System and method for analyzing multiwell plates

Abstract

To provide a system and method for analyzing a multi-well plate that is used in analyzers.SOLUTION: A method includes: analyzing, using a machine learning model, a sample inserted in an analyzer to determine whether or not a plate insert matches a multi-well plate (MWP); in response to the machine learning model determining a mismatch between the plate insert and the MWP, generating an alert on the analyzer to notify a user of the mismatch; in response to the machine learning model determining that the plate insert matches the MWP, analyzing, using the machine learning model, the sample to determine whether or not the MWP is sealed with a foil; and, in response to the machine learning mode determining that the MWP is not sealed with the foil, generating an alert on the analyzer, and otherwise, enabling the analyzer to perform its analysis on the sample.SELECTED DRAWING: Figure 3

Description

Technical Field

[0005] ,

[0004] , , ,

[0001] Embodiments of the present disclosure generally relate to analyzers, and more specifically, to the analysis of multi-well plates used in analyzers.

Background Art

[0002] In some cases, the foil on a test sample such as a multi-well plate (MWP) may not be properly applied before passing the sample through an analyzer such as a polymerase chain reaction (PCR) analyzer, or an incorrect plate insert may be used for a given MWP. For example, a technician may fail to properly apply the foil to the MWP before running the sample on the analyzer. Additionally, some analyzers may be configured to analyze more than one type of MWP. For example, the MWP can be a 96 (96) or 384 (396) well plate. However, in some cases, a plate insert for a 96 well plate may be applied to a 384 well plate, and vice versa. If the foil or plate insert is not properly applied, the sample material may evaporate during the thermal cycle, and in that case, the temperature may reach 95°C. This can lead to a loss of the measured sample results and the analyzer itself may be contaminated by the sample. Since contaminants in the optical path can result in inappropriate results, the analyzer needs to be cleaned. Further, if the sample spills due to movement within the analyzer, it can lead to cross-contamination and thus incorrect results.

[0003] Therefore, it is desirable to provide a new system and method for recognizing whether the correct insert has been used or the foil has been properly applied to the sample before using the analyzer.

Summary of the Invention

[0004] Embodiments of the present disclosure generally relate to analyzers, and more specifically, to the analysis of multi-well plates used in analyzers.

[0005] In one embodiment, the method includes using a machine learning model to analyze a sample inserted into an analyzer to determine whether the plate insert matches a multiwell plate (MWP). The method further includes generating an alert in the analyzer to notify the user of the mismatch in response to the machine learning model determining a mismatch between the plate insert and the MWP. The method also includes using the machine learning model to analyze a sample to determine whether the MWP is foil-sealed in response to the machine learning model determining that the plate insert matches the MWP. The method further includes generating an alert in the analyzer in response to the machine learning model determining that the MWP is not foil-sealed, allowing the analyzer to perform that analysis of the sample if the MWP is foil-sealed.

[0006] In some embodiments, machine learning models are trained using a combination of real and augmented images.

[0007] In some embodiments, a machine learning model is trained using a training dataset containing images of MWPs with at least two different numbers of wells.

[0008] In some embodiments, images of an MWP having at least two different numbers of wells include images with and without foil on the MWP.

[0009] In some embodiments, machine learning models are trained using a training dataset that includes images of both correct and incorrect combinations of MWP and foil.

[0010] In some embodiments, a machine learning model is trained using a training dataset that includes images of MWPs with multiple filling volumes, multiple dyes, and multiple foil types.

[0011] In some embodiments, machine learning models are trained using a training dataset containing images of MWPs with multiple different filling patterns.

[0012] In some embodiments, machine learning models are trained using a training dataset that includes images of MWP with user errors.

[0013] In some embodiments, analyzing a sample to determine a mismatch between the plate insert and the MWP includes calculating a confidence score, and an alert is generated when the confidence score falls below a threshold.

[0014] In some embodiments, analyzing a sample to determine whether the MWP is foil-sealed includes calculating a confidence score, and an alert is generated when the confidence score falls below a threshold.

[0015] In some embodiments, the method also includes receiving instructions from the user to continue the operation of the analyzer when any of the alerts are generated.

[0016] In some embodiments, the machine learning model includes multiple machine learning submodels, where a first machine learning submodel is trained to determine whether a plate insert matches an MWP, and a second machine learning submodel is trained to determine whether a sample is sealed with foil.

[0017] In some embodiments, the first and second machine learning submodels are trained using different datasets.

[0018] In some embodiments, augmented images are generated using multiple augmentation techniques.

[0019] In another embodiment, the system includes memory and a processor coupled to the memory. The processor is configured to use a machine learning model to analyze a sample inserted into an analyzer and determine whether the plate insert matches a multiwell plate (MWP). The processor is further configured to generate an alert in the analyzer in response to a machine learning model mismatch between the plate insert and the MWP, notifying the user of the mismatch. The processor is also configured to use the machine learning model to analyze a sample in response to the machine learning model determining that the plate insert matches the MWP and determine whether the MWP is foil-sealed. The processor is also configured to generate an alert in the analyzer in response to the machine learning model determining that the MWP is not foil-sealed, allowing the analyzer to perform that analysis of the sample if the MWP is foil-sealed.

[0020] Novel features of this disclosure are specifically described in the following claims. A better understanding of the features and advantages of this disclosure can be obtained from the following detailed description, which specifies exemplary embodiments in which the principles of this disclosure are utilized, and from reference to the accompanying drawings. [Brief explanation of the drawing]

[0021] [Figure 1] This is a block diagram showing one embodiment of an optical system for an analyzer according to the present disclosure. [Figure 2] A diagram illustrating an example of a machine learning model relating to the aspects of this disclosure is provided. [Figure 3] This is a block diagram showing one embodiment of a method 300 for analyzing an MWP inserted into an analyzer, according to an aspect of the present disclosure. [Figure 4] This is a block diagram showing one embodiment of a computer system relating to the aspect of this disclosure. [Modes for carrying out the invention]

[0022] The disclosure described herein is a system and method based on a machine learning algorithm for analyzing a sample inserted into an analyzer.

[0023] FIG. 1 is a block diagram showing one embodiment of an analyzer according to an aspect of the present disclosure. For example, as shown in FIG. 1, the analyzer 1000 includes an optical system 100 and a computing system 150. In some embodiments, the optical system 100 includes an imaging system 110, first and second reflecting surfaces 115 and 120, a lens 125, and an imaging surface 130. In some embodiments, the imaging system 110 may include a light source 110a configured to generate a light beam, an illumination lens 110b configured to focus the light beam, and an exciter 110c configured to transmit the focused light beam onto the first reflecting surface 115. In some embodiments, the collected light beam is reflected from the first reflecting surface 115 to the second reflecting surface 120 and then transmitted to the imaging surface 130 via the lens 125.

[0024] Next, the light is reflected from the imaging surface 130 through the lens 125 and then reflected from the first and second reflecting surfaces 115, 120 to the imaging system 110. In some embodiments, the imaging system 110 may further include an emitter 110d configured to receive the reflected light from the imaging surface 130, an imaging lens 110e configured to collect the reflected light, and a camera 110f configured to capture the reflected light from the imaging surface 130. In some embodiments, the camera 110f can be used for fluorescence imaging due to its high sensitivity, low noise, and high temporal stability.

[0025] In some embodiments, computing system 150 may execute machine learning model 155 to determine 1) whether a plate insert matches the MWP and 2) whether the foil has been correctly applied to the MWP. For example, machine learning model 155 may analyze an image captured by camera 110f based on epi-fluorescence imaging, which is a dark-field method for achieving a high signal-to-background ratio. Epi-fluorescence imaging is based on analyzing bright images of positive samples that are lit in front of a dark scene.

[0026] FIG. 2 depicts a diagram illustrating an example of a machine learning model according to an aspect of the present disclosure. In some embodiments, machine learning model 200 may include a first machine learning sub-model 210 and a second machine learning sub-model 220. The first machine learning sub-model 210 may be trained to identify non-matching plates, and the second machine learning sub-model 220 may be trained to identify whether the foil has been properly applied to the MWP. In some embodiments, machine learning model 200 may be a supervised machine learning model. For example, the first and second machine learning sub-models 210, 220 may be convolutional neural networks trained to perform object classification. That is, object classification can be regarded as a binary classification problem in each of the first and second machine learning sub-models 210, 220, that is, whether the plate is correct in the first machine learning sub-model 210, and whether the foil has been correctly applied in the second machine learning sub-model 220).

[0027] In some embodiments, the machine learning submodels 210,200 may be trained using a variety of conditions. For example, some conditions for generating deeper data variability include MWPs with different numbers of sample wells, e.g., MWPs with a 96-well format and / or MWPs with a 384-well format; MWPs with and without sealing foil; MWPs with different types of sealing foil; MWPs with different batches of samples; borderline cases, e.g., MWPs with bubbles in the fluid; MWPs with different volumes in the wells; images of MWPs with different settings in analyzer 1000 (e.g., PCR excitation and emission filters); images captured using different types of analyzers; and images of MWPs with both correct and incorrect plate combinations.

[0028] In some embodiments, the first machine learning submodel 210 and the second machine learning submodel 220 may be trained using different images from the training dataset. For example, images with mismatched plates used to train the first machine learning submodel 210 may be excluded from the training dataset used to train the second machine learning submodel 220. Similarly, images with different foil applications used to train the second machine learning submodel 220 may be excluded from the training dataset used to train the first machine learning submodel 210. In some embodiments, with respect to the machine learning submodels 210,200, the images in the training dataset containing a 96-well MWP may also include one or more 8-well inserts.

[0029] In some embodiments, with respect to the first machine learning submodel 210, the ground truth may be 0 (zero) for images depicting mismatched plates and 1 (one) for images with matching plates. In further embodiments, with respect to the second machine learning submodel 220, the ground truth may be 0 (zero) for images with improperly applied foil and 1 (one) for images with properly applied foil.

[0030] In some embodiments, the training dataset may also include images with any number of filling patterns. That is, the image may include several wells filled with samples, while other wells remain empty with the designed patterns. For example, the training dataset may include two, four, six, or eight different designed patterns. These are merely illustrative filling patterns, and it should be understood by those skilled in the art that other filling patterns may be used in accordance with aspects of this disclosure.

[0031] In some embodiments, machine learning submodels 210,220 may be trained using a combination of both real images and simulated (or augmented) images. In some embodiments, data augmentation may be used during preprocessing of real images to mimic the remaining different filling patterns. Augmented images may be created using several different augmentation techniques to introduce variability across the training data. For example, a first technique may be used to black out individual rectangles of varying sizes from a real image. As another example, a second technique may be used to black out small squares of varying numbers and angles scattered across the real image. In a further example, a third technique may be used to remove individual rectangles of varying sizes from the first image and replace them with rectangles of the same size from a second image. In yet another example, a fourth technique may be used to use a weighted average of the first and second images with randomly varying weights. In some embodiments, two or more augmentation techniques may be combined with each other. For example, the third and fourth techniques may be used to combine two real images from the training dataset to create a new batch of two images. In some embodiments, thousands of images were generated using the data augmentation techniques described herein.

[0032] In some embodiments, when combining two or more augmentation techniques, a pair of real images may have a common ground truth. To achieve this, augmented images may be created using two images from a common training dataset, for example, both images may depict a matched plate and a mismatched plate. Thus, the ground truth of the augmented images may be the same as that of the original images.

[0033] Furthermore, images with eight well inserts may be processed separately during data augmentation. That is, in any batch of two images, one image without eight well inserts was paired with one image with eight well inserts before data augmentation was applied.

[0034] In a further embodiment, the images in the training dataset used for the second machine learning submodel 200 may include images with user errors, including but not limited to wrinkles, wobbles, and bubbles. Furthermore, the images in the training dataset for either the first machine learning submodel 210 and / or the second machine learning submodel 200 may include modifications to the MWP to simulate variations across different instrument and plate configurations by raising each of the four corners of the MWP and / or sliding the MWP in a horizontal plane during the experiment. Furthermore, the images in the training dataset for either the first machine learning submodel 210 and / or the second machine learning submodel 200 may include images with physical changes to the MWP. As some examples, the physical changes may include placing adhesive material, e.g., tape, and / or solid covers, e.g., plastic layers, across sections of the MWP. By modifying the images used in the training set as described herein, the machine learning model 200 may be implemented on different analyzers, e.g., analyzers of different models, analyzers from different manufacturers, or analyzers of the same type but with different serial numbers.

[0035] In some embodiments, the images used in the training dataset may be preprocessed. Preprocessing may include scaling the pixel values ​​of the raw image to convert the raw image from a 16-bit image to a pixel value of 255 or less. Furthermore, preprocessing may include removing overexposed portions of the real image. For example, with respect to blurred portions of a given real image, preprocessing may include removing those blurred sections. Furthermore, a subset of pixels may be used as input to the model. For example, with respect to both 96MWP and 384MWP, the boundary region of a given image may be removed. This may be achieved by cropping one or more edges of the image, and the remaining central rectangle may be subdivided into nine equal rectangles. In some embodiments, a single region may be used to train the machine learning model 200, and the remaining region may be discarded. The region used to train the machine learning model may be downsampled to a 224x224 square using linear interpolation to make it the correct format for use with the machine learning model 200. In some embodiments, overexposure correction may be performed before any modifications to the image are made, for example, before any image cropping or image augmentation.

[0036] In some embodiments, the first and second machine learning submodels 210,220 may calculate confidence scores. For example, the confidence scores may be on a scale of 0.0 to 1.0. If the confidence score falls below a first threshold level, for example 0.8, the first and second machine learning submodels 210,220 may generate an alert, for example a visual or audible alert. If the confidence score falls below a second threshold, for example 0.5, the first and second machine learning submodels 210,220 may generate a command to prevent the analyzer 1000 from analyzing the sample, thereby preventing any potential contamination of the analyzer 1000.

[0037] During operation, the analyzer 1000 is configured to analyze either a 96-well MWP or a 384-well MWP so that the first machine learning submodel 210 recognizes which type of MWP is being analyzed. Once the MWP is inserted, the analyzer 1000 captures an image of the MWP using the optical system 100. The first machine learning submodel 210 then analyzes the captured image to determine whether the intermediate plate matches the upper and lower plates, for example, whether there is a plate mismatch. Several indicators of mismatch may include, but are not limited to, partial circles, such as semicircles or crescent shapes, based on the angle of the MWP relative to the optical system 100, caused by a mismatch between the upper and intermediate layers. That is, the first machine learning submodel 210 analyzes the wells of the MWP to identify semicircles or crescent shapes with inaccurate angles in the captured image, thereby indicating a mismatch between the upper layer and the plate insert.

[0038] In some embodiments, if the first machine learning submodel 210 determines that there is a plate mismatch, the analyzer 1000 may generate an alert, for example, a visual or auditory alert. Conversely, if the first machine learning submodel 210 determines that there is no plate mismatch, the second machine learning submodel 220 further analyzes the MWP to determine whether the foil was properly applied. For example, using the same capture image, the second machine learning submodel 220 analyzes each well of the MWP to definitively determine that the foil was properly applied. To this end, the second machine learning submodel 220 analyzes the intensity of the light reflected in each well. In some embodiments, a decrease in light intensity may indicate that the foil was not properly applied. In other words, if the foil is not stretched over the MWP, the intensity of the light reflected from the foil will be lower, thereby indicating that the foil was not properly applied. In some embodiments, if the second machine learning submodel 220 determines that the foil was not properly applied, the analyzer 1000 may generate an alert, for example, a visual or auditory alert.

[0039] This disclosure advantageously utilizes existing hardware components of the analyzer. That is, no hardware modifications are required to carry out the processes described herein. Furthermore, the processes described herein can be used without modifying consumables, such as foil and / or MWP.

[0040] Using the technologies described herein, this disclosure improves the functionality of analyzer 1000 and the computational resources required to perform MWP plate analysis. For example, this disclosure reduces the data required to run machine learning models to less than 10 megabytes, compared to over 20 megabytes for conventional machine learning models. Similarly, this disclosure reduces the runtime required to run machine learning models to less than 500 milliseconds, compared to over 2000 milliseconds for conventional machine learning models.

[0041] Figure 3 is a block diagram showing one embodiment of method 300 for analyzing an MWP inserted into an analyzer, according to aspects of the present disclosure. In step 310, method 300 may include using a machine learning model to analyze a sample inserted into an analyzer, for example, analyzer 1000 in Figure 1, to determine whether the plate insert matches the multiwell plate (MWP). For example, a first machine learning submodel, for example, the first machine learning submodel 210 in Figure 2, may analyze an image of the captured sample to determine whether the intermediate plate matches the upper and lower plates. Plate mismatch may occur if the image contains indicators that the plate insert and the upper layer are misaligned. For example, some indicators are, for example, semicircular or crescent-shaped subcircles as described herein, but are not limited to these.

[0042] In 320, method 300 may further include generating an alert in the analyzer to notify the user of the mismatch in response to the machine learning model determining a mismatch between the plate insert and the MWP. For example, the alert may include a visual alert and an audio alert.

[0043] In 330, method 300 may include using a machine learning model to analyze a sample to determine whether the MWP is sealed with foil, in response to the machine learning model determining that the plate insert matches the MWP. For example, using the same capture image, a second machine learning submodel, e.g., the second machine learning submodel 220 in Figure 2, may analyze each well of the MWP to definitively determine that the foil is properly applied. The second machine learning submodel 220 may also analyze the intensity of light reflected in each well, and a reduced light intensity may indicate that the foil is not properly applied.

[0044] In 340, method 300 further includes generating an alert in the analyzer in response to a machine learning model determining that the MWP is not sealed with foil, and enabling the analyzer 1000 to perform that analysis of the sample if the MWP is sealed with foil. For example, the alert may include a visual alert and an audible alert.

[0045] Figure 4 is a block diagram showing one embodiment of a computer system 400 configured to carry out one or more aspects of the present disclosure. For example, the machine learning model 200 and / or method 300 of Figure 3 may be carried out using the computer system 400.

[0046] As shown in Figure 4, the computing system 400 may include a processor 410, memory 420, storage device 430, and input / output device 440. The processor 410, memory 420, storage device 430, and input / output device 440 may be interconnected via a system bus 450. The processor 410 can process instructions for execution within the computing system 400. In some exemplary embodiments, the processor 410 may be a single-threaded processor. Alternatively, the processor 410 may be a multi-threaded processor. The processor 410 can process instructions stored in memory 420 and / or storage device 430 to display graphical information for a user interface provided via the input / output device 440.

[0047] Memory 420 is a computer-readable medium, such as volatile or non-volatile, that stores information within the computing system 400. Memory 420 can store data structures, for example, a configuration object database. Storage device 430 can provide persistent storage for the computing system 400. Storage device 430 may be a floppy disk device, a hard disk device, an optical disk device, a tape device, or other suitable persistent storage means. Input / output device 440 performs input / output operations for the computing system 400. In some exemplary embodiments, input / output device 440 includes a keyboard and / or a pointing device. In various embodiments, input / output device 440 includes a display unit for displaying a graphical user interface.

[0048] According to some exemplary embodiments, the input / output device 440 can perform input / output operations for network devices. For example, the input / output device 440 may include an Ethernet port or other network ports for communicating with one or more wired and / or wireless networks (e.g., a local area network (LAN), a wide area network (WAN), the Internet).

[0049] In some exemplary embodiments, the computing system 400 can be used to run various interactive computer software applications that can be used for organizing, analyzing, and / or storing various forms of data. Alternatively, the computing system 400 can be used to run any type of software application. These applications can be used to perform various functions, such as planning functions (e.g., generating, managing, and editing spreadsheet documents, word processing documents, and / or other objects), computing functions, communication functions, and so on. The application may include various add-in functions or may be a standalone computing product and / or function. When launched within the application, these functions can be used to generate a user interface provided via the input / output device 440. The user interface may be generated by the computing system 400 and presented to the user (e.g., on a computer screen monitor).

[0050] One or more aspects or features of the subject matter described herein can be realized in digital electronic circuits, integrated circuits, specially designed application-specific integrated circuits (ASICs), field-programmable gate array (FPGA) computer hardware, firmware, software, and / or combinations thereof. These various aspects or features may include implementation in one or more computer programs executable and / or interpretable on a programmable system which includes at least one programmable processor, which may be for special-purpose or general-purpose use, coupled to receive data and instructions from a storage system, at least one input device, and at least one output device, and to transmit data and instructions to the storage system, at least one input device, and at least one output device. The programmable system or computing system may include a client and a server. The client and the server are generally located far apart from each other and usually interact through a communication network. The client-server relationship arises from computer programs running on each computer having a client-server relationship with each other.

[0051] These computer programs, sometimes called programs, software, software applications, applications, components, or code, contain machine instructions for a programmable processor and may be implemented in high-level procedural and / or object-oriented programming languages, as well as / or assembly / machine languages. As used herein, the term “machine-readable medium” means any computer program product, apparatus, and / or device used to provide machine instructions and / or data to a programmable processor, including, for example, magnetic disks, optical disks, memory, and programmable logic devices (PLDs), and includes machine-readable medium that receives machine instructions as machine-readable signals. The term “machine-readable signals” means any signals used to provide machine instructions and / or data to a programmable processor. Machine-readable medium can store such machine instructions non-temporarily, for example, non-temporarily, solid-state memory, magnetic hard drives, or any equivalent storage medium. Machine-readable medium can, alternatively or additionally, store such machine instructions temporarily, for example, a processor cache or other random-access memory associated with one or more physical processor cores.

[0052] To provide user interaction, one or more aspects or features of the subject matter described herein may be implemented on a computer having, for example, a display device such as a cathode ray tube (CRT), liquid crystal display (LCD), or light-emitting diode (LED) monitor for displaying information to the user, and a keyboard, and a pointing device such as a mouse or trackball by which the user can provide input to the computer. Other types of devices may also be used to provide user interaction. For example, the feedback provided to the user may be any form of sensory feedback, such as visual feedback, auditory feedback, or tactile feedback, and input from the user may be received in any form, including acoustic input, voice input, or tactile input. Other possible input devices include touchscreens, or other touch-sensing devices such as single-point or multi-point resistive or capacitive trackpads, speech recognition hardware and software, optical scanners, optical pointers, digital image capture devices, and associated interpretation software.

[0053] When a feature or element is referred to herein as being "on top of" another feature or element, the feature or element may be directly on top of the other feature or element, or there may be intervening features and / or elements. In contrast, when a feature or element is referred to as being "directly" on another feature or element, there are no intervening features or elements. When a feature or element is referred to as being "connected," "attached," or "combined" to another feature or element, it will also be understood that it may be directly connected, attached, or combined with the other feature or element, or there may be intervening features or elements. In contrast, when a feature or element is referred to as being "directly connected," "directly attached," or "directly combined" with another feature or element, there are no intervening features or elements. Features and elements described or shown in relation to one embodiment may be applicable to other embodiments. It will also be understood by those skilled in the art that a structure or feature positioned "adjacent" to another feature may have portions that overlap with or lie beneath the adjacent feature.

[0054] The terms used herein are intended solely to describe specific embodiments and are not intended to limit the disclosure. For example, as used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context indicates otherwise. The terms “equipped with” and / or “equipped with” when used herein specify the presence of the described feature, step, action, element, and / or component, but are understood not to exclude the presence or addition of one or more other features, steps, actions, elements, components, and / or groups thereof. As used herein, the terms “and / or” include any and all combinations of one or more of the related enumerated items and may be abbreviated as “ / ”.

[0055] Spatially relative terms such as “under,” “below,” “lower,” “over,” and “upper” may be used herein to describe the relationship between one element or feature and another element or feature shown in the figure, for the sake of clarity. It should be understood that spatially related terms are intended to encompass different orientations of the device in use or operation, in addition to the orientation depicted in the figure. For example, if the device in the figure is upside down, an element described as “below” or “below” another element or feature becomes “above” that other element or feature. Therefore, the exemplary term “under” may encompass both up and down orientations. The device may be oriented differently (rotated 90 degrees or in other orientations), and spatially related descriptors used herein will be interpreted accordingly. Similarly, terms such as “upwardly,” “downwardly,” “vertical,” and “horizontal” are used herein for illustrative purposes only, unless otherwise specified.

[0056] The terms “First” and “Second” may be used herein to describe various features / elements (including steps), but unless the context indicates otherwise, these features / elements should not be limited by these terms. These terms may be used to distinguish one feature / element from another. Thus, without departing from the teachings of this disclosure, the first feature / element discussed below may be called the second feature / element, and similarly, the second feature / element discussed below may be called the first feature / element.

[0057] Throughout this specification and the following claims, unless otherwise specified in the context, the word “comprise,” and variations such as “comprises” and “comprising,” mean that various components may be used together in methods and articles (e.g., compositions and apparatus including devices and methods). For example, the term “comprise” shall be understood to imply the inclusion of any stated element or step and not the exclusion of any other element or step.

[0058] Where used herein and in the claims, including in the examples, all numbers may be read as if they begin with the words “about” or “approximately,” even if the term is not explicitly indicated. The expressions “about” or “approximately” may be used when describing a degree and / or location to indicate that the value and / or location described is within a reasonable predictable range of the value and / or location. For example, a number may have a value of + / -0.1% of the stated value (or range of value), + / -1% of the stated value (or range of value), + / -2% of the stated value (or range of value), + / -5% of the stated value (or range of value), + / -10% of the stated value (or range of value), and so on. Also, any number given herein should be understood to include approximately that value unless the context indicates otherwise. For example, if the value “10” is disclosed, “about 10” is also disclosed. Any numerical range described herein is intended to include all subranges contained therein. Furthermore, as those skilled in the art will understand appropriately, when a value is disclosed as "less than or equal to," it is also understood that "greater than or equal to" and the possible range between values ​​are disclosed. For example, if a value "X" is disclosed, "less than or equal to X" and "greater than or equal to X" (for example, X is a number) are also disclosed. Also, throughout this patent application, the data is provided in many different forms, and it is understood that this data represents a range of endpoints and start points, and any combination of data points. For example, if a particular data point "10" and a particular data point "15" are disclosed, it is understood that greater than, greater than, less than, less than, and equal to 10 and 15 are disclosed along with the range between 10 and 15. It is also understood that each unit between two particular units is also disclosed. For example, if 10 and 15 are disclosed, 11, 12, 13, and 14 are also disclosed.

[0059] While various exemplary embodiments have been described above, many of these embodiments may be modified without departing from the scope of the disclosure as described in the claims. For example, the order in which various described method steps are performed may often be changed in alternative embodiments, and in other alternative embodiments, one or more method steps may be skipped together. Optional features of various device and system embodiments may be included in some embodiments and not in others. Therefore, the foregoing descriptions are provided primarily for illustrative purposes and should not be construed as limiting the scope of the disclosure as described in the claims.

[0060] The examples and figures included herein illustrate, not limiting, specific embodiments in which the subject matter may be implemented. As stated above, other embodiments may be used and derived therefrom, resulting in structural and logical substitutions and modifications without departing from the scope of this disclosure. Such embodiments of the subject matter of the present invention may be referred to individually or collectively by the term “invention” simply for convenience and without the intention of voluntarily limiting the scope of this application to any single invention or concept of invention, if two or more are actually disclosed. Thus, while specific embodiments have been illustrated and described herein, any configuration calculated to achieve the same objective may be used instead of the specific embodiments shown. This disclosure is intended to cover any and all adapted or variant forms of various embodiments. Combinations of the above embodiments, and other embodiments not specifically described herein, will be obvious to those skilled in the art in considering the above description.

Claims

1. The steps include using a machine learning model (200) to analyze an image of a sample in a multiwell plate (MWP) inserted into an analyzer (1000) and determining whether the plate insert matches the MWP, The steps include: in response to the machine learning model (200) determining a mismatch between the plate insert and the MWP, the analyzer generates an alert and notifies the user of the mismatch; In response to the machine learning model (200) determining that the plate insert matches the MWP, the machine learning model is used to analyze an image of the sample to determine whether the MWP is sealed with foil. The steps include: generating the alert in the analyzer in response to the machine learning model (200) determining that the MWP is not sealed with foil, and enabling the analyzer to perform analysis of the sample itself in response to the machine learning model (200) determining that the MWP is sealed with foil; Methods that include...

2. The method according to claim 1, wherein the machine learning model (200) is trained using a combination of real images and augmented images.

3. The method according to claim 1, wherein the machine learning model (200) is trained using a training dataset containing images of MWPs with at least two different numbers of wells.

4. The method according to claim 3, wherein the image of an MWP having at least two different numbers of wells includes an image with foil on the MWP and an image without foil.

5. The method according to claim 1, wherein the machine learning model (200) is trained using a training dataset that includes images of both correct and incorrect combinations of MWP and foil.

6. The method according to claim 1, wherein the machine learning model (200) is trained using a training dataset that includes images of MWPs with multiple filling volumes, multiple dyes, and multiple foil types.

7. The method according to claim 1, wherein the machine learning model (200) is trained using a training dataset containing images of MWP with multiple different filling patterns.

8. The method according to claim 1, wherein the machine learning model (200) is trained using a training dataset that includes images of MWP with user errors.

9. The method according to claim 1, wherein the step of analyzing the sample image using the machine learning model (200) to determine whether there is the mismatch between the plate insert and the MWP includes calculating a confidence score, and when the confidence score falls below a threshold, the alert is generated.

10. The method according to claim 1, wherein the step of analyzing an image of the sample using the machine learning model (200) to determine whether the MWP is sealed with foil includes calculating a confidence score, and when the confidence score falls below a threshold, the alert is generated.

11. The method according to claim 1, further comprising the step of receiving a command from a user to continue the operation of the analyzer when any of the alerts are generated.

12. The method according to claim 1, wherein the machine learning model comprises a plurality of machine learning submodels (210, 220), the first machine learning submodel (210) of the plurality of machine learning submodels being trained to determine whether the plate insert matches the MWP, and the second machine learning submodel (220) of the plurality of machine learning submodels being trained to determine whether the sample is sealed with the foil.

13. The method according to claim 12, wherein the first machine learning submodel (210) and the second machine learning submodel (220) are trained using different datasets.

14. The method according to claim 2, wherein the augmented image is generated using a plurality of augmentation techniques.

15. Memory (420) and A processor (410) coupled to the memory (420), Using a machine learning model (200), the image of the sample in a multiwell plate (MWP) inserted into the analyzer (1000) is analyzed to determine whether the plate insert matches the MWP. In response to the machine learning model (200) determining a mismatch between the plate insert and the MWP, the analyzer generates an alert and notifies the user of the mismatch. In response to the machine learning model (200) determining that the plate insert matches the MWP, the machine learning model is used to analyze the image of the sample to determine whether the MWP is sealed with foil. In response to the machine learning model (200) determining that the MWP is not sealed with foil, the analyzer generates the alert, and in response to the machine learning model (200) determining that the MWP is sealed with foil, the analyzer is enabled to perform analysis of the sample itself. A processor (410) is configured as follows: A system equipped with these features.

Images