Classification of events in flow cytometry data

Abstract

The present disclosure relates to a method for analyzing particles within a flow cytometry system using a computing device. The method includes: receiving flow cytometry data from a flow cytometry system; assigning a probability value to the at least one event, the probability value indicating a probability that the at least one event is within a defined threshold rule to classify the event as belonging to a cluster; displaying the probability value to the user; allowing the user to adjust the probability threshold; and generating an output. An output is generated by comparing the probability value to a probability threshold, identifying at least one inclusion event when the probability value satisfies the probability threshold, and displaying the at least one inclusion event to a user.

Description

[0001] Cross-references to related applications

[0002] This application was filed on October 25, 2024 as a PCT international application and claims the benefit and priority of U.S. Application No. 63 / 594,165, filed on October 30, 2023, entitled CLASSIFICATION OF EVENTS IN FLOW CYTOMETRY DATA, the disclosure of which is incorporated herein by reference in its entirety. Background Technology

[0003] Flow cytometry is a technique used to detect and analyze the chemical and physical properties of cells or particles in fluid samples. For example, a flow cytometer can be used to evaluate cells from blood, bone marrow, tumors, or other bodily fluids. Typically, the sample is passed through a fluid nozzle that aligns particles within the sheath fluid in a single file line. As the particles pass through in this single file line, a laser beam illuminates them to generate radiation, including forward-scattered light, side-scattered light, and fluorescence. This radiation can then be detected and analyzed to determine one or more properties of the particles. Summary of the Invention

[0004] Generally, this disclosure relates to the analysis of particles using flow cytometry. In one possible configuration, an AI algorithm is used to automatically classify events using one or more automatically adjusted control variables / thresholds. In another possible configuration, one or more control variables / thresholds are manually adjusted to correct the classification of events, and the training dataset is further expanded if necessary. Various aspects are described in this disclosure, including but not limited to the following.

[0005] One aspect relates to a method for analyzing particles within a flow cytometry system using a computing device. The method includes: receiving flow cytometry data from the flow cytometry system, the flow cytometry data including at least one event; assigning probability values ​​to the at least one event, the probability values ​​indicating the probability that the at least one event belongs to a cluster / category of events; displaying the probability values ​​to a user, wherein a probability threshold categorizes the event into groups of interest; allowing the user to adjust the probability threshold; and generating output. The output is generated by comparing the probability values ​​of the at least one event with the probability threshold and assigning classification information to at least one inclusive event, wherein the at least one inclusive event includes probability values ​​exceeding the probability threshold.

[0006] On the other hand, a method for adjusting the classification of results within a flow cytometry system is provided. The method includes: examining flow cytometry data, which includes at least one event, the at least one event including a probability value indicating the probability that the at least one event is associated with a population of interest; allowing adjustment of a probability threshold on the flow cytometry system to filter out one or more of the at least one events when the at least one probability value is greater than a probability threshold input; and receiving an output from the flow cytometry system. The output includes at least one inclusive event exceeding the probability threshold input.

[0007] Another aspect relates to a flow cytometry system for analyzing particles. The flow cytometry system includes: at least one processing device; and a non-transitory computer-readable storage medium storing instructions, which, when executed by the processing device, cause the at least one processing device to: receive flow cytometry data from the flow cytometry system, the flow cytometry data including at least one event; assign probability values ​​to the at least one event, the at least one probability value indicating the probability that the at least one event is within a population of interest; display the at least one probability value to a user; allow adjustment of a probability threshold from the user; and generate output. The output is generated by: comparing the at least one probability value to a probability threshold; identifying at least one inclusive event when the at least one probability value exceeds the probability threshold; and displaying the at least one inclusive event to the user.

[0008] Various additional aspects will be set forth in the following description. These aspects may involve individual features and combinations of features. It should be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only, and do not limit the broad inventive concept on which the embodiments disclosed herein are based. Attached Figure Description

[0009] The following figures illustrate examples of this disclosure and therefore do not limit the scope of this disclosure. Examples of this disclosure will be described below in conjunction with the accompanying figures, in which the same reference numerals indicate the same elements.

[0010] Figure 1 An example of a flow cytometer system is illustrated schematically.

[0011] Figure 2 schematically shown Figure 1 An example of waveform analysis equipment for a flow cytometer system.

[0012] Figure 3 An example of a method for generating output is shown, which is adjusted based on a default probability threshold or a probability threshold received from the user, showing events that may be included in clusters / categories of events.

[0013] Figure 4The manual adjustment method is shown. Figures 2 to 3 An example of a method for autonomously generating flow cytometry data using an artificial intelligence model.

[0014] Figure 5 It shows Figures 2 to 4 An example of an artificial intelligence model, which has at least one input, at least one node in a first hidden layer, at least one node in a second hidden layer, and an output generated by the artificial intelligence model.

[0015] Figure 6 The training method is shown. Figure 2 An example of an artificial intelligence model used to evaluate flow cytometry data.

[0016] Figure 7 It shows the result of Figure 1 The waveform analysis device performs an example of gate adjustment.

[0017] Figure 8 It shows the result of Figures 1 to 2 An example of a waveform analysis device performing probabilistic event classification corresponding to the selection of a 50% probability threshold for the received data.

[0018] Figure 9 It shows the result of Figures 1 to 2 An example of probability event classification performed by a waveform analysis device, corresponding to the selection of a probability threshold of 95%.

[0019] Figure 10 Includes example histograms showing probability distributions of variation between two cell populations, which reduce the impact of... Figures 1 to 2 The confidence rating generated by the waveform analysis equipment.

[0020] Figure 11 Flow cytometry data are shown at three different resolutions.

[0021] Figure 12 An exemplary architecture of a computing device that can be used to implement various aspects of this disclosure is shown.

[0022] In the accompanying drawings, similar parts and / or features may have the same reference numerals. Furthermore, various parts of the same type may be distinguished by adding a dash after the reference numeral and a second reference numeral to differentiate between similar parts. If only the first reference numeral is used in the description, the description applies to any of the similar parts having the same first reference numeral, regardless of the second reference numeral. Detailed Implementation

[0023] Various examples will be described in detail with reference to the accompanying drawings, in which the same reference numerals denote the same parts and components throughout multiple views.

[0024] Figure 1 An example of a flow cytometry system 100 is illustrated schematically. In some cases, the flow cytometry system 100 may include aspects and features described in U.S. Provisional Patent Application No. 63 / 410,984, filed September 28, 2022, which is incorporated herein by reference in its entirety.

[0025] Typically, flow cytometry is a technique used to measure and analyze the properties of particles or cells as they flow in a fluid stream 114. Data from millions of particles or cells can be collected by the flow cytometer system 100 within minutes and displayed in various formats. Illustrative example applications of flow cytometry include: phenotypic analysis to identify and count specific cell types within a population; analysis of DNA or RNA content within cells; determination of the presence of antigens on or within cell surfaces; and assessment of cell health status.

[0026] like Figure 1 As illustrated in the illustrative example, a flow cytometry system 100 typically includes three main component subsystems: a fluid system 110, an optical system 120, and an electronic system 130. The fluid system 110 includes a nozzle 112 that receives a sample containing particles or cells suspended in a fluid. The nozzle 112 generates and ejects a fluid flow 114 of particles or cells arranged in a single chord. Each particle or cell passes through one or more beams generated by a light source 102. The point where a particle or cell intersects with a beam is referred to as an interrogation zone 116. In some examples, the light source 102 includes one or more lasers.

[0027] The optical system 120 includes a light source 102, an optical element 122, and a detector 124. At the interrogation area 116, light from the light source 102 strikes particles or cells in the fluid flow 114 and is scattered. The optical element 122 guides the scattered light toward the detector 124. The detector 124 may include: a forward scattering (FSC) detector for measuring scattering in the path of the light source 102; a side scattering (SSC) detector for measuring scattering at a 90-degree angle relative to the light source 102; and one or more fluorescence detectors (FL1, FL2, FL3, ..., FLn) for measuring the emission fluorescence intensity at different wavelengths of light.

[0028] Typically, the intensity of the fluorescence spectrum (FSC) is proportional to the size or diameter of the particle due to light diffraction around it. Therefore, FSC can be used to differentiate particles by size. On the other hand, the fluorescence spectrum (SSC) is generated by light refracted or reflected from the internal structure of the particle, and thus can provide information about the particle's internal complexity or particle size. By adding fluorescent labels to the sample, different fluorescence signals / channels (e.g., green, yellow, and red) can be analyzed for the functional characteristics of the cells. For example, since T cells have CD3 binding sites, a sample containing T cells can be "stained" with an anti-CD3 antibody conjugated to a fluorescent molecule. When these cells pass through interrogation zone 116, light from the light source excites the fluorescent tag or dye to emit photons at a wavelength detectable by the fluorescence detector. Therefore, detector 124 can simultaneously measure several parameters, enabling particle classification based on the wavelength of the detected light according to particle function.

[0029] Electronic system 130 includes waveform acquisition device 140 and waveform analysis device 150. Waveform acquisition device 140 is communicatively coupled to detector 124 to receive analog waveform data 126 generated by detector 124. Waveform acquisition device 140 includes analog-to-digital converter (ADC) 142 configured to digitize the waveform data.

[0030] Waveform analysis device 150 is configured to receive digital waveform data and display it to a user of flow cytometry system 100. In some embodiments, waveform analysis device 150 includes computing devices such as those communicatively coupled to flow cytometry system 101 via a network. Flow cytometry system 101 may include fluid system 110, optical system 120, and waveform acquisition device 140. In other embodiments, waveform analysis device 150 is integrated with flow cytometry system 101.

[0031] Current flow cytometers use a field-programmable gate array (FPGA) in waveform acquisition device 140 to acquire information about individual particles passing through the beam. Waveform acquisition device 140 uses a single threshold to determine when the detector output begins to convert from analog to digital. Therefore, digitization begins if detector 124 outputs a voltage value exceeding the threshold, or when detector 124 outputs a voltage value exceeding the threshold. As waveform data is digitized, the FPGA calculates the height, width, and area of ​​each pulse. Other waveform-related data (including data not exceeding the voltage threshold) are not captured, stored, or otherwise made available for analysis. If the user wishes to adjust the threshold, the experiment must be rerun with the new threshold, resulting in resource and time costs.

[0032] Furthermore, if a user wishes to gating the results (i.e., selecting a specific population of cells or particles based on their unique characteristics), region-based gating is typically used to define the boundaries around a specific population on a two-dimensional scatter plot (e.g., forward scatter (FSC) versus side scatter (SSC) plot). Region-based gating involves several constraints, where the gating is often subjectively defined by the user based on a visual inspection of the data, and the defined region may include overlapping populations. This makes it challenging to define separate gatings for each population based on scatter plot data, and some subpopulations of interest may be discarded during the gating process when one subpopulation is selected instead of another.

[0033] To solve the above problems, Figure 1 The example shown illustrates an event classifier 154 included as a component of a waveform analysis device 150. Compared to standard region-based gating methods, event classifier 154 allows users to more efficiently and effectively classify events into portions of groups of interest or clusters of events. In some examples, event classifier 154 allows users to classify samples using an artificial intelligence model (e.g., a machine learning algorithm) that uses multi-parameter data to define regions through statistical probabilities rather than one-dimensional and two-dimensional subjective gating, which refers to... Figures 3 to 11 This is described and illustrated in further detail. Furthermore, the event classifier 154 allows users to cluster / classify events by comparing multiple dimensions of the data at once (e.g., forward scattering, impedance, side scattering, fluorescence channels related to the detection of light scattering or emission from the fluorescent material or fluorescent dye, time spent by particles passing through the query point, etc.) rather than analyzing only one or two dimensions of the data at a time. Additionally, during the gating process, the event classifier 154 can cluster populations without discarding specific subgroups.

[0034] In addition, the flow cytometer system 100 includes a graphics processing unit (GPU) 152. Figure 1 In the example shown, GPU 152 is shown as a component included in waveform analysis device 150. GPU 152 processes a continuous digital stream generated by waveform acquisition device 140. The digital stream processed by GPU 152 is continuous because waveform acquisition device 140 collects data related to one or more pulses, which is then processed by artificial intelligence model 202 (such as...). Figure 2 (As shown) This data is used to generate probabilistic clusters / categories of events.

[0035] Waveform acquisition device 140 captures electrical signals (also referred to as “pulses”) generated when particles pass through light source 102 and detected by detector 124. In doing so, waveform acquisition device 140 captures pulse attributes, such as pulse peak value, area, width, and half-height, measured on a two-dimensional histogram. For each detected pulse, these parameters are transmitted to GPU 152.

[0036] Based on the foregoing description, waveform analysis device 150 calculates a digital version of waveform data with added data points, and the waveform data used for experiments is displayed and can be processed by GPU 152 in its entirety. Event classifier 154 allows users to classify one or more events across several dimensions while performing objective analysis and limiting data loss during the event classification process. Because probabilistic data is maintained, users have the ability to dynamically adjust thresholds and update graphs in real time without retraining the artificial intelligence model 202. Further details regarding operation and advantages are discussed below.

[0037] The flow cytometer system 100 includes elements shown and described for discussion purposes, and it should be understood that many variations in components and functions are possible. Optical elements 122 may include a series of filters, dichroic mirrors, and / or beam splitters to select different wavelengths of light and provide those wavelengths to an appropriate detector 124. Detector 124 may include, for example, a photomultiplier tube (PMT), an avalanche photodiode (APD), or a single-photon counting device.

[0038] Figure 2 schematically shown Figure 1 An example of a waveform analysis device 150 in a flow cytometry system 100. The waveform analysis device 150 receives, stores, and displays waveform data. The waveform analysis device 150 includes an interface 210 for receiving digitized raw waveform data 232, a persistent storage device 230 for storing the digitized raw waveform data 232, and may include a graphical user interface (GUI) 220 for displaying the digitized raw waveform data 232. The persistent storage device 230 may also store multiple event probability values ​​that allow for thresholding and real-time updating and display of applied thresholds (as further described below). The persistent storage device 230 may include system memory such as random access memory (RAM) and / or long-term non-volatile memory such as a hard disk drive.

[0039] The waveform analysis device 150 may also include an event classifier 154, which includes a software application or a set of related software applications configured to instruct the GPU 152 to process the digitized raw waveform data 232. The event classifier 154 may execute on one or more processors or in the cloud to provide the functionality described herein in conjunction with the GPU 152, such as receiving user input via a GUI 220. The event classifier 154 may include one or more artificial intelligence models 202 (e.g., machine learning algorithms) to analyze the data received from the waveform acquisition device 140. One or more components of the waveform analysis device 150 may reside in a cloud computing application within a networked distributed system. In this respect, the waveform analysis device 150 may be any of a variety of computing devices, including but not limited to personal computing devices, server computing devices, or distributed computing devices.

[0040] Figure 3 An example of method 300 for generating output is shown, which shows events that may be clustered based on probability thresholds received from the user.

[0041] At step 302, event classifier 154 receives flow cytometry data including at least one event. The flow cytometry data is collected by flow cytometry system 100 (as described above). Figure 1 and Figure 2 (As shown and described). Flow cytometry data includes at least one event. An event is a single measurement or detection of a particle (e.g., cell, microorganism, compensating microsphere, protein, etc.) when it passes through the light source 102 at interrogation zone 116. In some examples, the at least one event measures the characteristics of blood cells (e.g., white blood cells and red blood cells). The event is detected by detector 124 before it is received by waveform acquisition device 140. Waveform acquisition device 140 digitizes the waveform data using ADC 142 and sends the waveform data to waveform analysis device 150. Waveform analysis device 150 receives the waveform data and prepares it for further analysis, which may be performed by components such as event classifier 154.

[0042] At step 304, event classifier 154 assigns a probability value to each event among the events received within the flow cytometry data. The probability value relates to the probability that each event falls within the desired category (i.e., the probability that the particle detected by detector 124 is part of a population of interest (e.g., a specific cell type included within the sample)). In some examples, the probability value is a numerical value ranging from 0 to 1. In some examples, the probability value is a percentage ranging from 0 to 100, or it can be formatted for flow cytometry histogram classification at a 10-bit resolution from 0 to 1023.

[0043] In some examples, the artificial intelligence model 202 is used to analyze flow cytometry data and assign probability values ​​to each event within the flow cytometry data. (See above for reference.) Figure 1 The AI ​​model 202 can assign a probability value to each event by analyzing flow cytometry data and determining the likelihood that each event is included in a desired category (i.e., a cluster / category of events with one or more characteristics associated with the group of interest). In some examples, the AI ​​model 202 analyzes whether each event is included in the desired category by analyzing flow cytometry data that includes multiple dimensions (e.g., data related to forward scattering, side scattering, fluorescence channels related to the detection of light emitted from a fluorophore or fluorescent dye, or the time taken for a particle to pass through an interrogation point). Analyzing data with multiple dimensions allows the AI ​​model 202 to assign a more accurate probability value to each event compared to the process of analyzing flow cytometry data using only one or two dimensions. Furthermore, the AI ​​model 202 can objectively assign probability values ​​to one or more events within the flow cytometry data without relying on human intervention, which would require plotting the area around certain events within a two-dimensional histogram (e.g., a side scattering and forward scattering plot).

[0044] Flow cytometry data can be used to train an artificial intelligence model 202 (e.g., a machine learning algorithm) to assign accurate and precise probability values ​​to each event. In some examples, a supervised learning process is used to train the AI ​​model 202. The supervised learning process involves teaching the AI ​​model 202 how to correctly identify desired event clusters within the input flow cytometry data. In some examples, this includes a process where the user manually identifies the desired event clusters and the AI ​​model 202 develops a method for mapping the input data to the user-identified event clusters. Once the AI ​​model 202 is properly trained (i.e., generating confidence ratings based on the correlation between the AI ​​model 202's output and the user-identified desired clusters, and these confidence ratings are above a threshold), the AI ​​model 202 generates probabilistic event classifications. In some examples, the probabilistic event classifications can be manually adjusted by the user. Reference Figure 6 The process of training the artificial intelligence model 202 is described and illustrated in further detail.

[0045] At step 306, at least one probability value is displayed to the user. In some examples, such as reference... Figure 12 As further described and shown, the at least one probability value is displayed to the user on a display device 1242 (not shown).

[0046] At step 308, event classifier 154 receives a classification of at least one event based on a probability threshold. The probability threshold can be a default threshold or a probability threshold received from a user. In some examples, such as reference... Figure 12 As shown and described in further detail, a probability threshold is received from the user via one or more input devices 1226 and then transmitted to a processing device 1202 (1226, 1202, 1236 and 1206 are not shown) via an input / output interface 1236 and a system bus 1206.

[0047] The probability threshold is the threshold that the event classifier 154 uses to determine whether an event should be classified as being within the group of interest (ROI). (i.e., any probability value greater than the probability threshold is determined to be within the ROI and can be assigned a value of 1 if binary classification information is assigned; conversely, any probability value less than the probability threshold is determined to be outside the ROI and can be assigned a value of 0 if binary classification information is assigned.) Users can adjust the probability threshold to require events to have a higher or lower probability of being included in the ROI. A higher probability threshold will output fewer events, which are more likely to be included in the ROI. Conversely, a lower probability threshold will output more events, but these events may be less likely to be included in the ROI. In some examples, users can optimize the probability threshold based on the visual histogram appearance, or optimize the probability threshold to include a desired number of events that have the highest probability of being included in the ROI.

[0048] At step 310, classifier 154 generates output. This output is generated by comparing each probability value to a probability threshold, identifying at least one inclusive event (i.e., an event within the group of interest) when the probability value exceeds the threshold (or some function of the threshold), and displaying that at least one inclusive event to the user. In some examples, when the probability value does not exceed the probability threshold, event classifier 154 identifies at least one non-inclusive event as not within the group of interest. Event classifier 154 can filter out the at least one non-inclusive event from at least one event.

[0049] Each probability value is compared to a probabilistic numerical value to produce an output as a binary representation. Events identified as being within the group of interest (i.e., events with assigned probability values ​​greater than a probability threshold) are assigned an output value of 1. Conversely, events identified as being outside the group of interest (i.e., events with assigned probability values ​​less than a probability threshold) are assigned an output value of 0. After comparing each probability value with a probability threshold, one or more inclusive events, including those with an output value of 1, are identified. Each inclusive event can then be displayed to the user (e.g., in the form of a table, scatter plot, or histogram) to show the events identified as being within the group of interest based on the probability threshold provided by the user.

[0050] Furthermore, when reviewing the output, the user can choose to optimize the output by adjusting the probability thresholds provided to the event classifier 154. If the user chooses to adjust the probability thresholds, the event classifier 154 returns to step 306 to display at least one probability value to the user and allow the user to input a new probability threshold. In some examples, the event classifier 154 receives one or more additional probability thresholds from the user and generates one or more additional outputs corresponding to the one or more additional probability thresholds.

[0051] Once the adjusted probability threshold is received from the user, an adjusted output can be generated based on that threshold. Users can adjust the probability threshold without rerunning the experiment with the new threshold, reducing the resource and time costs of retraining the AI ​​model. In some examples, the binary representations associated with one or more events can be manually adjusted by the user.

[0052] Figure 4 The manual adjustment method is shown. Figures 2 to 3 An example of a method for autonomously generating flow cytometry data using an artificial intelligence model 202.

[0053] Method 400 includes a step 402 of examining flow cytometry data. The user examines flow cytometry data comprising at least one event. Each event includes a probability value assigned to it by an event classifier 154, indicating the probability that at least one event falls within the population of interest. The user examines the flow cytometry data to determine various characteristics of the data, such as the number of events and the probability assigned to each event. (Refer to above) Figure 3 The process of receiving flow cytometry data, assigning probability values ​​to each event, and displaying the events to the user is described and explained.

[0054] Method 400 includes a step 404 of providing a probability threshold to event classifier 154. A user can provide the probability threshold input to event classifier 154 based on observations made during step 402. In some examples, the user may observe many events that include high probability values ​​and choose to provide a high probability threshold to correspond to the observed high probabilities. In another example, the user may observe fewer events that include high probability values ​​and choose to provide a lower probability threshold to increase the number of inclusive events with probability values ​​greater than the probability threshold.

[0055] Method 400 includes step 406 of receiving output from event classifier 154. (See above reference...) Figure 3 As described, the output includes binary representations, where events include non-inclusive events assigned a value of 0 and inclusive events assigned a value of 1. The output is displayed to the user, who can repeat method 400 by re-examining the flow cytometry data and output, providing an adjusted probability threshold to event classifier 154, and receiving the adjusted output from event classifier 154.

[0056] Figure 5 It shows Figures 2 to 4 An example of an artificial intelligence model 202 is provided, which has at least one input 500, at least one node 512 in a first hidden layer 510, at least one node 522 in a second hidden layer 520, and an output 530 produced by the artificial intelligence model 202. In some examples, the artificial intelligence model 202 includes a machine learning algorithm.

[0057] In some examples, at least one input 500 includes a first input 502, a second input 504, or any number of inputs 506. At least one input 500 includes flow cytometry data collected from the flow cytometry system 100.

[0058] The first hidden layer 510 includes at least one node 512. In some examples, the first hidden layer includes multiple nodes 512, 514, 516, or any number of nodes 518. At least one node 512 receives at least one input 500 and uses a mathematical function to transform the at least one input 500 into an output sent to the second hidden layer 520. In some examples, the specific mathematical function used by at least one node 512 is manually entered by the user. In some examples, the specific mathematical function used by at least one node 512 is determined by the architecture of the artificial intelligence model 202 and the machine learning algorithm used to train the artificial intelligence model 202. In some examples, the machine learning algorithm iteratively adjusts the weights and biases of the connections between nodes within the hidden layers to identify and extract desired features from the flow cytometry data. In some examples, the number of hidden layers, the number of nodes in each hidden layer, and other hyperparameters can be adjusted to optimize the accuracy and computational efficiency of the artificial intelligence model 202.

[0059] The second hidden layer 520 includes at least one node 522. In some examples, the second hidden layer 520 includes multiple nodes 522, 524, 526, or any number of nodes 528. The second hidden layer 520 transforms mathematical functions to transform the data received from the first hidden layer 510 and generate an output 530 (e.g., Figures 3 to 4 (The output is described and shown in the diagram). In some examples, the mathematical function performed by the second hidden layer 520 is the same as or substantially similar to the mathematical function performed by the first hidden layer 510. In other examples, the mathematical function performed by the second hidden layer 520 is different from the mathematical function performed by the first hidden layer 510. The artificial intelligence model 202 may include any number of hidden layers (i.e., one hidden layer, two hidden layers, or more than two hidden layers).

[0060] Figure 6 The training method is shown. Figures 2 to 5 An example of an artificial intelligence model 202 for evaluating flow cytometry data 600.

[0061] Method 600 includes step 602 of collecting flow cytometry data. Using the above reference... Figure 1 The flow cytometer system 100 shown and described is used to collect flow cytometry data.

[0062] Method 600 includes a step 604 of verifying flow cytometry data. At step 604, the collected flow cytometry data is verified by a user. In some examples, the user is an expert in analyzing flow cytometry data. The user can verify and analyze the flow cytometry data by creating one-dimensional and / or two-dimensional scatter plots that visually represent the data.

[0063] Method 600 includes step 606 of identifying one or more populations of interest (e.g., white blood cells) within flow cytometry data by labeling events corresponding to one or more populations of interest. In some examples, one or more populations of interest are manually identified by the user. The flow cytometry data may be arbitrated or compared with independent datasets to ensure that one or more populations of interest are accurately provided to the event classifier 154.

[0064] Method 600 includes step 608 of generating a machine learning algorithm. Event classifier 154 uses stream cytometry data to create a machine learning algorithm, providing one or more groups of interest, and creates output including an indication of whether each event should be classified within one or more groups of interest.

[0065] In some examples, supervised learning is used to enhance the machine learning algorithm in one or more training iterations, using a labeled dataset corresponding to one or more correct outputs. The user can provide flow cytometry data, including scatter plots of one or more populations of interest, and indicators within the scatter plots indicating whether an event is included in one or more populations of interest. The machine learning algorithm can then be generated by mapping the input data (including the scatter plots and populations of interest) to the output data (indicators of whether each event is within a population of interest) to identify patterns and relationships within the data that can be used to predict whether an event is within a population of interest. The supervised training process involves iteratively adjusting the parameters of the machine learning algorithm based on the differences between the algorithm's output and the correct outputs provided by the user. Once the machine learning algorithm is trained, it can be tested on new data (as discussed in step 610 below).

[0066] In other examples, unsupervised learning is used to enhance machine learning algorithms in one or more training iterations by discovering inherent patterns or structures in the provided data without explicit guidance from the user.

[0067] Method 600 includes step 610, using a machine learning algorithm to evaluate new flow cytometry data, generating output predictions of whether an event is within one or more populations of interest, and comparing the output predictions from the machine learning algorithm with labeled events corresponding to the populations of interest to generate a confidence rating. The confidence rating measures the ability of the machine learning algorithm to analyze an event and correctly predict whether the event is within the population of interest. In some examples, the confidence rating is a numerical value between 0 and 1. In some examples, the confidence rating is a percentage between 0 and 100.

[0068] Method 600 includes a step 612 to assess whether the confidence rating exceeds a threshold. This is done by comparing the confidence rating generated in step 610 with a threshold input by the user. Method 600 includes a step 616, whereby if the confidence rating produces a result that exceeds or meets the accuracy specification, the machine learning algorithm is locked (i.e., the event classifier 154 will not make any additional adjustments to the machine learning algorithm). Method 600 includes a step 614, allowing the user to collect more training data and repeat the process.

[0069] Furthermore, once additional training data has been collected, any of steps 602, 604, 606, and 608 can be repeated to improve the training of the machine learning algorithm and better predict the correct output.

[0070] Figure 7 An exemplary gate adjustment performed using manual gate setting is shown, wherein the amorphous region is adjusted. Figure 7 The graph includes the emission fluorescence intensity of the fluorescent label (“CD45-FITC”) 704 along the x-axis and the intensity of the side-scattered light 702 measured by the side-scatter detector along the y-axis. The graph includes a number of events 708 detected by the detector 124 within the flow cytometry system 100. Furthermore, the graph includes a gating region 706 identified by a polygonal shape plotted in the lower right corner of the graph.

[0071] In some examples, and as referenced above. Figure 6 As described and illustrated, the gate region is provided by the user to train a machine learning algorithm. In some examples, the machine learning algorithm proactively identifies events within the gate region 706 based on unique characteristics of events from the waveform acquisition device 140.

[0072] Figure 8 An example of probabilistic event classification performed by waveform analysis device 150 corresponding to the selection of a 50% probability threshold is shown. In some examples, the probability threshold may include a default setting (e.g., 50%) that can be adjusted by the user and overridden. Figure 8 Scatter plot 800 and histogram 810 are shown. The scatter plot illustrates the intensity of lateral scattered light 802 on the x-axis and the intensity of forward scattered light 804 on the y-axis. Furthermore, scatter plot 800 shows multiple events 806 that have been selected by event classifier 154 as events occurring in… Figure 7 The gate area 706 shown in the diagram.

[0073] Histogram 810 includes the probability 812 of an event being within the population of interest on the x-axis, and the number of events with a certain probability value 814 on the y-axis. Furthermore, histogram 810 shows a probability threshold 814 (which is set to 50%). In some examples, the probability of an event being within the population of interest is a value between 0 and 1. In some examples, the probability of an event being within the population of interest is a percentage between 0 and 100. In other examples, the probability of an event being within the population of interest is scaled to include a range that includes both minimum and maximum values ​​to correspond to existing flow cytometry software (e.g., such as...). Figures 8 to 10 (The range of 0 to 1023 shown).

[0074] As shown in histogram 810, and as a non-limiting example, the flow cytometry data includes a total of 25,939 events. Adjusting the probability threshold 814 to 50% will filter out all events determined by the machine learning algorithm to have a probability value less than 50%, while displaying all inclusive events. As shown in histogram 810, and as a non-limiting example, the number of inclusive events is 10,134, which is approximately 39.1% of the total events included in the flow cytometry data. Generally, increasing the probability threshold 814 will reduce the number of inclusive samples and provide a greater probability that the inclusive samples displayed to the user are within the population of interest. Conversely, decreasing the probability threshold will increase the number of inclusive samples while providing a reduced probability that the inclusive samples displayed to the user are within the population of interest. Figure 9 Examples of increasing the probability threshold are shown and described in the figure.

[0075] Figure 9 An example of probability event classification performed by waveform analysis device 150 corresponding to the selection of a probability threshold of 95 percent is shown. Figure 9 It shows something similar to Figure 8 Example scatter plot 800 and example histogram 810 are shown in example scatter plot 900 and example histogram 910. In example histogram 910, the probability threshold 912 has been increased to a 95% threshold probability, and inclusive events 914 that include the assigned probability values ​​exceeding the probability threshold 912 are shown. Figure 9 As shown, and as a non-limiting example, the flow cytometry data includes a total of 25,939 events. Adjusting the probability threshold 912 to 95% shows 6,693 events determined by the machine learning algorithm to have a probability value greater than 95%. Therefore, by comparing the probability value of each event to the 95% probability threshold 912, the machine learning algorithm determines... Figure 7 Approximately 25.8% of the events 708 shown (6,693 of the total 25,939 events) are located within the gate region 706.

[0076] Figure 10 Examples of histograms 1000 and 1010 are included, representing different group classifiers. These two classifiers have different probability distributions. When the groups are more difficult to separate, the probability distribution is more dispersed (as in case 1010), which reduces the probability distribution caused by the classifiers' differences in probability distribution. Figures 1 to 2 The confidence rating generated by the waveform analysis equipment. Figure 10 This includes a high-confidence histogram (1000) and a low-confidence histogram (1010). Similar to... Figure 8 The high-confidence histogram 1000 and low-confidence histogram 1010 show the probability of an event being within the group of interest on the x-axis and the number of inclusive events 816 with probability values ​​exceeding the probability threshold 814 on the y-axis. Furthermore, the high-confidence histogram 1000 and low-confidence histogram 1010 include... Figure 8 The 50% probability threshold is 814. The high-confidence histogram 1000 and low-confidence histogram 1010 comprise the distribution of probability values ​​assigned to each event within the flow cytometry data by the machine learning algorithm. A concentration of probability values ​​within a finite range at the upper and lower ends of the histogram (as shown in the high-confidence histogram 1000) indicates that the output generated by the machine learning algorithm (i.e., determining which events are inclusive and which are not) is correct and has a higher confidence rating. Conversely, a larger distribution of probability values ​​(as shown in the low-confidence histogram 1010) indicates that the output generated by the machine learning algorithm is correct and has a lower confidence rating.

[0077] Figure 11 Flow cytometry data are shown at three different resolutions. Figure 11 This includes a 6-bit display 1100 showing events at a 6-bit resolution 1102, an 8-bit display 1110 showing events at an 8-bit resolution 1104, and a 9-bit display 1120 showing events at a 9-bit resolution 1106. Each display includes the same number of events, but the resolution affects the size of the event clusters and the accuracy with which users can classify events into groups of interest. When performing a visual inspection of the data, higher resolution tends to increase the accuracy of limiting events to groups of interest, while lower resolution tends to decrease the accuracy. Using a probability threshold to determine whether an event is included in the group of interest eliminates potential errors introduced by the subjectivity of visual inspection. Furthermore, the event classifier 154 is able to interpret the data in more than two dimensions to perform this determination, while Figure 11 The visual inspection of the diagram shown is limited to two dimensions.

[0078] Figure 12An exemplary architecture of a computing device that can be used to implement various aspects of this disclosure is shown. The computing device 1200 can be used to implement various aspects of this disclosure, including aspects of the waveform acquisition device 140 and the waveform analysis device 150 as described above. Furthermore, the computing device 1200 can be used to execute the operating system, applications, and software modules (including software engines) described herein.

[0079] Computing device 1200 includes at least one processing device 1202, such as a central processing unit (CPU). In this example, computing device 1200 also includes system memory 1204 and a system bus 1206 that couples various system components, including system memory 1204, to at least one processing device 1202. System bus 1206 is one of any number of types of bus structures using any of a variety of bus architectures, including a memory bus or memory controller, a peripheral bus, and a local bus.

[0080] System memory 1204 includes read-only memory (ROM) 1208 and random access memory (RAM) 1210. A basic input / output system 1212 is typically stored in ROM 1208, containing basic routines for transferring information within computing device 1200, such as during startup. In some examples, system memory 1204 has a large memory capacity, such as RAM equal to or greater than 1 terabyte. RAM can be used to load and subsequently analyze waveform data (e.g., raw waveform data stored in a raw waveform data file, which may include digitized waveform data).

[0081] In some embodiments, computing device 1200 further includes an auxiliary storage device 1214, such as a hard disk drive, for storing digital data. Auxiliary storage device 1214 is connected to system bus 1206 via auxiliary storage interface 1216. In some examples, auxiliary storage device 1214 and its associated computer-readable medium provide non-volatile storage for computer-readable instructions (including application programs and program modules), data structures, and other data to computing device 1200.

[0082] While the exemplary environment described herein employs a hard disk drive as a secondary storage device, other types of computer-readable storage media are used in other embodiments. Examples of these other types of computer-readable storage media include magnetic tape cassettes, flash memory cards, digital video disks, Bernoulli cassette tapes, optical disc read-only memory, digital universal disk read-only memory, random access memory, or read-only memory. Some embodiments include non-transitory media. Additionally, such computer-readable storage media may include local storage or cloud-based storage.

[0083] Several program modules, including an operating system 1218, one or more application programs 1220, other program modules 1222 (such as the software engine described herein), and program data 1224, can be stored in the secondary storage device 1214 or system memory 1204. The computing device 1200 can use any suitable operating system, such as Microsoft Windows™, Google Chrome™, Apple OS, and any other operating system suitable for computing devices.

[0084] In some examples, a user provides input to computing device 1200 through one or more input devices 1226. Examples of input devices 1226 include a keyboard 1228, a mouse 1230, a microphone 1232, and a touch sensor 1234 (e.g., a touchpad or touch-sensitive display). Additional examples include additional types of input devices 1226 or fewer types of input devices 1226. Input devices 1226 are connected to at least one processing device 1202 via input / output interfaces 1236 coupled to system bus 1206. Input / output interfaces 1236 may include any number of input / output interfaces, such as parallel ports, serial ports, game ports, or universal serial buses. In some possible implementations, wireless coupling between input devices 1226 and input / output interfaces 1236 is also possible, such as via infrared, Bluetooth®, 802.11a / b / g / n, cellular, or other radio frequency communication systems.

[0085] In this example embodiment, the display device 1242 (e.g., a monitor, liquid crystal display, projector, or touch-sensitive display) is also connected to the system bus 1206 via the video adapter 1240. In addition to the display device 1242, the computing device 1200 may also include various other peripheral devices (not shown), such as speakers or printers.

[0086] When used in a local area network (LAN) or wide area network (WAN) environment (e.g., the Internet), computing device 1200 is typically connected to a network via a network interface 1238 (e.g., an Ethernet interface). Other possible implementations use other communication devices. For example, some implementations of computing device 1200 include a modem for cross-network communication.

[0087] Computing device 1200 typically includes at least some form of computer-readable medium. Computer-readable medium includes any available medium that can be accessed by computing device 1200. By way of example, computer-readable medium includes computer-readable storage media and computer-readable communication media.

[0088] Computer-readable storage media include volatile and non-volatile, removable and non-removable media implemented in any device configured to store information such as computer-readable instructions, data structures, program modules or other data. Computer-readable storage media include, but are not limited to, random access memory, read-only memory, electrically erasable programmable read-only memory, flash memory, optical disc read-only memory, digital universal disk or other optical storage devices, magnetic tape cassettes, magnetic tape, disk storage devices or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computing device. Computer-readable storage media do not include computer-readable communication media.

[0089] Computer-readable communication media typically embody computer-readable instructions, data structures, program modules, or other data in the form of modulated data signals, such as carrier waves or other transmission mechanisms, and include any information delivery medium. The term "modulated data signal" refers to a signal whose characteristics are set or altered in a manner that encodes information within the signal. By way of example, computer-readable communication media include: wired media, such as wired networks or direct wired connections; and wireless media, such as acoustic media, radio frequency media, infrared media, and other wireless media. Any combination of the above is also included within the scope of computer-readable media.

[0090] Figure 12 The computing device 1200 shown is also an example of a programmable electronic device that may include one or more such computing devices, and when multiple computing devices are included, such computing devices may be coupled together with a suitable data communication network to jointly perform the various aspects disclosed herein.

[0091] This disclosure includes the subjects set forth in the following numbered clauses:

[0092] Clause 1. A method for analyzing particles within a flow cytometry system using a computing device, the method comprising: receiving flow cytometry data from the flow cytometry system, the flow cytometry data including at least one event; assigning probability values ​​to the at least one event, the probability values ​​indicating the probability that the at least one event is associated with a group of interest; displaying a probability threshold to a user, wherein the probability threshold classifies the event as belonging to the group of interest; allowing the user to adjust the probability threshold; and generating output by: comparing the probability values ​​of the at least one event with the probability threshold; and assigning classification information to at least one inclusive event, wherein the at least one inclusive event includes probability values ​​exceeding the probability threshold.

[0093] Clause 2. The method according to Clause 1 further includes identifying at least one non-inclusive event as not within the group of interest when the probability value does not exceed the probability threshold.

[0094] Clause 3. The method according to any one of Clauses 1 to 2 further includes filtering the at least one non-inclusive event from the at least one event.

[0095] Clause 4. The method according to Clause 2, wherein the at least one inclusive event and the at least one non-inclusive event are classified using binary representation.

[0096] Clause 5. The method according to any one of Clauses 1 to 4, wherein the probability threshold comprises a value between zero and one.

[0097] Clause 6. The method according to any one of Clauses 1 to 5, wherein the at least one inclusion event is identified by applying one or more filters.

[0098] Clause 7. The method according to Clause 6, wherein the at least one inclusion event is identified by a series of filters including at least one of a bandpass filter, a low-pass filter, and a high-pass filter.

[0099] Clause 8. The method according to Clause 6, wherein the at least one inclusive event is classified into a blood cell population by a Boolean logical binary representation of the event, wherein the at least one event is represented as an inclusive event or a non-inclusive event.

[0100] Clause 9. The method according to any one of Clauses 1 to 8, wherein the output is generated using a machine learning algorithm.

[0101] Clause 10. The method according to Clause 9 further includes training the machine learning algorithm by a supervised method, the supervised method generating confidence ratings by comparing the output with an independent dataset.

[0102] Clause 11. The method according to Clause 10 further comprises: adjusting the output generated by the machine learning algorithm by changing the probability threshold; and generating at least one adjusted output.

[0103] Clause 12. The method according to Clause 10, wherein the independent dataset is manually adjusted by the user.

[0104] Clause 13. The method according to Clause 10 further includes locking the machine learning algorithm for use when the confidence rating exceeds a threshold.

[0105] Clause 14. The method according to Clause 1 further includes: receiving one or more probability value events; and generating one or more additional outputs corresponding to one or more additional probability thresholds.

[0106] Clause 15. The method according to any one of Clauses 1 to 14 further comprises separating a desired number of events for further analysis by: generating a histogram including the at least one event, wherein the histogram corresponds to the number of events having the probability value; identifying the desired number of events to be analyzed; and adjusting the probability threshold to include the desired number of events while excluding other events.

[0107] Clause 16. The method according to any one of Clauses 1 to 15 further includes displaying the at least one containing event to the user.

[0108] Clause 17. A method for classifying events within a flow cytometry system, the method comprising: examining flow cytometry data including at least one event, the at least one event including a probability value indicating the probability that the at least one event is associated with a population of interest; allowing adjustment of a probability threshold on the flow cytometry system to filter one or more of the at least one events when the probability value is greater than the probability threshold; and receiving an output from the flow cytometry system including at least one inclusive event exceeding the probability threshold.

[0109] Clause 18. The method according to Clause 17 further comprises training the flow cytometry system by: generating a confidence rating by comparing the output with an expected output; adjusting the probability threshold and retraining the machine learning algorithm when the confidence rating is less than a threshold; instructing the machine learning algorithm to provide at least one adjusted output; and checking the at least one adjusted output.

[0110] Clause 19. The method according to Clause 18, wherein the at least one adjusted output is examined by comparing the at least one adjusted output with the independent dataset;

[0111] Generate at least one adjusted confidence rating; and compare the at least one adjusted confidence rating with the threshold.

[0112] Clause 20. The method according to Clause 18 further includes locking the machine learning algorithm when the confidence rating exceeds the threshold.

[0113] Clause 21. A flow cytometry system for analyzing particles, the flow cytometry system comprising: at least one processing device; and a non-transitory computer-readable storage medium storing instructions, the instructions, when executed by the processing device, causing the at least one processing device to: receive flow cytometry data from the flow cytometry system, the flow cytometry data including at least one event; assign probability values ​​to the at least one event, the probability values ​​indicating the probability that the at least one event is located within a population of interest; display the at least one probability value to a user; receive a selection of a probability threshold from the user; and generate output by: comparing the at least one probability value to the probability threshold; and identifying at least one inclusion event when the at least one probability value exceeds the probability threshold.

[0114] Clause 22. The flow cytometry system according to Clause 21 further includes the flow cytometry system comprising: a light source for generating a light beam toward an interrogation zone; and an optical system including a detector for detecting radiant light from particles passing through the light beam in the interrogation zone.

[0115] Clause 23. The flow cytometry system according to Clause 22, wherein the non-transitory computer-readable storage medium stores instructions, which, when executed by the processing device, further cause the processing device to use the light source and the optical system to detect flow cytometry data from particles passing through the interrogation area.

[0116] Clause 24. The flow cytometry system according to any one of Clauses 21 to 23 further includes one or more electrodes for measuring the electrical impedance of particles passing through the interrogation zone.

[0117] Clause 25. The flow cytometry system according to Clause 24, wherein the non-transitory computer-readable storage medium stores instructions, which, when executed by the processing device, further cause the processing device to detect flow cytometry data from particles passing through the interrogation region using the one or more electrodes.

[0118] Clause 26. A flow cytometry system according to any one of Clauses 21 to 25, wherein the non-transitory computer-readable storage medium stores instructions, which, when executed by the processing device, further cause the processing device to display the at least one inclusive event to the user.

[0119] The above description is illustrative only and not restrictive. Many variations of the invention will become apparent to those skilled in the art upon reading this disclosure. Therefore, the scope of the invention should not be determined by reference to the above description, but rather by reference to the pending claims and their full scope or equivalents.

[0120] Without departing from the scope of the invention, one or more features from any embodiment may be combined with one or more features from any other embodiment.

Claims

1. A method for analyzing particles within a flow cytometry system using a computing device, the method comprising: Receive flow cytometry data from the flow cytometry system, the flow cytometry data including at least one event; Assign a probability value to the at least one event, the probability value indicating the probability that the at least one event is associated with a group of interest; Display a probability threshold to the user, wherein the probability threshold categorizes events into those belonging to the group of interest; Allow the user to adjust the probability threshold; and The output is generated as follows: Compare the probability value of the at least one event with the probability threshold; and Assign classification information to at least one inclusive event, wherein the at least one inclusive event includes a probability value exceeding the probability threshold.

2. The method according to claim 1, further comprising: When the probability value does not exceed the probability threshold, at least one non-inclusive event is identified as not being in the group of interest.

3. The method according to claim 2, further comprising: Filter the at least one non-inclusive event from the at least one event.

4. The method according to claim 2, wherein, The at least one containing event and the at least one non-containing event are classified using binary representation.

5. The method according to claim 1, wherein, The probability threshold includes values ​​between zero and one.

6. The method according to claim 1, wherein, The at least one inclusion event is identified by applying one or more filters.

7. The method according to claim 6, wherein, The at least one inclusive event is identified by a series of filters including at least one of a bandpass filter, a low-pass filter, and a high-pass filter.

8. The method according to claim 6, wherein, The at least one inclusive event is classified into the blood cell population by using the Boolean logic binary representation of the event, wherein the at least one event is represented as either an inclusive event or a non-inclusive event.

9. The method according to claim 1, wherein, The output is generated using a machine learning algorithm.

10. The method of claim 9, further comprising: Confidence ratings are generated by comparing the outputs with independent datasets, and the machine learning algorithm is trained using a supervised method.

11. The method of claim 10, further comprising: The output generated by the machine learning algorithm is adjusted by changing the probability threshold; as well as Generate at least one adjusted output.

12. The method according to claim 10, wherein, The independent dataset is manually adjusted by the user.

13. The method of claim 10, further comprising: When the confidence rating exceeds a threshold, the machine learning algorithm is locked for use.

14. The method according to claim 1, further comprising: Receive one or more probability value events; as well as Generate one or more additional outputs corresponding to one or more additional probability thresholds.

15. The method according to claim 1, further comprising: The desired number of events are separated for further analysis using the following method: Generate a histogram including the at least one event, wherein the histogram corresponds to the number of events having the probability value; Identify the desired number of events to be analyzed; and The probability threshold is adjusted to include the desired number of events while excluding other events.

Images