Cell activity detection method based on cell image analysis and related device

Abstract

The embodiment of the invention provides a cell activity detection method based on cell image analysis and a related device. The method comprises the following steps: acquiring a target cell image; performing batch automatic segmentation on the target cell image by using a Cellpose platform to generate a segmented mask image; inputting the target cell image and the segmented mask image into an illic platform for classification, and determining the number of normal cells and the number of abnormal cells in the target cell image; and obtaining a cell activity detection result corresponding to the target cell image according to the ratio of the number of the abnormal cells to the total cell number, the total cell number being the sum of the number of the normal cells and the number of the abnormal cells. Based on this, the embodiment of the invention does not need to depend on labeling training on a large amount of data, adopts a simple image analysis means, can efficiently and accurately evaluate the cell activity by calculating the ratio of the number of abnormal cells to the number of total cells, and has the advantages of low cost, rapidness and high efficiency.

Description

Technical Field

[0001] Embodiments of the present invention relate to the technical field of cell image recognition, and in particular, to a method and related device for detecting cell viability based on cell image analysis. Background Art

[0002] Cell viability assessment and toxicity analysis are indispensable links in biomedical research, especially in the research fields such as environmental pollutants and drug toxicity. Traditional cell viability detection methods, such as colorimetric methods (CCK8 method, MTT method, WST-1 method, etc.), flow cytometry, etc., although they can provide quantitative data on cell viability, these methods usually rely on a variety of biochemical indicators and chemical dyes, with high costs and complex operations.

[0003] With the rapid development of computer vision and deep learning technologies, methods based on cell image analysis have gradually become an effective means for cell viability assessment. Through microscope image analysis, especially the classification and segmentation of cell images, researchers can more directly extract information from the morphological changes of cells to evaluate cell damage. For example, the U-Net and YOLOv3 models are used for segmentation and classification to predict the degree of oxidative damage of red blood cells. The U-Net model is used for the segmentation of cell images, while the YOLOv3 model is used for object detection and classification. This method has achieved certain success in the detection of red blood cell damage. However, it also has some significant limitations.

[0004] The effective training of the U-Net and YOLOv3 models depends on a large amount of high-quality image annotation data. Although in red blood cell images, the cell morphology is relatively regular and the annotation process is relatively simple, for images with irregular cell morphology or cell overlap, the complexity and cost of manual annotation increase significantly. Deep learning models such as U-Net and YOLOv3 usually require a large amount of training data, and these models are sensitive to changes in cell types and experimental conditions. In the case of different cell types, changes in pollutant types, or inconsistent experimental conditions, the generalization ability of the model is often poor, and it is necessary to re-train and tune, increasing the complexity and time cost of the experiment. Summary of the Invention

[0005] Embodiments of the present invention provide a method and related device for detecting cell viability based on cell image analysis, which do not need to rely on the annotation training of a large amount of data, adopt simple image analysis means, and can efficiently and accurately evaluate cell viability and toxicity by calculating the ratio of the number of abnormal cells to the total number of cells, with the advantages of low cost, fast speed, and high efficiency.

[0006] In a first aspect, embodiments of the present invention provide a method for detecting cell viability based on cell image analysis, including:

[0007] Obtain a target cell image to be detected;

[0008] Use the Cellpose platform to perform batch automated segmentation on the target cell image to generate a segmentation mask image;

[0009] Input the target cell image and the segmentation mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image;

[0010] Obtain the cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, where the total number of cells is the sum of the number of normal cells and the number of abnormal cells.

[0011] In some embodiments, the obtaining of the target cell image to be detected includes:

[0012] Configure the aperture parameter and focal length parameter of the microscope;

[0013] Take a picture of the target sample cells through the microscope to obtain the target cell image.

[0014] In some embodiments, the target sample cells include red blood cells and adherent cells.

[0015] In some embodiments, the using of the Cellpose platform to perform batch automated segmentation on the target cell image to generate a segmentation mask image includes:

[0016] Identify the cell type of the target sample cells in the target cell image;

[0017] Determine the segmentation parameters according to the cell type, where the segmentation parameters include the cell diameter range and the flow threshold;

[0018] Based on the segmentation parameters, input the target cell image into the cytoplasmic recognition model loaded on the Cellpose platform for semantic segmentation to generate the segmentation mask image, where the cytoplasmic recognition model uses cyto2 pre-trained weights.

[0019] In some embodiments, the inputting of the target cell image and the segmentation mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image includes:

[0020] Configure the classification feature parameters of the ilastic platform, where the classification feature parameters include standard object features, 3D convex hull features, and 2D skeleton features;

[0021] Input the target cell image and the segmentation mask image into the ilastic platform, so that the ilastic platform classifies the target sample cells according to the standard object features, the 3D convex hull features, and the 2D skeleton features, and outputs cell classification information;

[0022] Determine the number of normal cells and the number of abnormal cells in the target cell image according to the cell classification information.

[0023] In some embodiments, obtaining the cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells includes:

[0024] Calculate the sum of the number of normal cells and the number of abnormal cells to obtain the total number of cells;

[0025] Calculate the ratio of the number of abnormal cells to the total number of cells to obtain the cell viability detection result corresponding to the target cell image, where the cell viability detection result includes the degree of cell viability decline, and the degree of cell viability decline is used to characterize the damage condition of the cells.

[0026] In some embodiments, the cell viability detection result includes:

[0027] The higher the ratio, the greater the degree of cell viability decline;

[0028] The lower the ratio, the smaller the degree of cell viability decline.

[0029] In a second aspect, an embodiment of the present invention further provides a cell viability detection device based on cell image analysis, and the device includes:

[0030] An acquisition module, configured to acquire a target cell image to be detected;

[0031] A segmentation module, configured to automatically segment the target cell image in batches by using the Cellpose platform to generate a segmentation mask image;

[0032] A classification module, configured to input the target cell image and the segmentation mask image into the ilastic platform for classification to determine the number of normal cells and the number of abnormal cells in the target cell image;

[0033] A calculation module, configured to obtain the cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, where the total number of cells is the sum of the number of normal cells and the number of abnormal cells.

[0034] In a third aspect, an embodiment of the present invention further provides an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, it implements the cell viability detection method based on cell image analysis as described in the first aspect.

[0035] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium storing computer-executable instructions for executing the cell viability detection method based on cell image analysis as described in the first aspect.

[0036] According to the cell viability detection method and related device based on cell image analysis provided by the embodiments of the present invention, the cell viability detection method based on cell image analysis includes: obtaining a target cell image to be detected; using the Cellpose platform to perform batch automatic segmentation on the target cell image to generate a segmentation mask image; inputting the target cell image and the segmentation mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image; obtaining a cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, where the total number of cells is the sum of the number of normal cells and the number of abnormal cells. This method does not rely on a complex label annotation process, nor does it require the use of expensive deep learning models. Instead, by using two relatively lightweight deep learning tools, Cellpose and ilastic, where Cellpose is used for cell segmentation and ilastic is used for cell classification. Compared with traditional methods, both of them have lower requirements for training data and can provide relatively accurate segmentation and classification results under limited data, with the technical advantages of lower operation complexity, higher applicability, and lower resource consumption. Especially Cellpose, which has a powerful adaptive segmentation ability and can handle cells with different morphologies and densities, and can also perform well with a small amount of data. Based on this, the embodiments of the present invention do not need to rely on annotating and training a large amount of data, adopt simple image analysis means, and can efficiently and accurately evaluate cell viability and toxicity by calculating the ratio of the number of abnormal cells to the total number of cells, with the advantages of low cost, fast speed, and high efficiency. Description of the Drawings

[0037] Figure 1 is a flowchart of the cell viability detection method based on cell image analysis provided by an embodiment of the present invention;

[0038] Figure 2 is a comparison diagram of the original cell image, the image after cell segmentation, and the image after cell classification provided by an embodiment of the present invention;

[0039] Figure 3It is a graph showing the change trend of the proportion of abnormal cells under different concentrations of diniconazole provided by an embodiment of the present invention;

[0040] Figure 4 It is a graph showing the corresponding relationship between cell viability and diniconazole concentration provided by an embodiment of the present invention;

[0041] Figure 5 It is a graph showing the corresponding relationship between the proportion of abnormal cell number and cell viability provided by an embodiment of the present invention;

[0042] Figure 6 It is a schematic structural diagram of a cell viability detection device based on cell image analysis provided by an embodiment of the present invention;

[0043] Figure 7 It is a schematic diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0044] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0045] It should be noted that although functional module division is performed in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different module division in the device or a different order in the flowchart. Terms such as "first" and "second" in the description, claims and the following drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence.

[0046] In the embodiments of the present invention, words such as "furthermore", "exemplarily" or "optionally" are used to represent examples, illustrations or explanations, and should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Using words such as "furthermore", "exemplarily" or "optionally" is intended to present related concepts in a specific manner.

[0047] In order to more conveniently describe the working principle of the embodiments of the present invention later, the following first gives an introduction to the related technical scenarios.

[0048] Cell viability assessment and toxicity analysis are indispensable links in biomedical research, especially in the research fields such as environmental pollutants and drug toxicity. Traditional cell viability detection methods, such as colorimetric methods (CCK8 method, MTT method, WST-1 method, etc.), flow cytometry, etc., although they can provide quantitative data on cell viability, these methods usually rely on a variety of biochemical indicators and chemical dyes, with high costs and complex operations.

[0049] With the rapid development of computer vision and deep learning technologies, methods based on cell image analysis have gradually become an effective means of evaluating cell viability. Through microscopic image analysis, especially the classification and segmentation of cell images, researchers can more directly extract information from the morphological changes of cells to evaluate cell damage. For example, the U-Net and YOLOv3 models are used for segmentation and classification to predict the degree of oxidative damage to red blood cells. The U-Net model is used for the segmentation of cell images, while the YOLOv3 model is used for object detection and classification. This method has achieved some success in the detection of red blood cell damage. However, it also has some significant limitations.

[0050] The effective training of the U-Net and YOLOv3 models depends on a large amount of high-quality image annotation data. Although in red blood cell images, the cell morphology is relatively regular and the annotation process is relatively simple, for images with irregular morphology or cell overlap, the complexity and cost of manual annotation increase significantly. Deep learning models such as U-Net and YOLOv3 usually require a large amount of training data, and these models are sensitive to changes in cell types and experimental conditions. In cases where cell types are different, the types of pollutants vary, or experimental conditions are inconsistent, the generalization ability of the model is often poor, and retraining and tuning are required, increasing the complexity and time cost of the experiment.

[0051] Based on this, the present invention provides a method and related device for detecting cell viability based on cell image analysis. Among them, the method for detecting cell viability based on cell image analysis includes: obtaining a target cell image to be detected; using the Cellpose platform to perform batch automatic segmentation on the target cell image to generate a segmentation mask image; inputting the target cell image and the segmentation mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image; obtaining a cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, where the total number of cells is the sum of the number of normal cells and the number of abnormal cells. This method does not rely on a complex label annotation process, nor does it require the use of expensive deep learning models. Instead, by using two relatively lightweight deep learning tools, Cellpose and ilastic, Cellpose is used for cell segmentation and ilastic is used for cell classification. Compared with traditional methods, both of them have lower requirements for training data and can provide relatively accurate segmentation and classification results under limited data conditions, with technical advantages of lower operation complexity, higher applicability, and lower resource consumption. Especially Cellpose, which has a powerful adaptive segmentation ability and can handle cells with different morphologies and densities, and can also perform well with a small amount of data. Based on this, the embodiments of the present invention do not need to rely on annotating and training a large amount of data. By using simple image analysis means and calculating the ratio of the number of abnormal cells to the total number of cells, it can efficiently and accurately evaluate cell viability and toxicity, with the advantages of low cost, fast speed, and high efficiency.

[0052] The following further elaborates on the embodiments of the present invention in conjunction with the accompanying drawings.

[0053] As Figure 1 shown, Figure 1 FIG. 10 is a flowchart of a method for detecting cell viability based on cell image analysis provided by an embodiment of the present invention. The method for detecting cell viability based on cell image analysis may include but is not limited to steps S101 to S104.

[0054] Step S101: Obtain a target cell image to be detected;

[0055] Step S102: Use the Cellpose platform to perform batch automatic segmentation on the target cell image to generate a segmentation mask image;

[0056] Step S103: Input the target cell image and the segmentation mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image;

[0057] Step S104: Obtain the cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, where the total number of cells is the sum of the number of normal cells and the number of abnormal cells.

[0058] First, an ordinary microscope can be used to take images of the cells to be detected to ensure clear images of the cells are captured for subsequent analysis.

[0059] Second, use the Cellpose platform to perform batch automated segmentation and counting on the collected cell images. The Cellpose platform automatically identifies and segments the contours of each cell through deep learning algorithms to generate segmentation results.

[0060] Third, input the segmented images into the ilastic platform for classification. During the classification process, the ilastic platform can classify cells into two categories: "normal cells" and "abnormal cells" by analyzing the features of the segmented images. Generally speaking, normal cells usually have regular morphologies and complete structures, while abnormal cells may exhibit characteristics such as morphological variations, structural damage, or volume changes, such as unclear cell contours, abnormal cytoplasmic accumulation, and nuclear offset. In this way, the ilastic platform can effectively distinguish normal and abnormal cells, providing reliable data support for subsequent damage assessment.

[0061] Finally, count and calculate the number of abnormal cells (x), and calculate the ratio with the total number of cells (y). This ratio (a) represents the degree of decline in cell viability and reflects the damage condition of the cells (a = x / y).

[0062] This method does not rely on a complex label annotation process, nor does it require the use of expensive deep learning models. Instead, it uses two relatively lightweight deep learning tools, Cellpose and ilastic. Cellpose is used for cell segmentation, and ilastic is used for cell classification. Compared with traditional methods, these two have lower requirements for training data and can provide relatively accurate segmentation and classification results under limited data, with technical advantages of lower operation complexity, higher applicability, and lower resource consumption. Especially Cellpose has a powerful adaptive segmentation ability and can handle cells with different morphologies and densities, and can also perform well with a small amount of data. Based on this, the embodiments of the present invention do not need to rely on the annotation training of a large amount of data, adopt simple image analysis means, and can efficiently and accurately evaluate cell viability and toxicity by calculating the ratio of the number of abnormal cells to the total number of cells, with the advantages of low cost, fast speed, and high efficiency.

[0063] It should be noted that this method can not only be applied to the detection of red blood cell viability, but also be extended to the detection of the viability of other types of cells, especially the application of adherent cell types (such as HUVEC, A549, etc.). Cellpose can effectively segment adherent cells, which is different from the simple morphology of red blood cells and provides a wider applicability. This method can be adaptively adjusted in different types of cell images, avoiding the need to retrain the model due to cell type differences, and improving the generality and practical application scope of the method. The operation process of this method is simple, and it can use an ordinary microscope and standard image analysis tools to evaluate the cell viability level without complex experimental equipment or chemical reagents. The processes of image segmentation and classification are automated, and the changes in cell viability can be obtained quickly, which is suitable for rapid experiments. At the same time, through the correlation analysis of the viability results detected by the CCK8 method, the method provided by the present invention shows similar accuracy to the traditional method in cell viability evaluation, and has strong reliability and broad application prospects.

[0064] It can be understood that for the preparation of experimental samples and microscopic imaging, a cell suspension treated with gradient damage can be used as the research sample, and a inverted microscope system (OLYMPUS CKX53) equipped with a 10-fold objective lens is used for microscopic observation. By adjusting the aperture and focal length, ensure that the images of the cells are clearly visible. To avoid the image being too dark or too bright, resulting in blurred cell morphology or inability to clearly identify, special attention is paid to the adjustment of the exposure time during the operation process to ensure clear cell outlines and complete cell morphology. In addition, ensure that each cell is completely presented during image acquisition to avoid the loss of cell morphology. Through these fine operation steps, the quality of the image data is guaranteed, laying a solid foundation for subsequent cell segmentation and classification.

[0065] It can be understood that for the background segmentation of the original cell images, image segmentation can be implemented based on the deep learning Cellpose platform: First, the present invention imports the original microscopic images into the Cellpose platform and makes personalized settings for the segmentation parameters according to the morphological characteristics of the experimental samples. During the parameter setting stage, the present invention specifies different cell diameter ranges (Cell diameter) according to different cell types, and adjusts the flow threshold (flow_threshold) between 0.6 and 0.9 to ensure the accuracy of cell segmentation. Then, an optimized cytoplasmic recognition model (cyto2 pre-trained weights) is selected for semantic segmentation. Through this step, Cellpose can automatically segment the cell regions while avoiding the influence of cell morphology overlap or other artifacts on the results.

[0066] It is understandable that for multi-modal image feature classification, the original image and the segmented mask image can be input into the ilastik classification platform simultaneously. When using ilastik for classification, it first enters the feature engineering stage. In this stage, the classification effect of cell images is enhanced mainly by selecting appropriate feature parameters. To obtain more representative cell features, the present invention selects three-dimensional features: Standard Object Features (including geometric basic features such as the shape and size of cells), 3D Convex Hull Features, and 2D Skeleton Features. Then, target classification is performed. Click "Live Update" and "Show Predictions" in the settings column on the left side of the software to obtain the expected classification image, ensuring the accuracy and reliability of cell classification. After classification is completed, the final classification result will be saved as image data for subsequent statistics and analysis, such as Figure 2 shown

[0067] It is understandable that for the evaluation of cell viability level, the number of abnormal cells (x) can be counted and calculated, and the ratio is calculated with the total number of cells (y). First, ilastic will output the classification result, marking the classification information of each cell, and then the number of abnormal cells is counted through an algorithm. Subsequently, by calculating the ratio of abnormal cells to total cells (a = x / y), the viability state of cells can be quantified. This ratio (a) reflects the change in cell viability. The higher the value, the stronger the toxic effect of the pollutant on the cells, and the greater the degree of cell viability reduction. This ratio provides an intuitive and quantitative way for cell viability evaluation, and can efficiently and accurately reflect the toxic effect of pollutants or other stimuli on cells.

[0068] To better demonstrate the scientificity and practicality of this method, the present invention verifies the feasibility and reliability of this method through experimental analysis. This experiment uses the cell image analysis method to quantitatively characterize the toxic reactions of three types of cells (HeLa, HUVEC, A549) under different pollutant stimulation conditions, and compares with the cell viability results detected by the traditional CCK-8 method to evaluate the accuracy and applicability of this method.

[0069] The specific operations are as follows:

[0070] (1) The change in the proportion of abnormal-shaped cells under the stimulation of different concentrations of pollutants (diniconazole)

[0071] The three types of cells were respectively inoculated into 6-well culture plates at a density of 1×10^6 cells per well and cultured in a cell incubator at 37°C and 5% CO2 for 24 h to allow the cells to adhere and enter the logarithmic growth phase. After 24 h of cell culture, different concentrations of diniconazole (0, 2.5, 5, 10, 20 μmol / L) were added to the culture medium, and incubation was continued for 24 h. After the treatment, Cellpose was used for cell segmentation, and ilastik was used for classification to obtain the classification results of normal cells and abnormal cells. Calculate the proportion of abnormal cells under different concentrations of diniconazole treatment, and draw a grayscale histogram to visually display the change trend of the number of abnormal cells, as Figure 3 shown. The experimental results obtained by this method show that diniconazole is toxic to all three types of cells, and the higher the concentration, the stronger the toxicity.

[0072] (2) Detection of cell viability

[0073] The three types of cells to be detected were respectively inoculated into 96-well plates, with the number of cells per well between 1,000 and 10,000, and cultured in an incubator at 37°C and 5% CO2 for 24 h to allow the cells to enter the logarithmic growth phase. After the cell culture was completed, different concentrations of diniconazole (0, 2.5, 5, 10, 20 μmol / L) were added to the culture medium, and incubation was carried out for 24 h. After the treatment, 10 μL of CCK-8 reagent (Shanghai Beyotime Biotechnology Co., Ltd.) was added to each well, and incubation was continued for 1 - 4 h. After the incubation, a microplate reader was used to measure the absorbance (OD value) at a wavelength of 450 nm to reflect the metabolic activity of the cells. The untreated cells (0 μmol / L group) were used as the control group, and the relative cell viability of the pollutant treatment group was calculated, and a cell viability curve was drawn, as Figure 4 shown.

[0074] The experimental results show that as the concentration of diniconazole increases, the cell viability shows a downward trend, which has an obvious negative correlation with the upward trend of the abnormal cell ratio obtained by image analysis. To further evaluate the reliability of this method, the present invention analyzed the correlation between the proportion of abnormal cell numbers and cell viability detection (CCK-8 method). By calculating the correlation coefficient (R 2 value) between the abnormal cell ratio and the CCK-8 cell viability detection data, the linear relationship between the two was evaluated. The results show that the fitting degree (R 2 ) of the proportion of abnormal morphological numbers of the three types of cells and the CCK-8 cell viability detection results all reached above 0.9, as Figure 5 shown.

[0075] The above results prove that the method proposed by the present invention is feasible. This method can not only obtain the viability state of cells instantaneously, but also visually observe the real-time state changes of cells, so as to judge the toxic effects of pollutants on cells. It is simple and fast to operate and has a low cost.

[0076] Based on this, this method does not need to rely on the annotation training of a large amount of data, is applicable to multiple cell types, and combines automated segmentation and classification tools to improve the detection efficiency. At the same time, this method has the characteristics of low cost, fast speed and high efficiency, and its reliability has been verified through experiments, providing a new technical path for pollutant toxicity detection.

[0077] To sum up, this method does not rely on a complex label annotation process, nor does it need to use expensive deep learning models. Instead, it uses two relatively lightweight deep learning tools, Cellpose and ilastic. Cellpose is used for cell segmentation, and ilastic is used for cell classification. Compared with traditional methods, these two have lower requirements for training data, can provide relatively accurate segmentation and classification results under limited data, and have the technical advantages of lower operation complexity, higher applicability and lower resource consumption. Especially Cellpose has a powerful adaptive segmentation ability, can handle cells with different morphologies and densities, and can also perform well with a small amount of data. Based on this, the embodiment of the present invention does not need to rely on the annotation training of a large amount of data, adopts simple image analysis means, and can efficiently and accurately evaluate cell viability and toxicity by calculating the ratio of the number of abnormal cells to the total number of cells, with the advantages of low cost, fast speed and high efficiency.

[0078] In addition, as Figure 6 shown, an embodiment of the present invention also discloses a cell viability detection device based on cell image analysis. The device includes:

[0079] An acquisition module 110, configured to acquire a target cell image to be detected;

[0080] A segmentation module 120, configured to use the Cellpose platform to perform batch automated segmentation on the target cell image to generate a segmentation mask image;

[0081] A classification module 130, configured to input the target cell image and the segmentation mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image;

[0082] A calculation module 140, configured to obtain a cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, where the total number of cells is the sum of the number of normal cells and the number of abnormal cells.

[0083] The cell viability detection device based on cell image analysis according to the embodiments of the present invention is used to execute the cell viability detection method based on cell image analysis in the above embodiments. The specific processing process is the same as that of the cell viability detection method based on cell image analysis in the above embodiments, and will not be repeated here one by one.

[0084] In addition, as Figure 7 shown, an embodiment of the present invention also discloses an electronic device, including: at least one processor 210; at least one memory 220 for storing at least one program; when the at least one program is executed by the at least one processor 210, the cell viability detection method based on cell image analysis in any of the previous embodiments is implemented.

[0085] In addition, an embodiment of the present invention also discloses a computer-readable storage medium, in which computer-executable instructions are stored, and the computer-executable instructions are used to execute the cell viability detection method based on cell image analysis in any of the previous embodiments.

[0086] The system architecture and application scenarios described in the embodiments of the present invention are for more clearly explaining the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those skilled in the art know that with the evolution of the system architecture and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present invention are equally applicable to similar technical problems.

[0087] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, and their appropriate combinations.

[0088] In a hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed by several physical components working together. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transitory medium). As is well known to those of ordinary skill in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer storage media includes but is not limited to RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those of ordinary skill in the art that communication media typically embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and can include any information delivery media.

[0089] The terms "component", "module", "system", etc. used in this specification are used to refer to computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution. For example, a component may be, but is not limited to, a process running on a processor, a processor, an object, an executable, an execution thread, a program, or a computer. By way of illustration, both an application running on a computing device and the computing device can be components. One or more components may reside in a process or execution thread, and components may be located on one computer or distributed between two or more computers. In addition, these components may execute from various computer-readable media having various data structures stored thereon. Components may communicate, for example, by signals according to one or more data packets (e.g., data from two components interacting with each other from a local system, a distributed system, or a network, e.g., the Internet interacting with other systems via signals) through local or remote processes.

Claims

1. A method for detecting cell viability based on cell image analysis, comprising: Obtaining a target cell image to be detected; Automatically segmenting the target cell image in batches using the Cellpose platform to generate a segmented mask image; Inputting the target cell image and the segmented mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image; Obtaining a cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, wherein the total number of cells is the sum of the number of normal cells and the number of abnormal cells.

2. The method according to claim 1, wherein The obtaining of the target cell image to be detected includes: Configuring the aperture parameter and focal length parameter of the microscope; Taking a picture of the target sample cells through the microscope to obtain the target cell image.

3. The method according to claim 1, wherein The target sample cells include red blood cells and adherent cells.

4. The method according to claim 1, wherein The automatically segmenting the target cell image in batches using the Cellpose platform to generate a segmented mask image includes: Identifying the cell type of the target sample cells in the target cell image; Determining segmentation parameters according to the cell type, where the segmentation parameters include the cell diameter range and the flow threshold; Inputting the target cell image into the cytoplasmic recognition model loaded on the Cellpose platform for semantic segmentation based on the segmentation parameters to generate the segmented mask image, wherein the cytoplasmic recognition model uses cyto2 pre-trained weights.

5. The method according to claim 1, wherein The inputting the target cell image and the segmented mask image into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell image includes: Configuring the classification feature parameters of the ilastic platform, where the classification feature parameters include standard object features, 3D convex hull features, and 2D skeleton features; Inputting the target cell image and the segmented mask image into the ilastic platform so that the ilastic platform classifies the target sample cells according to the standard object features, the 3D convex hull features, and the 2D skeleton features and outputs cell classification information; Determining the number of normal cells and abnormal cells in the target cell image according to the cell classification information.

6. The method according to claim 1, characterized in that, The obtaining of the cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells includes: Calculating the sum of the number of normal cells and the number of abnormal cells to obtain the total number of cells; Calculating the ratio of the number of abnormal cells to the total number of cells to obtain the cell viability detection result corresponding to the target cell image, where the cell viability detection result includes the degree of cell viability decline, and the degree of cell viability decline is used to characterize the damage condition of the cells.

7. The method according to claim 1, wherein The cell viability detection result includes: The higher the ratio, the greater the degree of cell viability decline; The lower the ratio, the smaller the degree of cell viability decline.

8. A cell viability detection device based on cell image analysis, characterized in that, The device includes: An acquisition module for acquiring a target cell image to be detected; A segmentation module for automatically segmenting the target cell images in batches using the Cellpose platform to generate segmented mask images; A classification module for inputting the target cell images and the segmented mask images into the ilastic platform for classification to determine the number of normal cells and abnormal cells in the target cell images; A calculation module for obtaining the cell viability detection result corresponding to the target cell image according to the ratio of the number of abnormal cells to the total number of cells, where the total number of cells is the sum of the number of normal cells and the number of abnormal cells.

9. An electronic device, comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, the cell viability detection method based on cell image analysis according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing computer-executable instructions for executing the cell viability detection method based on cell image analysis according to any one of claims 1 to 7.

Images