Determining parameter values in a biological analyzer

Abstract

Mean corpuscular parameter values for a sample, such as a mean corpuscular hemoglobin or mean corpuscular volume for the red blood cells in a sample, can be calculated based on individual cell corpuscular parameter values for the red blood cells in the sample. These values may be obtained using a machine learning algorithm. Such a machine learning algorithm may be trained using sets of training images annotated with transducer derived mean corpuscular parameter values, thereby allowing a flow imaging based analyzer to provide parameter values which may otherwise not be available with that type of analysis tool.

Description

DETERMINING PARAMETER VALUES IN A BIOLOGICAL ANALYZER Cross Reference to Related Applications

[0001] This is a PCT International application of, and claims the benefit of priority to, provisional patent application 63 / 542,627 for “Determining Parameter Values in a Biological Analyzer,” filed on October 5, 2023 in the U.S. patent office, the disclosure of each of which is hereby incorporated by reference in its entirety. BACKGROUND

[0002] Blood cell analysis is one of the most commonly performed medical tests for providing an overview of a patient's health status. A blood sample can be drawn from a patient's body and stored in a test tube containing an anticoagulant to prevent clotting. A whole blood sample normally comprises three major classes of blood cells including red blood cells (erythrocytes), white blood cells (leukocytes) and platelets (thrombocytes), and various parameters can be determined for the sample based on information from those cells.

[0003] Some of these parameters are related to corpuscular (red-blood cell) related values, such as mean corpuscular volume and mean corpuscular hemoglobin. Traditionally these values are derived from impedance, fluorescence, or volume / conductivity / scatter (VCS) techniques. These values can be obtained, for instance, by taking an average of corpuscular-related measurements for individual red blood cells from the sample, or can be determined through indirect techniques. These values may then be used, either on their own or in combination with information from other tests, to provide information about the health of a patient from whom the blood sample was obtained, such as whether he or she suffers from anemia.

[0004] Imaging (e.g., slide imaging) can be used to obtain images of cells. However, it can be difficult to obtain these corpuscular related values from an image. For instance, a measurement such as mean corpuscular volume involves a volumetric analysis which can be difficult to determine from a 2-dimensional image. Similarly, it can be difficult to assess hemoglobin content just from an image. - 1 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0005] Accordingly, there is a need for improvements in the art related to blood cell analysis, and in particular for improvements in deriving mean corpuscular parameters using imaging analysis systems. SUMMARY

[0006] Described herein are methods and devices for obtaining sample mean corpuscular parameter values from individual cell corpuscular values, and for training machine learning algorithms to individual cell corpuscular parameter values for images using training data derived from non-imaging techniques (e.g., impedance-derived).

[0007] A first aspect of the disclosed technology refers to a computer-implemented method, comprising: obtaining a plurality of red blood cell images for a sample; determining a plurality of individual cell corpuscular parameter values by using a trained machine learning algorithm, the trained machine learning algorithm being configured to determine, for each red blood cell image from the plurality of red blood cell images, a respective individual cell corpuscular parameter value based on that blood cell image; and determining a sample mean corpuscular parameter value for the sample by applying an aggregation function to the plurality of individual cell corpuscular parameter values. In particular, each image of the plurality of red blood cell images for the sample depicts a respective red blood cell of the sample.

[0008] A biological sample may be a blood sample of a human or animal subject. The biological sample may comprise one or more components that may potentially comprise analytes of interest for performing at least one clinical test. A whole blood sample normally comprises three major classes of blood cells including red blood cells (erythrocytes), white blood cells (leukocytes) and platelets (thrombocytes), and various parameters can be determined for the sample based on information from those cells.

[0009] According to some aspects of the disclosed technology, a mean corpuscular parameter may be a parameter comprising information about the red blood cells. In particular, a mean corpuscular - 2 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryparameter provides information about one or more of a hemoglobin content, a morphology and a size of red blood cells. The mean corpuscular parameter is in particular an index of the red blood cell indices. For example, the mean corpuscular parameter may be one of a Mean Corpuscular Volume (MCV), a Mean Corpuscular Hemoglobin (MCH), a Mean Corpuscular Hemoglobin Concentration (MCHC) and a Red Blood Cell Distribution Width (RDW).

[0010] In particular, an individual cell corpuscular parameter for a red blood cell provides information about one or more of a hemoglobin content, a morphology and a size of that red blood cell. For example, the individual cell corpuscular parameter for a red blood cell may be the volume of that red blood cell, the hemoglobin content in that red blood cell or the hemoglobin concentration in that red blood cell.

[0011] According to some aspects of the disclosed technology, the sample mean corpuscular parameter value may be the value of the mean corpuscular parameter for the sample. In particular, the sample mean corpuscular parameter value provides information about one or more of a hemoglobin content, a morphology and a size of the red blood cells of the sample. The mean corpuscular parameter is in particular an index of the red blood cell indices for the sample. For example, the mean corpuscular parameter may be one of a MCV, a MCH, a MCHC, and an RDW for the sample.

[0012] Exemplarily, the individual cell corpuscular parameter and the mean corpuscular parameter may respectively be the volume and MCV. In particular, the individual cell corpuscular parameter and the mean corpuscular parameter may respectively be the hemoglobin content and the MCH. For example, the individual cell corpuscular parameter and the mean corpuscular parameter may respectively be hemoglobin concentration and the MCHC.

[0013] Exemplarily, obtaining the plurality of red blood cell images may comprise accessing, by the computing device, the plurality of red blood cell images. In the present disclosure, accessing data (e.g., the plurality of images or other data such as one or more mean corpuscular parameter values) may comprise retrieving said data e.g., from the at least one memory of the computing device, from the memory of another computing device, from another remote data storage (a - 3 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarydatabase, a secondary memory, a cloud storage or the like). In particular, accessing the plurality of images my comprise retrieving said images from an image capture device that has captured these images or from another computing device. Accordingly, in some cases, retrieving the plurality of images may comprise downloading the images. Additionally or alternatively, accessing data may comprise receiving said data, e.g., from a user or a computing device different from the computing device accessing the images. The two options are not mutually exclusive. For instance, accessing data may comprise receiving said data, storing the data in the memory of the computer device and retrieving the data by accessing said memory.

[0014] Obtaining the plurality of red blood cell images may comprise instructing, by the computing device, an image capture device to capture the plurality of red blood cell images. Obtaining the plurality of red blood cell images may comprise capturing, by an image capture device the plurality of blood cell images.

[0015] According to aspects of the present disclosure, the trained machine learning algorithm is particularly a machine learning algorithm that is trained so that it is configured to determine, for each red blood cell image from the plurality of red blood cell images, the respective individual cell corpuscular parameter value based on that blood cell image. In particular, the trained machine learning algorithm is configured to determine, for each red blood cell image from the plurality of red blood cell images, the respective individual cell corpuscular parameter by processing respective input data, wherein the input data may comprise, or be based on that blood cell image. For example, the respective input data may be generated by processing the blood cell image via a pre-processing procedure.

[0016] According to aspects of the disclosed technology, determining, for each red blood cell image from the plurality of red blood cell images, the respective individual cell corpuscular parameter value based on that blood cell image may comprise processing that blood cell image by using the pre-processing procedure thereby generating the respective input data and processing said input data by using the trained machine learning algorithm thereby determining the respective individual cell corpuscular parameter value. In particular, for each red blood cell image from - 4 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarythe plurality of red blood cell images, the respective individual cell corpuscular parameter value is the value of the corpuscular parameter for the respective red blood cell depicted in that red blood cell image.

[0017] According to some aspects of the disclosed technology, the aggregation function may be a function that associates one or more aggregated values to a plurality of input values. Exemplarily, the aggregation function computes the one or more aggregated values by using the plurality of input values. In particular, an aggregated value is a value that provides information about the plurality of input values, e.g., the distribution of the input values. For instance, the one or more aggregated value may be one or more of an arithmetic mean, a geometric mean, a count, a median and a standard deviation of the plurality of input values. In particular, applying the aggregation function to the plurality of individual cell corpuscular parameter values comprises computing the sample mean corpuscular parameter value by using the plurality of individual cell corpuscular parameter values as inputs.

[0018] Exemplarily the method according to the first aspect of the disclosed technology may be carried out by a computing device. According to the present disclosure, a computing device may comprise one or more processors. For instance, a processor may be a CPU or a GPU. Additionally, the computing device may comprise a memory. For example, the computing device may be a computer, a laptop, a smartphone, a tablet and the like. According to the present disclosure, a computing device may be a distributed system such as a computer network and / or a cloud computing architecture. In particular, the memory of the computing device comprises computer executable instructions implementing the trained machine learning algorithm.

[0019] For example, the machine learning algorithm is trained on impedance-derived data. Exemplarily, the machine learning algorithm is trained on spectrophotometry-derived data.

[0020] In particular, impedance-derived data comprise data measured by using an impedance-based analyzer or impedance-based analytical techniques. For example, an impedance-based analyzer is a hematology analyzer based on the Coulter principle. In some examples, the - 5 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryimpedance-derived data comprise a training set of annotated red blood cell images. Each image of the training set of annotated red blood cell images depicts a respective red blood cell and e.g., is captured by an image capturing device. Exemplarily, each image of the training set of annotated red blood cell images is annotated with information specifying a respective individual cell corpuscular parameter value for the respective red blood cell depicted in that image, the respective individual cell corpuscular parameter value being measured by using the impedance-based analyzer.

[0021] In particular, spectrophotometry-derived data may comprises data measured by using a spectrophotometry-based analyzer, a biological or hematological analyzer with spectrophotometric analytical module(s), or spectrophotometric analytical techniques. In some examples, the spectrophotometry-derived data comprise a training set of annotated red blood cell images. Each image of the training set of annotated red blood cell images depicts a respective red blood cell and e.g., is captured by an image capturing device. Exemplarily, each image of the training set of annotated red blood cell images is annotated with information specifying a respective individual cell corpuscular parameter value for the respective red blood cell depicted in that image, the respective individual cell corpuscular parameter value being measured by using the spectrophotometry-based analyzer.

[0022] The method according to the first aspect of the present disclosure may further comprise, for each red blood cells set of a plurality of red blood cells sets: obtaining a respective impedance- derived training mean corpuscular parameter value, causing a respective plurality of images of the red blood cells in that red blood cells set to be captured, and annotating each image of the respective plurality of images with the respective impedance-derived training mean corpuscular parameter value; and wherein the method further comprises: training the machine learning algorithm using the annotated images of red blood cells from the plurality of training sets of red blood cells.

[0023] The method according to the first aspect of the present disclosure may further comprise, for each red blood cells set of a plurality of red blood cells sets: obtaining a respective - 6 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryspectrophotometry-derived training mean corpuscular parameter value, causing a respective plurality of images of the red blood cells in that red blood cells set to be captured, and annotating each image of the respective plurality of images with the respective spectrophotometry-derived training mean corpuscular parameter value; and wherein the method further comprises: training the machine learning algorithm using the annotated images of red blood cells from the plurality of training sets of red blood cells.

[0024] In particular, obtaining, for each red blood cells set of a plurality of red blood cells sets, the respective (e.g., impedance-derived or spectrophotometry-derived) training mean corpuscular parameter value may comprise accessing, by the computing device, the respective training mean corpuscular parameter value.

[0025] Obtaining, for each red blood cells set of a plurality of red blood cells sets, the respective impedance-derived training mean corpuscular parameter value may comprise instructing, by the computing device, an impedance-based analyzer to measure said value. Obtaining, for each red blood cells set of a plurality of red blood cells sets, the respective impedance-derived training mean corpuscular parameter value may comprise measuring, by the impedance-based analyzer, said value. In particular, for each red blood cells set of the plurality of red blood cells sets, the respective impedance-derived training mean corpuscular parameter value is the mean corpuscular parameter value for that red blood cells set, as measured by an impedance-based analyzer.

[0026] Obtaining, for each red blood cells set of a plurality of red blood cells sets, the respective spectrophotometry-derived training mean corpuscular parameter value may comprise instructing, by the computing device, a spectrophotometry -based analyzer to measure said value. Obtaining, for each red blood cells set of a plurality of red blood cells sets, the respective spectrophotometry-derived training mean corpuscular parameter value may comprise measuring, by the spectrophotometry-based analyzer, said value. In particular, for each red blood cells set of the plurality of red blood cells sets, the respective spectrophotometry-derived - 7 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarytraining mean corpuscular parameter value is the mean corpuscular parameter value for that red blood cells set, as measured by a spectrophotometry-based analyzer.

[0027] Exemplarily, causing, for each red blood cells set of a plurality of red blood cells sets, the respective plurality of images of the red blood cells in that red blood cells set to be captured, comprises instructing, by the computing device, an image capture device to capture the respective plurality of images. In particular, causing, for each red blood cells set of a plurality of red blood cells sets, the respective plurality of images of the red blood cells in that red blood cells set to be captured, comprises capturing, by an image capture device, the respective plurality of images.

[0028] Exemplarily, according to the present disclosure, annotating each image of the respective plurality of images with the respective (impedance-derived or spectrophotometry-derived) training mean corpuscular parameter value comprises associating to that image the respective training mean corpuscular parameter value, said value being the ground truth associated with that image.

[0029] For instance, for each red blood cells set of the plurality of red blood cells sets, the respective (impedance-derived or spectrophotometry-derived) training mean corpuscular parameter value is a mean corpuscular volume value for that red blood cells set. In particular, according to some aspects of the disclosed technology, the mean corpuscular volume of a set of blood cells is defined as the average volume of the red blood cells in the set of blood cells.

[0030] In particular, for each red blood cells set of the plurality of red blood cells sets, the respective impedance-derived training mean corpuscular parameter value is a mean corpuscular hemoglobin value for that red blood cells set. Exemplarily, for each red blood cells set of the plurality of red blood cells sets, the respective spectrophotometry -derived training mean corpuscular parameter value is a mean corpuscular hemoglobin value for that red blood cells set. In particular, according to some aspects of the present disclosure, the mean corpuscular hemoglobin of a set of blood cells is defined as the average amount of hemoglobin per red blood cell of said set of blood cells. - 8 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0031] Exemplarily, each individual cell corpuscular parameter value from the plurality of cell mean corpuscular parameter values is a mean corpuscular volume value; and the sample mean corpuscular parameter value for the sample is a mean corpuscular volume value for the sample.

[0032] For instance, each individual cell corpuscular parameter value from the plurality of cell mean corpuscular parameter values is a mean corpuscular hemoglobin value; and the sample mean corpuscular parameter value for the sample is a mean corpuscular hemoglobin value for the sample.

[0033] According to some aspects of the present disclosure, the machine learning algorithm may a deep-learning algorithm.

[0034] In particular, according to the present disclosure, the deep learning algorithm may comprise a deep neural network. A deep neural network is in particular an artificial neural network comprising a plurality of hidden layers. For example, a deep neural network may comprise two or more hidden layers, in particular three or more hidden layers, more particularly four or more hidden layers, more particularly five or more hidden layers.

[0035] According to the aspects of the present disclosure, the machine learning algorithm may be a convolutional neural network.

[0036] In particular, according to the present disclosure, a convolutional neural network is an artificial neural network comprising one or more convolutional layers. Typically, a convolutional neural network comprises an input layer, a plurality of hidden layers and an output layer. In particular, the plurality of hidden layers comprises one or more convolutional layers. Typically, the plurality of hidden layers comprises one or more pooling layers, e.g., at least a pooling layer after each convolutional layer. Exemplarily, the plurality of hidden layers comprises one or more fully connected layers that may be arranged immediately before the output layer.

[0037] Exemplarily, obtaining the plurality of red blood cell images may comprise receiving the plurality of red blood cell images. - 9 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0038] According to aspects of the disclosed technology, the aggregation function may comprise a median function of the plurality of individual cell corpuscular parameter values. In some cases, the aggregation function may comprise the median function, i.e., it may associate to the plurality of individual cell corpuscular parameter values the median of that plurality of individual cell corpuscular parameter values.

[0039] In particular, obtaining the plurality of red blood cell images may comprise capturing, by the image capturing device, the plurality of red blood cell images as the blood sample flows through a flow cell.

[0040] A second aspect of the present disclosure refers to a biological analyzer comprising an image capture device and a computing device. The biological analyzer is configured to obtain, by the image capture device, a plurality of red blood cell images for a sample; determine, by the computing device, a plurality of individual cell corpuscular parameter value by using a machine learning algorithm, the trained machine learning algorithm being configured to determine, for each red blood cell image from the plurality of red blood cell images, a respective individual cell corpuscular parameter value based on that blood cell image; and determine a sample mean corpuscular parameter value for the sample by applying an aggregation function to the plurality of individual cell corpuscular parameter values.

[0041] According to the present disclosure, an image capture device may be a camera, e.g. a photo camera, a video camera or the like. For instance, the image capture device is a device configured to capture image, in particular a digital image, more particularly a digital image depicting a red blood cell of a blood sample.

[0042] The biological analyzer may further comprise a flow cell with an imaging region, the image capture device configured to capture images from the imaging region of the flow cell. In particular, the image capture device is configured to capture images depicting red blood cells of a red blood sample as the red blood sample flows through the imaging region of the flow cell. - 10 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0043] Exemplarily, each individual mean corpuscular parameter value from the plurality of mean corpuscular parameter values is a mean corpuscular volume value; and the sample mean corpuscular parameter value for the sample is a mean corpuscular volume value for the sample.

[0044] In particular, each individual mean corpuscular parameter value from the plurality of mean corpuscular parameter values is a mean corpuscular hemoglobin value; and the sample mean corpuscular parameter value for the sample is a mean corpuscular hemoglobin value for the sample.

[0045] A third aspect of the present disclosure refers to method, e.g., a computer-implemented method, of training a machine learning algorithm to determine, for each red blood cell image from a plurality of red blood cell images, a respective individual cell corpuscular parameter value based on that blood cell image, wherein the machine learning algorithm is trained on spectrophotometry-derived data or on impedance-derived data.

[0046] In particular, the third aspect of the disclosed technology refers to method of training a machine learning algorithm, comprising, for each set from a plurality of sets of training images of red blood cells: obtaining, by a computing device, a respective impedance-derived mean corpuscular parameter value; annotating, by the computing device, each training image from that set of training images with the respective impedance-derived mean corpuscular parameter value; and wherein the method further comprises: training, by the computing device, the machine learning algorithm using the annotated training images from the plurality of sets of training images of red blood cells.

[0047] Additionally or alternatively, the third aspect of the disclosed technology refers to method of training a machine learning algorithm, comprising, for each set from a plurality of sets of training images of red blood cells: obtaining, by a computing device, a respective spectrophotometry-derived mean corpuscular parameter value; annotating, by the computing device, each training image from that set of training images with the respective spectrophotometry-derived mean corpuscular parameter value; and wherein the method further - 11 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarycomprises: training, by the computing device, the machine learning algorithm using the annotated training images from the plurality of sets of training images of red blood cells.

[0048] In particular, the machine learning algorithm that is trained so that it is configured to determine, for each red blood cell image from the plurality of red blood cell images, the respective individual cell corpuscular parameter value based on that blood cell image.

[0049] In particular, each image of each set from a plurality of sets of training images of red blood cells depicts a respective red blood cell. Exemplarily, for each set from a plurality of sets of training images of red blood cells, the respective impedance-derived mean corpuscular parameter value is the mean corpuscular parameter value for the set of red blood cells depicted in that set of training images of red blood cells, as measured by an impedance-based analyzer. For example, for each set from a plurality of sets of training images of red blood cells, the respective spectrophotometry-derived mean corpuscular parameter value is the mean corpuscular parameter value for the set of red blood cells depicted in that set of training images of red blood cells, as measured by a spectrophotometry-based analyzer.

[0050] Exemplarily, obtaining, for each set from the plurality of sets of training images of red blood cells, the respective impedance-derived mean corpuscular parameter value may comprise one or more of: accessing, by the computing device, the respective impedance-derived mean corpuscular parameter value; instructing, by the computing device, an impedance-based analyzer to measure said value; measuring by the impedance-based analyzer said value.

[0051] For example, obtaining, for each set from the plurality of sets of training images of red blood cells, the respective spectrophotometry-derived mean corpuscular parameter value may comprise one or more of: accessing, by the computing device, the respective spectrophotometry-derived mean corpuscular parameter value; instructing, by the computing device, a spectrophotometry-based analyzer to measure said value; measuring by the spectrophotometry-based analyzer said value. - 12 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0052] Exemplarily, for each set of the plurality of sets of training images, the respective (impedance- derived or spectrophotometry-derived) mean corpuscular parameter value is a mean corpuscular volume value.

[0053] In particular, for each of the plurality of sets of training images, the (impedance-derived or spectrophotometry-derived) mean corpuscular parameter for that set of training images is a mean corpuscular hemoglobin value.

[0054] For instance, the method according to the third aspect of the present disclosure may further comprise obtaining the plurality of sets of training images, wherein, for each set of the plurality of sets of training images, each training image in that set depicts a respective red blood cell from a respective sample of a plurality of samples, the respective sample corresponding to that set of the plurality of sets of training images. The method according to the third aspect of the present disclosure may comprise receiving the plurality of samples.

[0055] For example, obtaining the plurality of sets of training images comprises, for each sample from the plurality of samples, capturing, e.g., by an image capture device, the respective set of training images corresponding to that sample. In particular, obtaining the plurality of sets of training images may comprise, for each sample from the plurality of samples, instructing, by the computing device, an image capture device to capture the respective set of training images corresponding to that sample.

[0056] In particular, for each set of the plurality of sets of training images, obtaining the respective impedance-derived mean corpuscular parameter value comprises: obtaining a respective plurality of impedance-derived parameter values for the red blood cells from the respective sample corresponding to that set of training images; and calculating a mean of the respective plurality of impedance-derived parameter values.

[0057] In particular, obtaining, for each set of the plurality of sets of training images, the respective impedance-derived mean corpuscular parameter value may be based on, for each training image from that set of training images, obtaining parameter values for the individual red blood - 13 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarycells from the sample corresponding to that set of training images e.g. by using an electrode assembly comprising a first electrode and a second electrode.

[0058] For instance, for each set of the plurality of sets of training images, obtaining the respective spectrophotometry-derived mean corpuscular parameter value comprises: obtaining a respective plurality of spectrophotometry-derived parameter values for the red blood cells from the respective sample corresponding to that set of training images; and calculating a mean of the respective plurality of spectrophotometry-derived parameter values.

[0059] In particular, obtaining, for each set of the plurality of sets of training images, the respective spectrophotometry-derived mean corpuscular parameter value may be based on, for each training image from that set of training images, obtaining parameter values for the individual red blood cells from the sample corresponding to that set of training images e.g., by using an spectrophotometry-based analyzer.

[0060] Exemplarily, calculating, for each set of the plurality of sets of training images, the mean of the respective plurality of (impedance-derived or spectrophotometry-derived) parameter values may be carried out by applying an aggregation function to the respective plurality of parameter values. In particular, the aggregation function may comprise the median function.

[0061] Exemplarily, for each set of the plurality of sets of training images, each impedance-derived parameter value of the respective plurality of impedance-derived parameter values is the corpuscular parameter value of a respective red blood cell for the respective sample corresponding, said corpuscular value being measured by using an impedance-based analyzer.

[0062] Exemplarily, for each set of the plurality of sets of training images, each spectrophotometry- derived parameter value of the respective plurality of spectrophotometry-derived parameter values is the corpuscular parameter value of a respective red blood cell for the respective sample corresponding, said corpuscular value being measured by using a spectrophotometry- derived analyzer. - 14 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0063] Exemplarily, for each of the plurality of sets of training images, each of the plurality of impedance-derived parameter values is a corpuscular volume value. For instance, for each of the plurality of sets of training images, each of the plurality of spectrophotometry-derived parameter values is a corpuscular hemoglobin value.

[0064] In particular, for each set of the plurality of sets of training images, obtaining the respective plurality of (impedance-derived or spectrophotometry-derived) parameter values comprises obtaining, for each training image from that set of training images, a respective volume value for the respective red blood cell depicted in that training image.

[0065] For example, for each of the plurality of sets of training images, each of the plurality of impedance-derived parameter values is a corpuscular volume value. In particular, for each of the plurality of sets of training images, each of the plurality of spectrophotometry-derived parameter values is a corpuscular hemoglobin value.

[0066] In particular, for each set of the plurality of sets of training images, obtaining the respective plurality of (impedance-derived or spectrophotometry-derived) parameter values comprises obtaining, for each training image from that set of training images, a respective hemoglobin value for the respective red blood cell depicted in that training image.

[0067] Exemplarily, for each set of the plurality of sets of training images, obtaining the respective plurality of impedance-derived parameter values is carried out by using at least an impedance- based analyzer and / or at least an electrode assembly comprising a first electrode and a second electrode. In particular, for each set of the plurality of sets of training images, obtaining the respective plurality of impedance-derived parameter values may comprise measuring, by using at least an impedance-based analyzer and / or least an electrode assembly comprising a first electrode and a second electrode, the respective plurality of impedance-derived parameter values.

[0068] For example, for each set of the plurality of sets of training images, obtaining the respective plurality of spectrophotometry-derived parameter values is carried out by using at least a - 15 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryspectrophotometry-based analyzer. In particular, for each set of the plurality of sets of training images, obtaining the respective plurality of impedance-derived parameter values may comprise measuring, by using at least a spectrophotometry-based analyzer, the respective plurality of spectrophotometry-derived parameter values.

[0069] A fourth aspect of the present disclosure refers to a computing device comprising a processor configured to perform one or more of the method according the first aspect of the present disclosure and the method according the third aspect of the present disclosure.

[0070] A fifth aspect of the present disclosure refers to a biological analyzer comprising an image capture device, a flow cell and a computing device, wherein the biological analyzer is configured to perform the method according the first aspect of the present disclosure.

[0071] A sixth aspect of the present disclosure refers to an assembly comprising an image capture device, a computing device and an electrode assembly comprising a first electrode and a second electrode, wherein the assembly is configured the method according the third aspect of the present disclosure.

[0072] A seventh aspect of the present disclosure refers to a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of one or more of the method according the first aspect of the present disclosure and the method according the third aspect of the present disclosure.

[0073] An eighth aspect of the present disclosure refers to a computer program product comprising instructions to cause the biological analyzer according to the fifth aspect of the present disclosure to carry out the method according to the first aspect of the present disclosure.

[0074] An ninth aspect of the present disclosure refers to a computer program product comprising instructions to cause the assembly according to the sixth aspect of the present disclosure to carry out the method according to the third aspect of the present disclosure. - 16 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0075] A tenth aspect of the present disclosure refers to computer-readable storage medium having stored thereon one or more of the computer program according to the seventh aspect of the present disclosure, the computer program according to the eighth aspect of the present disclosure and the computer program according to the ninth aspect of the present disclosure.

[0076] It should be noted that any of the various features of the aspects disclosed herein can be included or combined in each of those aspects.

[0077] While multiple examples are described herein, still other examples of the described subject matter will become apparent to those skilled in the art from the following detailed description and drawings, which show and describe illustrative examples of disclosed subject matter. As will be realized, the disclosed subject matter is capable of modifications in various aspects, all without departing from the spirit and scope of the described subject matter. Accordingly, the drawings and detailed description are to be regarded as illustrative in nature and not restrictive. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] While the specification concludes with claims which particularly point out and distinctly claim the invention, it is believed the present invention will be better understood from the following description of certain examples taken in conjunction with the accompanying drawings, in which like reference numerals identify the same elements and in which:

[0079] FIG.1 is a schematic illustration, partly in section and not to scale, showing operational aspects of an exemplary flowcell and high optical resolution imaging device for sample image analysis using digital image processing.

[0080] FIG. 2 depicts a cellular analysis system.

[0081] FIG. 3 depicts a transducer module and associated components.

[0082] FIG. 4 depicts an exemplary training method.

[0083] FIG. 5 depicts an exemplary machine learning algorithm architecture. - 17 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0084] FIG. 6 depicts an illustrative layer such as could be included in the architecture of FIG. 5.

[0085] FIG. 7 shows a method which may be used for obtaining mean corpuscular parameter values for patient samples.

[0086] The drawings are not intended to be limiting in any way, and it is contemplated that various embodiments of the invention may be carried out in a variety of other ways, including those not necessarily depicted in the drawings. The accompanying drawings incorporated in and forming a part of the specification illustrate several aspects of the present invention, and together with the description serve to explain the principles of the invention; it being understood, however, that this invention is not limited to the precise arrangements shown. DETAILED DESCRIPTION

[0087] The present disclosure relates to apparatus, systems, compositions, and methods for analyzing a sample containing particles. In one embodiment, the disclosure relates to an automated particle imaging system which comprises an analyzer which may be, for example, a visual analyzer. In some embodiments, the visual analyzer may further comprise a processor to facilitate automated analysis of the images.

[0088] According to some aspects of this disclosure, a system comprising a visual analyzer may be provided for obtaining images of a sample comprising particles suspended in a liquid. Such a system may be useful, for example, in characterizing particles in biological fluids, such as detecting and quantifying erythrocytes, reticulocytes, nucleated red blood cells, platelets, and white blood cells, including white blood cell differential counting, categorization and subcategorization and analysis. Other similar uses such as characterizing blood cells from other fluids are also contemplated.

[0089] The analysis of blood cells in a blood sample is an exemplary application for which the subject matter is particularly well suited, though other types of body fluid samples may be used. For example, aspects of the disclosed technology may be used in analysis of a non-blood body fluid sample comprising blood cells (e.g., white blood cells and / or red blood cells), such as - 18 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryserum, bone marrow, lavage fluid, effusions, exudates, cerebrospinal fluid, pleural fluid, peritoneal fluid, and amniotic fluid. It is also possible that the sample can be a solid tissue sample, e.g., a biopsy sample that has been treated to produce a cell suspension. The sample may also be a suspension obtained from treating a fecal sample. A sample may also be a laboratory or production line sample comprising particles, such as a cell culture sample. The term sample may be used to refer to a sample obtained from a patient or laboratory or any fraction, portion or aliquot thereof. The sample can be diluted, divided into portions, or stained in some processes.

[0090] In some aspects, samples are presented, imaged and analyzed in an automated manner. In the case of blood samples, the sample may be substantially diluted with a suitable diluent or saline solution, which reduces the extent to which the view of some cells might be hidden by other cells in an undiluted or less-diluted sample. The cells can be treated with agents that enhance the contrast of some cell aspects, for example using permeabilizing agents to render cell membranes permeable, and histological stains to adhere in and to reveal features, such as granules and the nucleus. In some cases, it may be desirable to stain an aliquot of the sample for counting and characterizing particles which include reticulocytes, nucleated red blood cells, and platelets, and for white blood cell differential, characterization and analysis. In other cases, samples containing red blood cells may be diluted before introduction to the flow cell and / or imaging in the flow cell or otherwise.

[0091] The particulars of sample preparation apparatus and methods for sample dilution, permeabilizing and histological staining, generally may be accomplished using precision pumps and valves operated by one or more programmable controllers. Examples can be found in patents such as U.S. Pat. No. 7,319,907. Likewise, techniques for distinguishing among certain cell categories and / or subcategories by their attributes such as relative size and color can be found in U.S. Pat. No. 5,436,978 in connection with white blood cells. The disclosures of these patents are hereby incorporated by reference in their entirety. - 19 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0092] Turning now to the drawings, FIG. 1 schematically shows an exemplary flowcell 22 for conveying a sample fluid through a viewing zone 23 of a high optical resolution imaging device 24 in a configuration for imaging microscopic particles in a sample flow stream 32 using digital image processing. Flowcell 22 is coupled to a source 25 of sample fluid which may have been subjected to processing, such as contact with a particle contrast agent, composition and heating. Flowcell 22 is also coupled to one or more sources 27 of a particle and / or intracellular organelle alignment liquid (PIOAL), such as a clear glycerol solution having a viscosity that is greater than the viscosity of the sample fluid.

[0093] The sample fluid is injected through a flattened opening at a distal end 28 of a sample feed tube 29, and into the interior of the flowcell 22 at a point where the PIOAL flow has been substantially established resulting in a stable and symmetric laminar flow of the PIOAL above and below (or on opposing sides of) the ribbon-shaped sample stream. The sample and PIOAL streams may be supplied by precision metering pumps that move the PIOAL with the injected sample fluid along a flowpath that narrows substantially. The PIOAL envelopes and compresses the sample fluid in the zone 21 where the flowpath narrows. Hence, the decrease in flowpath thickness at zone 21 can contribute to a geometric focusing of the sample stream 32. The sample fluid ribbon 32 is enveloped and carried along with the PIOAL downstream of the narrowing zone 21, passing in front of, or otherwise through the viewing zone 23 of, the high optical resolution imaging device 24 where images are collected, for example, using a Charge-Coupled Device (CCD) 48 through a viewing port 57 while the sample is illuminated by a light source 42. Processor 18 can receive, as input, pixel data from CCD 48. The sample fluid ribbon flows together with the PIOAL to a discharge 33.

[0094] As shown here, the narrowing zone 21 can have a proximal flowpath portion 21a having a proximal thickness PT and a distal flowpath portion 21b having a distal thickness DT, such that distal thickness DT is less than proximal thickness PT. The sample fluid can therefore be injected through the distal end 28 of sample tube 29 at a location that is distal to the proximal portion 21a and proximal to the distal portion 21b. Hence, the sample fluid can enter the PIOAL envelope as the PIOAL stream is compressed by the zone 21. wherein the sample fluid - 20 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryinjection tube has a distal exit port through which sample fluid is injected into flowing sheath fluid, the distal exit port bounded by the decrease in flowpath size of the flowcell.

[0095] The digital high optical resolution imaging device 24 with objective lens 46 is directed along an optical axis that intersects the ribbon-shaped sample stream 32. The relative distance between the objective 46 and the flowcell 33 is variable by operation of a motor drive 54, for resolving and collecting a focused digitized image on a photosensor array. Additional information regarding the construction and operation of an exemplary flowcell such as shown in FIG. 1 is provided in U.S. Patent 9,322,752, entitled “Flowcell Systems and Methods for Particle Analysis in Blood Samples,” filed on March 17, 2014, the disclosure of which is hereby incorporated by reference in its entirety.

[0096] As noted previously, it may be difficult or not possible for imaging analyzers to determine certain parameter values such as mean corpuscular volume may not be capable of determining some corpuscular parameter values, such as mean corpuscular volume and / or mean corpuscular hemoglobin due to the general use of two-dimension images. Accordingly, in some cases a different type of analyzer, such as an impedance-based analyzer, volume-conductivity- scatter based analyzer, or a fluorescence-based analyzer, may be used for this purpose. An example of an analyzer which may be used to obtain impedance-derived values is illustrated in FIG. 2. FIG. 2 schematically depicts a cellular analysis system 200. As shown here, system 200 includes a preparation system 210, a transducer module 220, and an analysis system 230. While system 200 is herein described at a very high level, with reference to the three core system blocks (210, 220, and 230), one of skill in the art would readily understand that system 200 includes many other system components such as central control processor(s), display system(s), fluidic system(s), temperature control system(s), user-safety control system(s), and the like. In operation, a whole blood sample (WBS) 240 can be presented to the system 200 for analysis. In some instances, WBS 240 is aspirated into system 200. Exemplary aspiration techniques are known to the skilled artisan. After aspiration, WBS 240 can be delivered to a preparation system 210. Preparation system 210 receives WBS 240 and can perform operations involved with preparing WBS 240 for further measurement and analysis. - 21 - 0133788.0794525 4862-2756-9898v7 Confidential - Company ProprietaryFor example, preparation system 210 may separate WBS 240 into predefined aliquots for presentation to transducer module 220. Preparation system 210 may also include mixing chambers so that appropriate reagents may be added to the aliquots. Preparation system 210 may also include temperature control components to control the temperature of the reagents and / or mixing chambers. Appropriate temperature controls can improve the consistency of the operations of preparation system 210.

[0097] In some instances, predefined aliquots can be transferred from preparation system 210 to transducer module 220. As described in further detail below, transducer module 220 can perform direct current (DC) impedance, radiofrequency (RF) conductivity, light transmission, and / or light scatter measurements of cells from the WBS passing individually therethrough. Measured DC impedance, RF conductivity, and light propagation (e.g., light transmission, light scatter) parameters can be provided or transmitted to analysis system 230 for data processing. In some instances, analysis system 230 may include computer processing features and / or one or more modules or components such as those described herein with reference to the system depicted in FIG. 6 and described further below, which can evaluate the measured parameters, identify and enumerate the WBS constituents, and correlate a subset of data characterizing elements of the WBS with an infection status. As shown here, cellular analysis system 200 may generate or output a report 250 containing information such as mean corpuscular volume, mean corpuscular hemoglobin, and / or other information for the same. In some instances, excess biological sample from transducer module 220 can be directed to an external (or alternatively internal) waste system 260.

[0098] FIG.3 illustrates in more detail a transducer module and associated components in more detail. As shown here, system 300 includes a transducer module 310 having a light or irradiation source such as a laser 312 emitting a beam 314. The laser 312 can be, for example, a 635 nm, 5 mW, solid-state laser. In some instances, system 300 may include a focus-alignment system 320 that adjusts beam 314 such that a resulting beam 322 is focused and positioned at a cell interrogation zone 332 of a flow cell 330. In some instances, flow cell 330 receives a sample aliquot from a preparation system 302. Various fluidic mechanisms and techniques can - 22 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarybe employed for hydrodynamic focusing of the sample aliquot within flow cell 330. In some instances, the aliquot generally flows through the cell interrogation zone 332 such that its constituents pass through the cell interrogation zone 332 one at a time. In some cases, a system 300 may include a cell interrogation zone or other feature of a transducer module or blood analysis instrument such as those described in U.S. Pat. Nos. 5,125,737; 6,228,652; 7,390,662; 8,094,299; and 8,189,187, the contents of which are incorporated herein by reference. For example, a cell interrogation zone 332 may be defined by a square transverse cross-section measuring approximately 50×50 microns, and having a length (measured in the direction of flow) of approximately 65 microns.

[0099] Flow cell 330 may include an electrode assembly having first and second electrodes 334, 336 for performing DC impedance and RF conductivity measurements of the cells passing through cell interrogation zone 332. Signals from electrodes 334, 336 can be transmitted to analysis system 304. The electrode assembly can analyze volume and conductivity characteristics of the cells using low-frequency current and high-frequency current, respectively. For example, low-frequency DC impedance measurements can be used to analyze the volume of each individual cell passing through the cell interrogation zone. Relatedly, high-frequency RF current measurements can be used to determine the conductivity of cells passing through the cell interrogation zone. Because cell walls act as conductors to high frequency current, the high frequency current can be used to detect differences in the insulating properties of the cell components, as the current passes through the cell walls and through each cell interior. High frequency current can be used to characterize nuclear and granular constituents and the chemical composition of the cell interior. Additional information on methods and systems which may be used to derive values from analysis of a sample are provided in U.S. Pat. No. 8,192,995 and 11,114,205, the contents of which are hereby incorporated by reference.

[0100] While, in some cases, parameter values may be obtained using a combination of systems (e.g., an imaging system such as shown in FIG. 1 with a laser or impedance-based system such as shown in FIGS. 2 and 3), it is also possible that the capabilities of one system - 23 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarymay be used to enhance another without requiring both systems to be available. For example, a system such as shown in FIGS 2 and / or 3 may be used to obtain impedance-based parameter values which could then be used to train a machine learning algorithm to derive those same parameter values from images of the type that would be captured by a system such as shown in FIG. 1. Once such a trained algorithm was available, it could be stored in a memory of an analyzer such as shown in FIG. 1 (or linked to the analyzer in a cloud or distributed environment), and used to derive parameter values (e.g., MCV and MCH) using images captured by the analyzer of FIG.1 even in cases were analyzers such as shown in FIGS.2 and 3 are not available. An illustrative example of this type of training method is provided in FIG. 4, discussed below.

[0101] The transducer module involves a conversion of energy from one form to another, and can apply to a range of analytical techniques. The term “transducer” and “transducer module” are inclusive of various analytical techniques, modules, systems, devices, and methods that involve conversion of one energy form to another for the purposes of biological analysis, for instance obtaining information on volume-related and hemoglobin-related parameters for red blood cells.

[0102] A transducer module is used in an impedance-based system to derive cell volume whereby cells are passed individually through an aperture or opening, and the correlating resistance or impedance associated with each cell translate to a volume of the cell (e.g., the Coulter technique). The mean corpuscular volume can then be derived in various ways, for instance by determining each corpuscular volume from each red blood cell and then obtaining an average for the corpuscular volume for the red blood cells in a blood sample to obtain a mean corpuscular volume, or by obtaining a total corpuscular volume for all the red blood cells in a blood sample and dividing this over the number of red blood cells in a blood sample to obtain a mean corpuscular volume. Alternatively, the mean corpuscular volume can be indirectly determined through other measurements related to red blood cell analysis. - 24 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0103] A transducer module can also be used in a spectrophotometry application whereby red blood cells are lysed with a lysing agent to expose their hemoglobin content. Light of a particular wavelength from a light source is directed towards the hemoglobin content, and a particular portion of this light will be at least partially absorbed by the hemoglobin, and a certain amount of light will be passed through and received by a sensor as a certain voltage. This received voltage is then corresponded to a particular hemoglobin content value to obtain a hemoglobin content for a blood sample (e.g., a corpuscular hemoglobin amount). The hemoglobin amount can further be divided over the cells in the blood sample to obtain a mean corpuscular hemoglobin, representing the average hemoglobin per red blood cell. Alternatively, the mean corpuscular volume can be indirectly determined through other measurements related to red blood cell analysis.

[0104] In some examples, a transducer module can utilize a fluorescence excitation analytical technique, where blood cells are fluorescently excited via the use of fluorescent compositions and laser excitation in order to establish cell data (e.g., cell sizes / volumes) that are used to establish particular cell information (e.g., corpuscular volumes, other related cell sizing data, or other cell data). As explained later and herein, this transducer-based data (e.g., obtained through impedance, spectrophotometry, or fluorescence) can be used as ground truth data or data sourcing as part of an imaging training workflow to correlate transducer-based data to imaging-based data.

[0105] In a method such as shown in FIG. 4, training a machine learning algorithm may begin with receiving 401 a plurality of samples, and obtaining 402 a plurality of sets of training images by, for each of those samples, capturing images of red blood cells comprised by that sample. Additionally, the method may also comprise receiving 403 a transducer derived mean corpuscular parameter value (e.g., a parameter value determined using impedance, spectrophotometry, or other technology such as may be included in a transducer module such as discussed above) for each of the sets of training images. This may be done, for example, by using an electrode assembly to obtain 404 impedance derived parameter values(e.g., volume values) and / or using a spectrophotometer to obtain 404 spectrophotometry derived parameter - 25 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryvalues (e.g., hemoglobin values) for the red blood cells from the samples, and then calculating 405 means of those parameter values to obtain a mean corpuscular parameter value for each of the sets of training images. The training images may then be annotated 406 by, for each set of training images, annotating each training image in that set with the mean corpuscular parameter value for the sample corresponding to that set of training images. The annotated training images may then be used to train 407 a machine learning algorithm to provide individual cell corpuscular parameter values (i.e., values predicting the mean corpuscular parameter value for a sample which comprises the individual cell) based on images such as those from the sets of training images. In other words, a blood sample is first run through a transducer module to assess a sample-level corpuscular parameter value associated with the sample (e.g., mean corpuscular volume or mean corpuscular hemoglobin). A different portion of the same blood sample (e.g., a different portion taken from the same patient sample, a different portion of the same aliquot taken from the same patient sample, or a different aliquot taken from the same sample) is then run through an imaging module to pull individual images of blood cells. Alternatively, the same exact blood sample (e.g., once prepared and run through the transducer module, the same sample portion is then used in the imager) is then run through an imaging module to pull individual images of blood cells. The blood cell images are then labelled with the same sample-level corpuscular parameter value (e.g., where the transducer run sample has a sample level MCV of X, then each image labelled or annotated with an MCV value of X). This is done for additional blood samples where each set of images is then labelled / annotated / associated with the sample-level MCV value in order to correlate images with a sample-level parameter value.

[0106] In one example, the sample-level parameter value is derived by getting a sample level value for a measurement from a transducer module (e.g., a sample level corpuscular volume, sample level corpuscular hemoglobin, or sample level platelet volume representing respectively the total corpuscular volume, corpuscular hemoglobin, or platelet volume associated with the sample) and then averaging this over the number of cells represented in the sample (e.g., dividing over the number of cells represented in the sample) to obtain an average measurement associated with each cell (e.g., a mean corpuscular volume, a mean corpuscular - 26 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryhemoglobin, or mean platelet volume). In one example, the sample-level corpuscular parameter value is derived by getting a cell-by-cell value from a transducer module (e.g., a cell-by-cell corpuscular volume, corpuscular hemoglobin, platelet volume) and then taking a function of all these values (e.g., a mean or median) to calculate a sample level parameter value (e.g., e.g., a mean corpuscular volume, a mean corpuscular hemoglobin, or mean platelet volume). For a red-cell distribution width, the range in cell volumes can be used to establish a statistical variance or statistical range associated with the cell type. However the sample- level parameter value is derived, once it is derived from the transducer module, the sample- level parameter value is then used to annotate or label each image of an imaging module (e.g., so that each image of a training sample receives the same sample-level parameter value). This is then used as part of a training process for training an algorithm to label cell images with algorithm-derived parameter values.

[0107] For example, the machine learning algorithm may comprise a Neural Network (NN), e.g., Convolutional Neural Network (CNN). Exemplarily, the NN may be trained by using about 20,000 training images annotated as described above. For instance, the loss function used for the training may be mean squared error, the learning rate may be set to 10^^, the number of epochs may be equal to 200 and the batch size equal of 512.

[0108] To provide a concrete example of how a process such as described in the context of FIG. 4 could be applied, consider FIG. 5, which illustrates an architecture which could be used for a machine learning algorithm such as could be trained to provide individual cell corpuscular parameter values. In the architecture of FIG. 5, an input image 501 would be analyzed in a series of stages 502a-502n, each of which may comprise one or more layers and which is illustrated in more detail in FIG. 6. As shown in FIG. 6, an input 601 (which, in the initial stage 602a of FIG. 5 would be the input image 501, and otherwise would be the output of the preceding stage) is provided to a stage 602 where it would be processed by a convolutional layer of the stage 602 to generate one or more transformed images 603a-603m. This processing may include convolving the input 601 with a set of filters 604a-604m, each of which would identify a type of feature from the underlying image that would then be captured - 27 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryin that filter’s corresponding transformed image. For instance, as a simple example, convolving an image with the filter shown in table 1 could generate a transformed image capturing the edges from the input 601. −1 −1 −1^−1 8 −1^−1 −1 −1Table 1

[0109] As shown in FIG. 6, in addition to generating transformed images 603a-603m a stage may also comprises a pooling layer that generates a pooled image 605a-605m for each of the transformed images 603a-603m. This may be done, for example, by organizing the appropriate transformed image into a set of regions, and then replacing the values in that region with a single value, such as the maximum value for the region or the average of the values for the region. The result would be a pooled image whose resolution would be reduced relative to its corresponding transformed image based on the size of the regions it was split into (e.g., if the transformed image 603a-603m had NxN dimensions, and it was split into 2x2 regions, then the pooled image 605a-605m would have size (N / 2)x(N / 2)). These pooled images 605a-605m could then be combined into a single output image 606, in which each of the pooled images 605a-605m is treated as a separate channel in the output image 606. This output image 606 can then be provided as input to the next stage 502a-502m as shown in FIG. 5.

[0110] Returning to the discussion of FIG. 5, after a final output image 503 has been created through the various stages 502a-502n of processing, the final output image 503 could be provided as input to a neural network 504. This may be done, for example, by providing the value of each channel of each pixel in the output image 503 to an input node of a densely connected single layer network. The output of the neural network 504 could then be treated as indicating the value of a parameter derived from the input image 501, such as an individual cell corpuscular volume or an individual cell corpuscular hemoglobin.

[0111] Exemplarily, the trained CNN may comprise the following layers: - 28 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarya) An input layer that receives an 128x128x3 RGB image depicting a red blood cell image, immediately followed by b) A convolutional layer with 64 5x5 filters and the ReLU activation function, immediately followed by c) A 2x2 max pooling that generates a 64x64x64 output, immediately followed by d) A convolutional layer with 128 5x5 filters and the ReLU activation function, immediately followed by e) A 2x2 max pooling that generates a 32x32x128 output, immediately followed by f) A convolutional layer with 256 5x5 filters and the ReLU activation function, immediately followed by g) A 2x2 max pooling that generates a 16x16x256 output, immediately followed by h) A convolutional layer with 512 5x5 filters and the ReLU activation function, immediately followed by i) A 2x2 max pooling that generates a 8x8x512 output, immediately followed by j) A convolutional layer with 512 5x5 filters and the ReLU activation function, immediately followed by k) A 2x2 max pooling that generates a 4x4x512 output, immediately followed by l) A fully connected layer that generates a single scalar value, the single scalar being the individual cell corpuscular value for the red blood cell obtained by using the CNN.

[0112] After a machine learning algorithm (e.g., a deep learning algorithm with a plurality of convolution layers as described in the context of FIGS. 5 and 6) had been trained using a method such as shown in FIG. 4, it may be used for obtaining mean corpuscular parameter values for patient samples using a method such as shown in FIG. 7. In that method, after a plurality of red blood cell images are received 701 (e.g., through capturing images of red blood cells as they pass through a flow cell in a system such as shown in FIG. 1), a plurality of individual cell corpuscular parameter values could be obtained 702 by providing those images to a machine learning algorithm trained to determine individual cell corpuscular parameter values for red blood cell images (e.g., an algorithm having the structure shown in FIGS. 5 and 6 which had been trained using a method such as shown in FIG. 4). These individual cell - 29 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarycorpuscular parameter values could then be used to obtain 703 a sample mean corpuscular parameter value for the sample by applying an aggregation function to the plurality of individual cell corpuscular parameter values (e.g., by averaging the various individual cell corpuscular parameter values – or in other words a mean value).

[0113] Additional examples can include utilizing a different statistical function to report the mean corpuscular parameter value – such as utilizing the median of the aggregation function and reporting this value as the mean corpuscular volume or mean corpuscular hemoglobin.

[0114] Additional examples can include reporting different blood-cell related or corpuscular-related parameter values – such as mean platelet volume (MPV) or red-cell distribution width (RDW). With MPV, a training approach outlined generally herein (getting transducer-derived sample level data for mean platelet volume and labelling individual cell images with the same mean platelet volume) is done for platelet volume. A neural network such as the one described herein is then trained on the image data. Similarly, this approach can be used for RDW which assesses the distribution associated with cell sizes or volumes. A training approach outlined generally herein (getting transducer-derived sample level data for RDW and labelling individual cell images with the RDW value) is done for red-cell distribution width.

[0115] While the discussion above and herein discusses training approaches, algorithms, devices, systems, and methods that leverage applying sample level data to individual cell images and exemplary concepts have included the use of corpuscular values (e.g., mean corpuscular volume, mean corpuscular hemoglobin, red-cell distribution width) or platelet values (e.g., mean platelet volume), these are provided by way of example and these techniques can be used for other measurements where it can be desirable to correlate sample level data with individual images. In other words, the inventive concepts apply to such training approaches, algorithms, devices, systems, and methods for correlating sample data and images. - 30 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0116] While the above discussion provided examples of machine learning algorithms which may be used to determine mean corpuscular parameter values, as well as methods and systems which may be used in training and applying such algorithms in practice, the examples from the above discussion should be understood as being illustrative only, and the disclosed technology could be applied in other manners by those of ordinary skill in the art without undue experimentation. As an example, consider an aggregation function which could be used to obtain 703 a mean corpuscular parameter value in a method such as shown in FIG. 7. While such an aggregation function could be calculating an average of individual cell corpuscular parameter values, other types of aggregation functions, such as identifying a median of individual mean corpuscular parameter values and treating that median value as the mean corpuscular parameter values for the same, are also possible. Similarly, while FIGS. 5-6 illustrated a machine learning algorithm architecture that could be used when implementing the disclosed technology, other architectures, such as architectures where multiple neural network layers are used to transform the outputs of convolution layers into a final output, or architectures which do not use convolution layers at all (e.g., decision trees support vector machines, etc.) may also be used in implementations of the disclosed technology. Other variations are also possible and will be immediately apparent to those of ordinary skill in the art in light of this disclosure. Accordingly, the particular examples and illustrations provided herein should not be treated as implying limitations on the protection provided by this document or any related document.

[0117] As a further illustration of potential implementations and applications of the disclosed technology, the following examples are provided of non-exhaustive ways in which the teachings herein may be combined or applied. It should be understood that the following examples are not intended to restrict the coverage of any claims that may be presented at any time in this application or in subsequent filings of this application. No disclaimer is intended. The following examples are being provided for nothing more than merely illustrative purposes. It is contemplated that the various teachings herein may be arranged and applied in numerous other ways. It is also contemplated that some variations may omit certain features referred to in the below examples. Therefore, none of the aspects or features referred to below should be - 31 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarydeemed critical unless otherwise explicitly indicated as such at a later date by the inventors or by a successor in interest to the inventors. If any claims are presented in this application or in subsequent filings related to this application that include additional features beyond those referred to below, those additional features shall not be presumed to have been added for any reason relating to patentability.

[0118] Example 1

[0119] A computer-implemented method, comprising: obtaining a plurality of blood cell images for a sample; determining a plurality of individual cell parameter values by using a trained machine learning algorithm, the trained machine learning algorithm being configured to determine, for each blood cell image from the plurality of blood cell images, a respective individual cell parameter value based on that blood cell image; and determining a sample parameter value for the sample by applying an aggregation function to the plurality of individual cell parameter values.

[0120] Example 2

[0121] The method of example 1, wherein the plurality of blood cell images are red blood cell images, the plurality of individual cell parameter values are corpuscular parameter values, and the sample parameter value is a sample mean parameter value.

[0122] Example 3

[0123] The method of example 2, wherein the corpuscular parameter values are mean corpuscular volumes, and the sample mean parameter value is a sample mean corpuscular volume.

[0124] Example 4 - 32 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0125] The method of example 2, wherein the corpuscular parameter values are mean corpuscular hemoglobin values, and the sample mean parameter value is a sample mean corpuscular hemoglobin.

[0126] Example 5

[0127] The method of any one of the preceding examples, wherein the machine learning algorithm is a deep-learning algorithm.

[0128] Example 6

[0129] The method of any one of the preceding examples, wherein the machine learning algorithm comprises a convolutional neural network.

[0130] Example 7

[0131] The method of any one of the preceding examples, wherein obtaining the plurality of blood cell images comprises receiving the plurality of blood cell images.

[0132] Example 8

[0133] The method of any one of the preceding examples, wherein the aggregation function comprises a mean function or a median function of the plurality of individual cell corpuscular values.

[0134] Example 9

[0135] The method of any of examples 1-7, wherein the aggregation function comprises a mean function of the plurality of individual cell parameter values.

[0136] Example 10

[0137] The method of any of examples 1-7, wherein the aggregation function comprises a median function of the plurality of individual cell parameter values. - 33 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0138] Example 11

[0139] The method of any one of the preceding examples, wherein obtaining the plurality of blood cell images comprises capturing the plurality of blood cell images as the blood sample flows through a flow cell.

[0140] Example 12

[0141] The method of any of examples 1-11, wherein the machine learning algorithm is trained on transducer-derived data, and the transducer includes an impedance module or a spectrophotometric module.

[0142] Example 13

[0143] The method of example 12, wherein the method further comprises: for each blood cells set of a plurality of blood cells sets: obtaining a sample-level transducer-derived training parameter value, causing a plurality of images of the blood cells in that blood cells set to be captured, and annotating each image of the plurality of images with the sample-level transducer-derived training parameter value; and training the machine learning algorithm using the annotated images of blood cells from the plurality of blood cells sets wherein, for each annotated image from the plurality of blood cells sets, the sample-level transducer-derived training parameter value with which that image is annotated is treated as a ground truth value for that image.

[0144] Example 14

[0145] The method of example 13, wherein, for each blood cells set of the plurality of blood cells sets, each image of the plurality of images is annotated with the same sample-level transducer-derived training parameter value.

[0146] Example 15 - 34 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0147] The method of any of examples 13-14, wherein, for each blood cells set from the plurality of blood cells sets, the blood cells from that set are red blood cells.

[0148] Example 16

[0149] The method of any of examples 13-14, wherein, for each blood cells set of the plurality of blood cells sets, the respective transducer-derived value is a mean corpuscular value for that blood cells set.

[0150] Example 17

[0151] The method of example 16, wherein the mean corpuscular value is a mean corpuscular volume or a mean corpuscular hemoglobin.

[0152] Example 18

[0153] An analyzer configured to perform the method of any of examples 1-17.

[0154] Example 19

[0155] A method of training a machine learning algorithm, comprising: for each set from a plurality of sets of training images of blood cells: obtaining, by a computing device, a sample- level transducer-derived parameter value; annotating, by the computing device, each training image from that set of training images with the sample-level transducer-derived parameter value; and training, by the computing device, the machine learning algorithm using the annotated training images from the plurality of sets of training images of blood cells wherein, for each annotated image from the plurality of sets, the sample-level transducer-derived parameter value with which that image is annotated is treated as a ground truth value for that image.

[0156] Example 20 - 35 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0157] The method of example 19, wherein for each set of the plurality of sets of training images, the sample-level transducer-derived parameter value is a mean corpuscular volume value, a mean corpuscular hemoglobin volume value, a mean platelet volume value, or a red- blood cell distribution width value.

[0158] Example 21

[0159] The method of example 19 wherein the blood cells are red blood cells.

[0160] Example 22

[0161] The method of any one of examples 19 to 21, wherein: the method further comprises obtaining the plurality of sets of training images wherein, for each set of the plurality of sets of training images, each training image in that set depicts a respective blood cell from a sample corresponding to that set of training images; and for each set of the plurality of sets of training images, obtaining the sample-level transducer-derived parameter value comprises at least one of: obtaining a plurality of transducer-derived parameter values for blood cells from the sample corresponding to that set of training images and calculating a mean of the respective plurality of transducer-derived parameter values; or obtaining a single transducer- derived parameter value for blood cells from the sample corresponding to that set of training images and averaging over the number of the blood cells from the samples to calculate a mean of the transducer-derived parameter values.

[0162] Example 23

[0163] The method of example 22, wherein, for each of the plurality of sets of training images, each of the plurality of transducer-derived parameter values is a corpuscular volume value derived from an impedance module.

[0164] Example 24 - 36 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0165] The method of example 23 wherein, for each set of the plurality of sets of training images, obtaining the plurality of transducer-derived parameter values for blood cells from the sample corresponding to that set of training images is carried out by using at least an electrode assembly comprising a first electrode and a second electrode.

[0166] Example 25

[0167] The method of example 22, wherein, for each of the plurality of sets of training images, each of the plurality of transducer derived parameter values is a corpuscular hemoglobin value derived from a spectrophotometric module.

[0168] Example 26

[0169] The method of any of examples 22-25, wherein a subject blood cell depicted in a training image from a set of training images is also included in the blood cells for which the plurality of transducer-derived parameter values is obtained.

[0170] Example 27

[0171] The method of example 26, wherein: for each set of the plurality of sets of training images, obtaining the plurality of transducer-derived parameter values for blood cells from the sample corresponding to that set of training images comprises, for each blood cell from the sample corresponding to that set of training images, obtaining a transducer-derived parameter value for that blood cell; and the ground truth value for the image depicting the subject blood cell is different from transducer-derived parameter value obtained for the subject blood cell.

[0172] Example 28

[0173] The method of any of examples 22-25, wherein, for at least one set of training images: obtaining that set of training images comprises capturing images of blood cells from a first portion of the sample corresponding to that set of training images; and obtaining the plurality of transducer-derived parameter values for blood cells from the sample corresponding - 37 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryto that set of training images comprises obtaining transducer-derived parameter values for blood cells from a second portion of the sample corresponding to that set of training images, wherein the first portion and the second portions are different portions.

[0174] Example 29

[0175] The method of any of examples 22-28, wherein, for each set of the plurality of set of training images, obtaining that set of training images comprises capturing that plurality of training images as the sample corresponding to that set of training images flows through a flow cell.

[0176] Example 30

[0177] The method of any of examples 19-29, wherein the machine learning algorithm is a deep learning algorithm.

[0178] Example 31

[0179] The method of any of examples 19-30, wherein the machine learning algorithm comprises a convolutional neural network.

[0180] Example 32

[0181] The method of any of examples 19-31, wherein, for each set from the plurality of sets of training images, the sample-level transducer-derived parameter value is a mean corpuscular value.

[0182] Example 33

[0183] A non-transitory computer readable medium having stored therein instructions for performing the method of any of examples 19-32.

[0184] Example 34 - 38 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0185] A biological analyzer, comprising: an image capture device; a computing device, wherein the biological analyzer is configured to: obtain, by the image capture device, a plurality of blood cell images for a sample; determine, by the computing device, a plurality of individual cell parameter values by using a machine learning algorithm, the machine learning algorithm being configured to determine, for each blood cell image from the plurality of blood cell images, a respective individual cell parameter value based on that blood cell image; and determine a sample parameter value for the sample by applying an aggregation function to the plurality of individual cell parameter values.

[0186] Example 35

[0187] The analyzer of example 34, wherein the plurality of blood cell images are red blood cell images, the plurality of individual cell parameter values are corpuscular parameter values, and the sample parameter value is a sample mean parameter value.

[0188] Example 36

[0189] The analyzer of example 35, wherein the corpuscular parameter values are corpuscular volumes, and the sample mean parameter value is a sample mean corpuscular volume.

[0190] Example 37

[0191] The analyzer of example 35, wherein the corpuscular parameter values are mean corpuscular hemoglobin values, and the sample mean parameter value is a sample mean corpuscular hemoglobin.

[0192] Example 38

[0193] The analyzer of any of examples 34-37, wherein the machine learning algorithm is a deep-learning algorithm.

[0194] Example 39 - 39 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0195] The analyzer of any of examples 34-38, wherein the machine learning algorithm comprises a convolutional neural network.

[0196] Example 40

[0197] The analyzer of any of examples 34-49, wherein obtaining the plurality of blood cell images comprises receiving the plurality of blood cell images.

[0198] Example 41

[0199] The analyzer of any of examples 34-40, wherein the aggregation function comprises a mean function or a median function of the plurality of individual cell corpuscular parameter values.

[0200] Example 42

[0201] The analyzer of any of examples 34-41, wherein obtaining the plurality of red blood cell images comprises capturing the plurality of red blood cell images as the blood sample flows through a flow cell.

[0202] Example 43

[0203] The analyzer of any of examples 34-42, wherein the machine learning algorithm is trained on transducer-derived data from one of: an impedance module; and a spectrophotometric module.

[0204] Example 44

[0205] The analyzer of example 43, wherein the machine learning algorithm is trained using a training method which comprises: for each blood cells set of a plurality of blood cells sets: obtaining a sample-level transducer-derived training parameter value, causing a plurality of images of the blood cells in that blood cells set to be captured, and annotating each image of the plurality of images with the sample-level transducer-derived training parameter value; - 40 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryand training the machine learning algorithm using the annotated images of blood cells from the plurality of blood cells sets wherein, for each annotated image from the plurality of blood cells sets, the sample-level transducer-derived training parameter value with which that image is annotated is treated as a ground truth value for that image.

[0206] Example 45

[0207] The analyzer of example 44, wherein, or each blood cells set, the blood cells from that set are red blood cells.

[0208] Example 46

[0209] The analyzer of example 44, wherein, for each blood cells set of the plurality of blood cells sets, the respective transducer-derived value is a mean corpuscular value for that blood cells set.

[0210] Example 47

[0211] The analyzer of example 46, wherein the mean corpuscular value is a mean corpuscular volume or a mean corpuscular hemoglobin.

[0212] Example 48

[0213] A method of determining a sample parameter value using the analyzer of any of examples 34-47.

[0214] Example 49

[0215] A biological analysis system comprising: a biological analyzer comprising an image capture device configured to capture a plurality of blood cell images for a sample; and a computing device configured to: determine a plurality of individual cell parameter values by using a machine learning algorithm, the machine learning algorithm being configured to determine, for each blood cell image from the plurality of blood cell images, a respective - 41 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryindividual cell parameter value based on that blood cell image; and determine a sample parameter value for the sample by applying an aggregation function to the plurality of individual cell parameter values.

[0216] Example 50

[0217] The system of example 49, wherein the plurality of blood cell images are red blood cell images, the plurality of individual cell parameter values are corpuscular parameter values, and the sample parameter value is a sample mean parameter value.

[0218] Example 51

[0219] The system of example 50, wherein the corpuscular parameter values are corpuscular volumes, and the sample mean parameter value is a sample mean corpuscular volume.

[0220] Example 52

[0221] The system of example 50, wherein the corpuscular parameter values are mean corpuscular hemoglobin values, and the sample mean parameter value is a sample mean corpuscular hemoglobin.

[0222] Example 53

[0223] The system of any of examples 49-52, wherein the machine learning algorithm is a deep-learning algorithm.

[0224] Example 54

[0225] The system of any of examples 49-53, wherein the machine learning algorithm comprises a convolutional neural network.

[0226] Example 55 - 42 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0227] The system of any of examples 49-54, wherein obtaining the plurality of blood cell images comprises receiving the plurality of blood cell images.

[0228] Example 56

[0229] The system of any of examples 49-55, wherein the aggregation function comprises a mean function or a median function of the plurality of individual cell corpuscular parameter values.

[0230] Example 57

[0231] The system of any of examples 49-56, wherein obtaining the plurality of red blood cell images comprises capturing the plurality of red blood cell images as the blood sample flows through a flow cell.

[0232] Example 58

[0233] The system of any of examples 46-54, wherein the machine learning algorithm is trained on transducer-derived data from one of: an impedance module; and a spectrophotometric module.

[0234] Example 59

[0235] The system of example 58, wherein the machine learning algorithm is trained using a training method which comprises: for each blood cells set of a plurality of blood cells sets: obtaining a sample-level transducer-derived training parameter value, causing a plurality of images of the blood cells in that blood cells set to be captured, and annotating each image of the plurality of images with the sample-level transducer-derived training parameter value; and training the machine learning algorithm using the annotated images of blood cells from the plurality of blood cells sets wherein, for each annotated image from the plurality of blood cells sets, the sample-level transducer-derived training parameter value with which that image is annotated is treated as a ground truth value for that image. - 43 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

[0236] Example 60

[0237] The system of example 59, wherein, or each blood cells set, the blood cells from that set are red blood cells.

[0238] Example 61

[0239] The system of example 59, wherein, for each blood cells set of the plurality of blood cells sets, the respective transducer-derived value is a mean corpuscular value for that blood cells set.

[0240] Example 62

[0241] The system of example 61, wherein the mean corpuscular value is a mean corpuscular volume or a mean corpuscular hemoglobin.

[0242] Example 63

[0243] The system of any of examples 49-62, wherein the computing device is located remotely from the biological analyzer.

[0244] Example 64

[0245] The system of any of examples 49-62, wherein the biological analyzer is configured to provide the plurality of blood cell images to the computing device, wherein providing the plurality of blood cell images to the computing device comprises transmitting the plurality of blood cell images over a wide area network from a location of the biological analyzer to a location of the computing device.

[0246] A method of determining a sample parameter value using the system of any of the examples 49-64.

[0247] Each of the calculations or operations described herein may be performed using a computer or other processor having hardware, software, and / or firmware. The various method - 44 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarysteps may be performed by modules, and the modules may comprise any of a wide variety of digital and / or analog data processing hardware and / or software arranged to perform the method steps described herein. The modules optionally comprising data processing hardware adapted to perform one or more of these steps by having appropriate machine programming code associated therewith, the modules for two or more steps (or portions of two or more steps) being integrated into a single processor board or separated into different processor boards in any of a wide variety of integrated and / or distributed processing architectures. These methods and systems will often employ a tangible media embodying machine-readable code with instructions for performing the method steps described above. Suitable tangible media may comprise a memory (including a volatile memory and / or a non-volatile memory), a storage media (such as a magnetic recording on a floppy disk, a hard disk, a tape, or the like; on an optical memory such as a CD, a CD-R / W, a CD-ROM, a DVD, or the like; or any other digital or analog storage media), or the like.

[0248] All patents, patent publications, patent applications, journal articles, books, technical references, and the like discussed in the instant disclosure are incorporated herein by reference in their entirety for all purposes.

[0249] Different arrangements of the components depicted in the drawings or described above, as well as components and steps not shown or described are possible. For example, in some cases, aspects of processing described herein (e.g., training of machine learning algorithms, application of machine learning algorithms to images taken of patient samples) may be performed in various configurations – for instance, using a processor which his comprised by (or local to) an analyzer, a parallel-processing arrangement, or processing being performed remotely from the analyzer which captures images (such as using a cloud based platform, or using a remotely linked computer or system to process the analyzer data). Similarly, some features and sub-combinations are useful and may be employed without reference to other features and sub-combinations. Embodiments of the invention have been described for illustrative and not restrictive purposes, and alternative embodiments will become apparent to readers of this patent. In certain cases, method steps or operations may be - 45 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryperformed or executed in differing order, or operations may be added, deleted or modified. It can be appreciated that, in certain aspects of the invention, a single component may be replaced by multiple components, and multiple components may be replaced by a single component, to provide an element or structure or to perform a given function or functions. Except where such substitution would not be operative to practice certain embodiments of the invention, such substitution is considered within the scope of the invention. Accordingly, the claims should not be treated as limited to the examples, drawings, embodiments and illustrations provided above, but instead should be understood as having the scope provided when their terms are given their broadest reasonable interpretation as provided by a general purpose dictionary, except that when a term or phrase is indicated as having a particular meaning under the heading Explicit Definitions, it should be understood as having that meaning when used in the claims.

[0250] Explicit Definitions

[0251] It should be understood that, in the above examples and the claims, a statement that something is “based on” something else should be understood to mean that it is determined at least in part by the thing that it is indicated as being based on. To indicate that something must be completely determined based on something else, it is described as being “based EXCLUSIVELY on” whatever it must be completely determined by.

[0252] It should be understood that, in the above examples and claims, the term “set” should be understood as one or more things which are grouped together. - 46 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary

Claims

What is claimed is:

1. A computer-implemented method, comprising: obtaining a plurality of blood cell images for a sample; determining a plurality of individual cell parameter values by using a trained machine learning algorithm, the trained machine learning algorithm being configured to determine, for each blood cell image from the plurality of blood cell images, a respective individual cell parameter value based on that blood cell image; and determining a sample parameter value for the sample by applying an aggregation function to the plurality of individual cell parameter values.

2. The method of claim 1, wherein the plurality of blood cell images are red blood cell images, the plurality of individual cell parameter values are corpuscular parameter values, and the sample parameter value is a sample mean parameter value.

3. The method of claim 2, wherein the corpuscular parameter values are mean corpuscular volumes, and the sample mean parameter value is a sample mean corpuscular volume.

4. The method of claim 2, wherein the corpuscular parameter values are mean corpuscular hemoglobin values, and the sample mean parameter value is a sample mean corpuscular hemoglobin.

5. The method of any one of the preceding claims, wherein the machine learning algorithm is a deep-learning algorithm.

6. The method of any one of the preceding claims, wherein the machine learning algorithm comprises a convolutional neural network.

7. The method of any one of the preceding claims, wherein obtaining the plurality of blood cell images comprises receiving the plurality of blood cell images. - 47 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary8. The method of any one of the preceding claims, wherein the aggregation function comprises a mean function or a median function of the plurality of individual cell parameter values.

9. The method of any one of the preceding claims, wherein obtaining the plurality of blood cell images comprises capturing the plurality of blood cell images as the blood sample flows through a flow cell.

10. The method of any of claims 1-9, wherein the machine learning algorithm is trained on transducer-derived data, and the transducer includes an impedance module or a spectrophotometric module.

11. The method of claim 10, wherein the method further comprises: for each blood cells set of a plurality of blood cells sets: obtaining a sample-level transducer-derived training parameter value, causing a plurality of images of the blood cells in that blood cells set to be captured, and annotating each image of the plurality of images with the sample-level transducer- derived training parameter value; and training the machine learning algorithm using the annotated images of blood cells from the plurality of blood cells sets wherein, for each annotated image from the plurality of blood cells sets, the sample-level transducer-derived training parameter value with which that image is annotated is treated as a ground truth value for that image.

12. The method of claim 11, wherein, for each blood cells set from the plurality of blood cells sets, the blood cells from that set are red blood cells. - 48 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary13. The method of claim 11, wherein, for each blood cells set of the plurality of blood cells sets, the respective transducer-derived value is a mean corpuscular value for that blood cells set.

14. The method of claim 13, wherein the mean corpuscular value is a mean corpuscular volume or a mean corpuscular hemoglobin.

15. An analyzer configured to perform the method of any of claims 1-14.

16. A method of training a machine learning algorithm, comprising: for each set from a plurality of sets of training images of blood cells: obtaining, by a computing device, a sample-level transducer-derived parameter value; and annotating, by the computing device, each training image from that set of training images with the sample-level transducer-derived parameter value; and training, by the computing device, the machine learning algorithm using the annotated training images from the plurality of sets of training images of blood cells wherein, for each annotated image from the plurality of sets, the sample-level transducer-derived parameter value with which that image is annotated is treated as a ground truth value for that image.

17. The method of claim 16, wherein for each set of the plurality of sets of training images, the sample-level transducer-derived parameter value is a mean corpuscular volume value, a mean corpuscular hemoglobin volume value, a mean platelet volume value, or a red-blood cell distribution width value.

18. The method of claim 16, wherein the blood cells are red blood cells.

19. The method of any one of claims 16 to 18, wherein: - 49 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarythe method further comprises obtaining the plurality of sets of training images wherein, for each set of the plurality of sets of training images, each training image in that set depicts a respective blood cell from a sample corresponding to that set of training images; and for each set of the plurality of sets of training images, obtaining the sample-level transducer-derived parameter value comprises at least one of: obtaining a plurality of transducer-derived parameter values for blood cells from the sample corresponding to that set of training images and calculating a mean of the respective plurality of transducer-derived parameter values; or obtaining a single transducer-derived parameter value for blood cells from the sample corresponding to that set of training images and averaging over the number of the blood cells from the samples to calculate a mean of the transducer-derived parameter values.

20. The method of claim 19, wherein, for each of the plurality of sets of training images, each of the plurality of transducer-derived parameter values is a corpuscular volume value derived from an impedance module.

21. The method of claim 20 wherein, for each set of the plurality of sets of training images, obtaining the plurality of transducer-derived parameter values for blood cells from the sample corresponding to that set of training images is carried out by using at least an electrode assembly comprising a first electrode and a second electrode.

22. The method of claim 19, wherein, for each of the plurality of sets of training images, each of the plurality of transducer derived parameter values is a corpuscular hemoglobin value derived from a spectrophotometric module.

23. The method of any of claims 19-22, wherein a subject blood cell depicted in a training image from a set of training images is also included in the blood cells for which the plurality of transducer-derived parameter values is obtained. - 50 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary24. The method of claim 23, wherein: for each set of the plurality of sets of training images, obtaining the plurality of transducer- derived parameter values for blood cells from the sample corresponding to that set of training images comprises, for each blood cell from the sample corresponding to that set of training images, obtaining a transducer-derived parameter value for that blood cell; and the ground truth value for the image depicting the subject blood cell is different from transducer-derived parameter value obtained for the subject blood cell.

25. The method of any of claims 19-22, wherein, for at least one set of training images: obtaining that set of training images comprises capturing images of blood cells from a first portion of the sample corresponding to that set of training images; and obtaining the plurality of transducer-derived parameter values for blood cells from the sample corresponding to that set of training images comprises obtaining transducer-derived parameter values for blood cells from a second portion of the sample corresponding to that set of training images, wherein the first portion and the second portions are different portions.

26. The method of any of claims 19-25, wherein, for each set of the plurality of set of training images, obtaining that set of training images comprises capturing that plurality of training images as the sample corresponding to that set of training images flows through a flow cell.

27. The method of any of claims 16-26, wherein the machine learning algorithm is a deep learning algorithm.

28. The method of any of claims 16-27, wherein the machine learning algorithm comprises a convolutional neural network.

29. The method of any of claims 16-28, wherein, for each set from the plurality of sets of training images, the sample-level transducer-derived parameter value is a mean corpuscular value. - 51 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary30. A non-transitory computer readable medium having stored therein instructions for performing the method of any of claims 16-29.

31. A biological analyzer, comprising: an image capture device; a computing device, wherein the biological analyzer is configured to: obtain, by the image capture device, a plurality of blood cell images for a sample; determine, by the computing device, a plurality of individual cell parameter values by using a machine learning algorithm, the machine learning algorithm being configured to determine, for each blood cell image from the plurality of blood cell images, a respective individual cell parameter value based on that blood cell image; and determine a sample parameter value for the sample by applying an aggregation function to the plurality of individual cell parameter values.

32. The analyzer of claim 31, wherein the plurality of blood cell images are red blood cell images, the plurality of individual cell parameter values are corpuscular parameter values, and the sample parameter value is a sample mean parameter value.

33. The analyzer of claim 32, wherein the corpuscular parameter values are corpuscular volumes, and the sample mean parameter value is a sample mean corpuscular volume.

34. The analyzer of claim 32, wherein the corpuscular parameter values are mean corpuscular hemoglobin values, and the sample mean parameter value is a sample mean corpuscular hemoglobin.

35. The analyzer of any of claims 31-34, wherein the machine learning algorithm is a deep- learning algorithm. - 52 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary36. The analyzer of any of claims 31-35, wherein the machine learning algorithm comprises a convolutional neural network.

37. The analyzer of any of claims 31-36, wherein obtaining the plurality of blood cell images comprises receiving the plurality of blood cell images.

38. The analyzer of any of claims 31-37, wherein the aggregation function comprises a mean function or a median function of the plurality of individual cell corpuscular parameter values.

39. The analyzer of any of claims 31-38, wherein obtaining the plurality of red blood cell images comprises capturing the plurality of red blood cell images as the blood sample flows through a flow cell.

40. The analyzer of any of claims 31-39, wherein the machine learning algorithm is trained on transducer-derived data from one of: an impedance module; and a spectrophotometric module.

41. The analyzer of claim 40, wherein the machine learning algorithm is trained using a training method which comprises: for each blood cells set of a plurality of blood cells sets: obtaining a sample-level transducer-derived training parameter value, causing a plurality of images of the blood cells in that blood cells set to be captured, and annotating each image of the plurality of images with the sample-level transducer- derived training parameter value; and - 53 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietarytraining the machine learning algorithm using the annotated images of blood cells from the plurality of blood cells sets wherein, for each annotated image from the plurality of blood cells sets, the sample-level transducer-derived training parameter value with which that image is annotated is treated as a ground truth value for that image.

42. The analyzer of claim 41, wherein, or each blood cells set, the blood cells from that set are red blood cells.

43. The analyzer of claim 41, wherein, for each blood cells set of the plurality of blood cells sets, the respective transducer-derived value is a mean corpuscular value for that blood cells set.

44. The analyzer of claim 43, wherein the mean corpuscular value is a mean corpuscular volume or a mean corpuscular hemoglobin.

45. A method of determining a sample parameter value using the analyzer of any of claims 31- 44.

46. A biological analysis system comprising: a biological analyzer comprising an image capture device configured to capture a plurality of blood cell images for a sample; and a computing device configured to: determine a plurality of individual cell parameter values by using a machine learning algorithm, the machine learning algorithm being configured to determine, for each blood cell image from the plurality of blood cell images, a respective individual cell parameter value based on that blood cell image; and determine a sample parameter value for the sample by applying an aggregation function to the plurality of individual cell parameter values. - 54 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietary47. The system of claim 46, wherein the plurality of blood cell images are red blood cell images, the plurality of individual cell parameter values are corpuscular parameter values, and the sample parameter value is a sample mean parameter value.

48. The system of claim 47, wherein the corpuscular parameter values are corpuscular volumes, and the sample mean parameter value is a sample mean corpuscular volume.

49. The system of claim 47, wherein the corpuscular parameter values are mean corpuscular hemoglobin values, and the sample mean parameter value is a sample mean corpuscular hemoglobin.

50. The system of any of claims 46-49, wherein the machine learning algorithm is a deep- learning algorithm.

51. The system of any of claims 46-50, wherein the machine learning algorithm comprises a convolutional neural network.

52. The system of any of claims 46-51, wherein obtaining the plurality of blood cell images comprises receiving the plurality of blood cell images.

53. The system of any of claims 46-52, wherein the aggregation function comprises a mean function or a median function of the plurality of individual cell corpuscular parameter values.

54. The system of any of claims 46-53, wherein obtaining the plurality of red blood cell images comprises capturing the plurality of red blood cell images as the blood sample flows through a flow cell.

55. The system of any of claims 46-54, wherein the machine learning algorithm is trained on transducer-derived data from one of: - 55 - 0133788.0794525 4862-2756-9898v7 Confidential - Company Proprietaryan impedance module; and a spectrophotometric module.

56. The system of claim 55, wherein the machine learning algorithm is trained using a training method which comprises: for each blood cells set of a plurality of blood cells sets: obtaining a sample-level transducer-derived training parameter value, causing a plurality of images of the blood cells in that blood cells set to be captured, and annotating each image of the plurality of images with the sample-level transducer- derived training parameter value; and training the machine learning algorithm using the annotated images of blood cells from the plurality of blood cells sets wherein, for each annotated image from the plurality of blood cells sets, the sample-level transducer-derived training parameter value with which that image is annotated is treated as a ground truth value for that image.

57. The system of claim 56, wherein, or each blood cells set, the blood cells from that set are red blood cells.

58. The system of claim 56, wherein, for each blood cells set of the plurality of blood cells sets, the respective transducer-derived value is a mean corpuscular value for that blood cells set.

59. The system of claim 58, wherein the mean corpuscular value is a mean corpuscular volume or a mean corpuscular hemoglobin.

60. A method of determining a sample parameter value using the system of any of claims 46- 59. -Confidential - Company Proprietary

Images