on the interquartile range: confidence - IQR = (Q3 - Q1) / (Q3 + Q1); and f) reporting an error or defect and the measurement result is not credible if the confidence - IQR from (e) exceeds a certain threshold, wherein the threshold is from training / evaluation data or from the physical rules governing the distribution of analytes.

36. The method according to claim 1, wherein the determination device has a sample holder, and the method comprises: a) using an imager to capture an image of the sample on the region of interest (AoI) of the sample holder for determination; b) performing machine learning-based inference using a trained machine learning model to perform dry texture detection and segmentation - to detect the dry texture area and determine the dry texture area in the AoI associated with the segmentation contour mask covering the dry texture area in the AoI of those sample images; c) determining the area ratio between the dry texture area in the AoI and the area of the AoI: dry texture area ratio in AoI = dry texture area in AoI / area of AoI; and d) If the ratio of the dry texture area in the AoI from (c) exceeds a specific threshold, an indication of error or defect is reported and the measurement result is not credible, where the threshold is obtained from training / evaluation data or from physical rules governing the analyte distribution.

37. The method according to claim 36, wherein the threshold for the ratio of the dry texture area in the AoI is 10%, 15%, 30%, 50% or a value between any two of them.

38. The method according to claim 36, wherein the threshold for the air bubble gap area in the AoI is 10%, 15%, 30%, 50% or a value between any two of them.

39. The method according to claim 3, wherein the measuring device has a sample holder, and the method comprises: a) taking an image of the sample on the region of interest (AoI) on the sample holder using an imager for measurement; b) performing machine learning-based inference using a trained machine learning model to detect and segment the monitoring structure with the analyte on top, and determining the area analyte on the pillar in the AoI related to the detected monitoring structure according to the segmentation contour mask in the AoI; c) determining the area ratio between the area of the analyte on the pillar in the AoI and the area of the AoI: Area ratio of analyte in AoI = Aggregate analyte area in AoI / Area of AoI; and d) If the ratio of the area of the analyte on the pillar in the AoI from (c) exceeds a specific threshold, an indication of error or defect is reported and the measurement result is not credible, where the threshold is obtained from training / evaluation data or from physical rules governing the analyte distribution.

40. The method according to claim 39, wherein, The threshold for the ratio of the analyte on the pillar to the AoI area is 10%, 15%, 30%, 50% or a value between any two of them.

41. The method according to claim 39, wherein the detection and segmentation of the sample image taken by the imager in the image-based analysis is based on image processing, machine learning or a combination of image processing and machine learning.

42. The method according to claim 39, further comprising a monitoring structure built into the sample holder, and the monitoring structure pillar is applied as a detectable anchor to make the estimation of the true lateral dimension or the field of view (FoV) accurate in the face of distortion in microscopic imaging.

43. The method according to claim 39, wherein the monitoring structure of the sample holder has some configurations with a prescribed periodic distribution in the sample holder to make the detection and positioning of the monitoring markers as anchors in the estimation of the true lateral dimension (TLD) or the field of view (FoV) reliable and robust.

44. The method according to claim 39, further comprising detecting and characterizing outliers in the image-based analysis according to non-overlapping sub-image blocks of the sample input image, and the detection of outliers during the analysis is based on non-parametric methods, parametric methods and a combination of both.

45. The method according to claim 1, wherein the algorithm comprises A machine learning model, where training data is used to train the machine learning model, and the training includes (i) adjusting hyperparameters and the model structure with the training and evaluation data until the model achieves satisfactory performance on the evaluation and test data; and (ii) performing inference on the test data using the trained machine learning model.

Images