Image sequence dynamic feature analysis method for EMT process of liver cancer cells

Abstract

The present application relates to the technical field of image processing, in particular to a kind of image sequence dynamic feature analysis method of liver cancer cell EMT process.This method first compares the correlation between different features in the feature descriptor of liver cancer cell in the microscopic image of live cell in static control group and EMT index, determines the contribution of different features in the feature descriptor to EMT index, classifies liver cancer cell in each live cell microscopic image, and obtains multiple categories;Analysis of live cell microscopic image in dynamic observation group and liver cancer cell in the same position category in each live cell microscopic image in static control group, determine EMT index similarity;EMT index in live cell microscopic image in dynamic observation group is adjusted with EMT index similarity as weight, and the optimized EMT index is obtained.The present application improves the dynamic analysis capability of EMT index.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, specifically to a method for dynamic feature analysis of image sequences during the EMT process of liver cancer cells. Background Technology

[0002] Epithelial-mesenchymal transition (EMT) is a phenomenon first observed in embryonic development. During EMT, epithelial cells gradually lose their characteristic cell polarity and intercellular adhesion, acquiring the phenotypic characteristics of mesenchymal cells. In adulthood, EMT may occur after epidermal cell damage. In the development of liver cancer tumor cells, EMT also causes tumor cells to lose some epithelial cell characteristics and acquire some mesenchymal cell characteristics, thus giving the tumor cells stronger invasive and detachment capabilities.

[0003] Current precise methods for measuring EMT transformation markers in tumor cells typically rely on gene or related protein assays. These processes involve staining and drug testing, which can directly impair cell viability. Furthermore, traditional gene or related protein assays are static endpoint detections, dependent on protein or gene endpoint indicators. This results in a specific EMT transformation stage of the cell, making it impossible to obtain real-time, continuous, dynamic results when assessing the impact of other conditions on tumor cell EMT transformation. Summary of the Invention

[0004] To address the technical problem of obtaining continuous dynamic EMT index detection results, the present invention aims to provide a method for dynamic feature analysis of image sequences of the EMT process in liver cancer cells. The specific technical solution adopted is as follows:

[0005] In a first aspect, embodiments of the present invention provide a method for dynamic feature analysis of image sequences during the EMT process of liver cancer cells, the method comprising:

[0006] We obtained live-cell microscopic images of liver cancer cells and EMT indices from a dynamic observation group during liver tumor cell culture, and constructed a feature descriptor for each liver cancer cell.

[0007] The correlation between different features in the feature descriptors of liver cancer cells and EMT indices in live cell micrographs of the static control group was compared to determine the contribution of different features in the feature descriptors to EMT indices.

[0008] By combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to EMT indicators, liver cancer cells in each live cell microscopic image are classified into multiple categories.

[0009] The similarity of EMT indices between the live cell micrographs of the dynamic observation group and the live cell micrographs of the static control group was determined by analyzing the liver cancer cells belonging to the same category at the same location in each live cell micrograph of the dynamic observation group and the live cell micrographs of the static control group.

[0010] Using EMT index similarity as a weight, the EMT index in the live cell microscopic images of the dynamic observation group was adjusted to obtain the optimized EMT index.

[0011] Furthermore, the comparison of the correlation between different features in the feature descriptors of liver cancer cells and EMT indices in the live-cell micrographs of the static control group, and the determination of the contribution of different features in the feature descriptors to the EMT indices, includes:

[0012] Based on the similarity of EMT indices between any two live cell microscopic images in the static control group, the live cell microscopic images are matched to obtain the corresponding contribution image pairs to be analyzed in the static control group.

[0013] Using any feature in the feature descriptor as the target feature, the contribution of the target feature to the EMT index is determined based on the difference in the target features in the feature descriptors of liver cancer cells in two live-cell micrographs of all the contributing image pairs to be analyzed.

[0014] Further, the step of matching the live cell microscopic images based on the similarity of the EMT indices of any two live cell microscopic images in the static control group to obtain the corresponding contributing image pair to be analyzed in the static control group includes:

[0015] Multiple EMT indices from each live-cell micrograph are combined into an EMT index description vector.

[0016] For any two live cell microscopic images of the static control group, the negative correlation normalized value of the Euclidean distance between the EMT index description vectors is calculated as the EMT index similarity between the two live cell microscopic images.

[0017] Two live cell microscopic images of the static control group with EMT index similarity greater than a preset similarity threshold were used as the contributing image pairs to be analyzed.

[0018] Further, the step of determining the contribution of the target feature to the EMT index based on the difference in the target features in the feature descriptors of liver cancer cells in two live-cell micrographs of all contributing image pairs to be analyzed includes:

[0019] Use any pair of contributing images to be analyzed as the target pair of contributing images to be analyzed.

[0020] The average of the absolute differences between the target features in the feature descriptors of each pair of liver cancer cells in two live-cell micrographs of the target image pair to be analyzed is used as the contribution of the target features to the image pair.

[0021] The contribution of the target features to the EMT index is obtained by averaging the contribution of all image pairs of the target features in the static control group.

[0022] Furthermore, by combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to the EMT index, liver cancer cells in each live-cell micrograph are classified into multiple categories, including:

[0023] For any live cell microscopic image, the cell feature deviation between any two liver cancer cells is determined by combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to the EMT index.

[0024] Based on the cellular characteristic deviation between liver cancer cells, liver cancer cells in live cell microscopic images are classified into multiple categories.

[0025] Furthermore, for any live-cell microscopic image, the determination of the cell feature deviation between any two liver cancer cells, combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to the EMT index, includes:

[0026] Any feature in the feature descriptor is used as the target feature; any liver cancer cell in the live cell micrograph is used as the first liver cancer cell, and any liver cancer cell in the same live cell micrograph other than the first liver cancer cell is used as the second liver cancer cell.

[0027] The differences in target features corresponding to the first and second hepatocellular carcinoma cells are identified as cellular feature differences.

[0028] The product of the differences in cell characteristics between the first and second hepatocellular carcinoma cells and the contribution of the target features to the EMT index is used as the initial bias of the target features of the first and second hepatocellular carcinoma cells.

[0029] The arithmetic square root of the sum of the initial deviations of all characteristics of the first and second hepatocellular carcinoma cells is taken as the intermediate deviation.

[0030] By combining the intermediate deviation between the first and second hepatocellular carcinoma cells and the cumulative contribution of all features to the EMT index, the degree of cellular feature deviation between the first and second hepatocellular carcinoma cells was determined.

[0031] Furthermore, based on the cellular characteristic deviation between liver cancer cells, liver cancer cells in live-cell microscopic images are classified into multiple categories, including:

[0032] When the cellular feature deviation between two liver cancer cells is less than a preset deviation threshold, the two liver cancer cells are classified into the same category; by traversing any pair of liver cancer cells in the live cell micrograph, multiple categories are obtained.

[0033] Furthermore, the method for obtaining the similarity of EMT indices between the live cell microscopic images of the dynamic observation group and each live cell microscopic image of the static control group is as follows:

[0034] An arbitrary number of live cell microscopic images from the static control group were used as the live cell microscopic images to be analyzed.

[0035] Based on the proportion of liver cancer cells within each category, the categories in the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed are matched to obtain multiple category pairs;

[0036] Using any category pair as the target category pair, the similarity of the feature descriptors of liver cancer cells within the target category pair corresponding to the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed is analyzed to obtain the feature similarity value of the target category pair;

[0037] The difference in the number of liver cancer cells within the target category pair is obtained by analyzing the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed.

[0038] By combining the feature similarity value and cell number difference of the target category pair, a single index similarity of the target category pair is obtained;

[0039] By summing the single index similarity of all category pairs, the EMT index similarity between the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed is obtained, where the EMT index similarity is a normalized value.

[0040] Furthermore, based on the proportion of liver cancer cells within each category, the categories in the live-cell micrographs of the dynamic observation group and the live-cell micrographs to be analyzed are matched to obtain multiple category pairs, including:

[0041] Based on the proportion of liver cancer cells within each category, the categories corresponding to the live cell microscopic images of the dynamic observation group are arranged in reverse order to obtain the first category proportion sequence; based on the proportion of liver cancer cells within each category, the categories corresponding to the live cell microscopic images to be analyzed are arranged in reverse order to obtain the second category proportion sequence; when the number of elements in the first category proportion sequence and the second category proportion sequence are different, the sequence with fewer elements is padded with 0s, and the lengths of the first category proportion sequence and the second category proportion sequence after 0 padding are the same; the categories with the same index in the first category proportion sequence and the second category proportion sequence are regarded as a category pair.

[0042] Furthermore, the step of adjusting the EMT indices in the live cell microscopic images of the dynamic observation group using EMT index similarity as weight to obtain optimized EMT indices includes:

[0043] For any EMT index, the similarity of the EMT index of each live cell microscopic image in the static control group is used as the weight to perform a weighted summation of the EMT index of each live cell microscopic image in the static control group, so as to obtain the optimized EMT index of each live cell microscopic image in the dynamic observation group.

[0044] Secondly, a dynamic feature analysis system for image sequences of the EMT process in liver cancer cells is provided, the system comprising the following modules:

[0045] The initial data acquisition module is used to acquire liver cancer cells and EMT indices in live cell microscopic images of the dynamic observation group during the liver tumor cell culture process, and to construct a feature descriptor for each liver cancer cell.

[0046] The contribution analysis module is used to compare the correlation between different features in the feature descriptors of liver cancer cells and EMT indices in live cell micrographs of the static control group, and to determine the contribution of different features in the feature descriptors to the EMT indices.

[0047] The cell classification module is used to classify liver cancer cells in each live cell microscopic image by combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to the EMT index, resulting in multiple categories.

[0048] The data analysis module is used to analyze the liver cancer cells belonging to the same category at the same location in each live cell microscopic image of the dynamic observation group and the static control group, and to determine the similarity of the EMT index between each live cell microscopic image of the dynamic observation group and the static control group.

[0049] The optimization index module is used to adjust the EMT index in the live cell microscopic images of the dynamic observation group by using the similarity of EMT index as weight, so as to obtain the optimized EMT index.

[0050] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the various possible implementations of the first aspect.

[0051] Fourthly, embodiments of the present invention provide a computer program product comprising: computer program code, which, when executed on a computer, causes the computer to perform the method described in the first aspect or any possible implementation thereof.

[0052] Fifthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the various possible implementations of the first aspect.

[0053] The embodiments of the present invention have at least the following beneficial effects:

[0054] This invention addresses the lack of effective dynamic analysis capabilities for the EMT process in existing liver tumor live cells. Through real-time microscopic imaging, it compares the changes in microscopic cellular structure between static control groups and real-time dynamic observation groups using live cell microscopic images. This indirectly obtains optimized EMT indices in real-time from samples with known EMT processes, thus achieving dynamic analysis capabilities. This invention achieves real-time quantitative analysis of the EMT process in liver cancer cells through non-invasive, dynamic, and continuous time-series prediction. Attached Figure Description

[0055] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a flowchart of a method for dynamic feature analysis of image sequences of liver cancer cells during EMT, provided in one embodiment of the present invention.

[0057] Figure 2 This is a system block diagram of an image sequence dynamic feature analysis system for the EMT process of liver cancer cells, provided as an embodiment of the present invention. Detailed Implementation

[0058] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for dynamic feature analysis of image sequences of the EMT process in liver cancer cells proposed according to the present invention.

[0059] In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form.

[0060] In the description of the embodiments of the present invention, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of the present invention, "multiple" means two or more.

[0061] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0063] The embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided by the embodiments of the present invention are also applicable to similar technical problems.

[0064] This invention provides a specific implementation method for dynamic feature analysis of image sequences during the EMT process in liver cancer cells. This method is applicable to scenarios where EMT indicators of liver cancer cells are identified through image-stained morphological features. The method ultimately achieves optimized EMT indicators from label-free dynamic microscopic images as input, through feature extraction and weight analysis, realizing numerical results and dynamic curve fitting, and completing full-process digital monitoring. This method can be applied to drug screening and high-throughput EMT blocking efficacy evaluation, tumor cell drug resistance mechanism monitoring, and the construction of core modules for AI microscopic diagnostic platforms, providing a new technical path for anti-tumor drug development and intelligent digital pathology analysis. It should be noted that this invention is based on a dedicated feature set constructed around the biological mechanism of EMT, belonging to a method specific to a biomedical scenario.

[0065] The following describes in detail, with reference to the accompanying drawings, a specific scheme for the image sequence dynamic feature analysis method of the EMT process of liver cancer cells provided by the present invention.

[0066] Please see Figure 1 The diagram illustrates a flowchart of a method for dynamic feature analysis of image sequences during the EMT process of liver cancer cells according to an embodiment of the present invention. The method includes the following steps:

[0067] Step S100: Obtain live cell microscopic images of liver cancer cells in the target group during the liver tumor cell culture process, and construct a feature descriptor for each liver cancer cell.

[0068] Set up a live cell dynamic observation group and a static control group:

[0069] The static control group consisted of live-cell microscopic images of tissue sections at different stages, followed by immunohistochemical (IHC) staining and other methods to detect EMT markers and determine the overall EMT conversion index, referred to as the EMT index in subsequent steps. Each image contained EMT indicators, including epithelial and stromal markers. Epithelial markers included E-cadherin; stromal markers included N-cadherin, vimentin, and fibronectin.

[0070] The dynamic observation group uses live cell culture medium and suitable culture conditions to acquire microscopic images of liver cancer cells over several hours using an inverted microscope live cell imaging system within a preset time period. The image acquisition parameters are consistent with those of the control group. In this embodiment of the invention, the preset time period is set to 10 minutes.

[0071] The EMT process in hepatocytes refers to the gradual transformation of hepatocytes from an epithelial-like state to a mesenchymal-like state under specific pathological or physiological stimuli. Epithelial characteristics include cell polarity and intercellular connections, while mesenchymal characteristics include spindle-shaped structure and migration ability. During this process, cells undergo significant morphological changes, including differences in the morphological characteristics of individual cells and differences in the texture of groups of cells.

[0072] In normal hepatocytes, most epithelial cells resemble pebbles or paving stones with regular shapes; while cells during or after EMT exhibit spindle or star-shaped shapes. This results in individual hepatocytes in different states displaying different characteristic indicators, such as significant differences in cell aspect ratio, roundness, and eccentricity. Furthermore, in the texture of cell populations, epithelial cells tend to aggregate more dispersedly, while stromal cells show migration, with the migration direction consistent across the aggregates.

[0073] Because liver tumor cells exhibit significant morphological differences under different EMT states, cells with similar morphological characteristics are generally considered to have similar EMT indices. Considering that protein and gene detection methods kill cells but accurately capture EMT transformation characteristics, a static control group can be established. This control group obtains complete EMT indices using existing techniques, including pre-staining microscopic images. When the morphological similarity between the live cell microscopic images of the dynamic observation group and the static control group is high, the likelihood of similar EMT indices between the two groups is also high. Therefore, the live cell microscopic images of the current dynamic observation group can be used to predict the EMT indices corresponding to the live cell microscopic images of the static control group, ultimately yielding the EMT indices for the live cell microscopic images.

[0074] This invention first obtains the morphological features of live cell microscopic images of the dynamic observation group and the static control group, then analyzes the feature similarity between the live cell microscopic images of the dynamic observation group and the live cell microscopic images of the static control group, and finally evaluates the EMT index of the live cell microscopic images of the dynamic observation group.

[0075] Feature extraction was performed on live cell micrographs of the static control group and the static control group, respectively:

[0076] In live-cell imaging, cells are typically observed directly under a microscope in unstained conditions, a process known as label-free imaging. This imaging principle is based on the differences in light absorption across different parts of the cell. For example, denser regions, such as the nucleus, absorb more light, resulting in a darker, lower-grayscale image; while less dense regions exhibit more pronounced highlights. This significant difference in grayscale allows image segmentation techniques to be used to determine the boundaries of the cell membrane and nucleus.

[0077] Segmenting live-cell micrographs reveals that cells in different EMT states typically exhibit different morphological characteristics. For instance, cells in the epithelial state generally show better overall roundness, while those in the mesenchymal state show lower roundness. However, the EMT transformation process of liver tumor cells in culture medium is usually not completely synchronous; that is, different cells may have different EMT states at the same time. Therefore, it is necessary to obtain the individual morphological characteristics of the segmented cells.

[0078] Therefore, it is necessary to identify complete individual cells, i.e. liver cancer cells, in live cell microscopic images, then extract cell characteristics, and combine them with multi-cell analysis for further analysis.

[0079] For the live cell microscopic images of the dynamic observation group and the live cell microscopic images of the static control group taken before staining during the liver tumor cell culture process, the cell boundary is identified using existing technology to obtain the cell membrane boundary of each liver cancer cell and the corresponding cell nucleus location region, such as using the Cellpose model for segmentation.

[0080] Because the EMT process in liver cancer cells typically causes significant changes in cell morphology, such as mesenchymal cells becoming more elongated, the cell area-to-perimeter ratio can be used as one of the EMT features of cells. The number of pixels representing the perimeter of each liver cancer cell is extracted; the more elongated the cell, the larger its area-to-perimeter ratio. This ratio is determined by the ratio of the area of ​​the liver cancer cell to its perimeter. The area of ​​the liver cancer cell is represented by the number of pixels corresponding to it.

[0081] It should be noted that each liver cancer cell contains a cytoplasmic region and a nuclear region.

[0082] Obtain the roundness of the nucleus in each liver cancer cell, denoted as nuclear roundness; obtain the area ratio of the nucleus to the cytoplasm, denoted as nuclear-cytoplasmic ratio.

[0083] Each liver cancer cell still exhibits different texture features. For example, cells with a high degree of EMT transformation are identified by obtaining common texture features of each cytoplasm to obtain cytoplasmic regions in liver cancer cells. Then, statistical parameters, including contrast and grayscale information, are obtained through the grayscale co-occurrence matrix.

[0084] Because epithelial cells are typically densely distributed and structurally close, possessing numerous neighboring cells, the number of cells belonging to the cell region within each of the current cell's boundary pixels is calculated by extending the algorithm outwards in eight-neighbor directions. This number is considered the neighboring cell count; a larger count indicates a more compact cellular structure at the current cell's location. This process yields the number of neighboring cells for each cell.

[0085] In this embodiment of the invention, the cell perimeter ratio, nuclear roundness, nucleocytoplasmic ratio, contrast, mean grayscale value, and number of neighboring cells of each cell are arranged in a fixed order to form a feature combination. Thus, each liver cancer cell generates features with a fixed parameter order, referred to as a feature descriptor. Furthermore, the feature descriptor for each liver cancer cell is obtained.

[0086] Step S200: Compare the characteristic descriptors of liver cancer cells in live cell micrographs of the dynamic observation group and the static control group to determine the contribution of liver cancer cells to the EMT index.

[0087] In step S100, feature descriptors characterizing the morphology and texture of each liver cancer cell were extracted. Since the EMT index of the static control group describes a complete image, a similarity difference analysis is needed between the live-cell micrographs of the current dynamic observation group and the static control group. Generally, when the descriptors of multiple liver cancer cells in the live-cell micrographs of the dynamic observation group are similar to the feature descriptors of liver cancer cells in the live-cell micrographs of the static control group, it indicates that the two images are quite similar in morphology and other aspects. Therefore, using the EMT index of the static control group to reflect the EMT index of the target image has a higher accuracy.

[0088] However, since the cells in the culture medium are in a dynamic process of division and constant movement, the total number of cells varies at different times. The total number of cells in the live cell micrographs of the static control group is static data. Furthermore, because the static control group was collected at different times, the total number of cells in each live cell micrograph within the static control group is also inconsistent. Therefore, when performing differential analysis on the characteristic descriptors between liver cancer cells, it is necessary to exclude the influence of cell number on the calculation.

[0089] Typically, a live-cell microscopy image contains cells at different EMT (electromyeloma transition) stages. The proportion of liver cancer cells at each EMT stage directly reflects the EMT status of cells in each live-cell microscopy image. Therefore, by performing similarity analysis on the feature descriptors of liver cancer cells in each image and calculating the proportion of each type of liver cancer cell to the total number of cells, the two images are considered to have a high degree of similarity if their average feature descriptors for cell types are similar and their cell number proportions are also similar. It should be noted that the EMT stage includes both epithelial and stromal characteristic states.

[0090] When performing similarity analysis between liver cancer cells using feature descriptors, the contributions of different features in the feature descriptors to the final classification are usually inconsistent. Some features often play a key role in classification, while others, even if they differ greatly, have little impact on the final classification. In other words, the scale of similarity contribution of different features is inconsistent.

[0091] Since the liver cancer cells in the live-cell micrographs of the static control group also had feature descriptors obtained, and each feature descriptor can indirectly reflect the final EMT index, it can be regarded as a regression model, that is, the EMT index is reflected by multiple features in the feature descriptor. Therefore, if two live-cell micrographs in the static control group have similar EMT indices, and a certain feature in the feature descriptor has a large difference, then the feature representing the large difference will contribute relatively little to the final classification and recognition, and thus have a lower weight when using this feature for similarity analysis.

[0092] This step calculates the contribution of each feature to the EMT index by comparing the differences among the features in the characteristic descriptors of liver cancer cells in live-cell micrographs of the static control group.

[0093] First, based on the similarity of the EMT index of any two live cell microscopic images in the static control group, the live cell microscopic images are matched to obtain the corresponding contribution image pairs to be analyzed in the static control group.

[0094] For two live-cell micrographs with similar EMT indices, the greater the difference between a feature, the smaller the contribution of that feature to the final EMT index. Therefore, similarity classification is performed using known EMT indices from a static control group, and then difference analysis is conducted on each feature descriptor. The greater the difference, the lower the contribution of that feature to the EMT index.

[0095] Similarity analysis of EMT indices in the static control group was performed. First, multiple EMT indices were combined into an EMT index description vector. For any two live cell microscopic images of the static control group, the negative correlation normalized value of the Euclidean distance between the EMT index description vectors was calculated as the EMT index similarity between the two live cell microscopic images.

[0096] Two live-cell microscopic images of the static control group with an EMT index similarity greater than a preset similarity threshold are used as the contributing image pairs to be analyzed. In this embodiment of the invention, the preset similarity threshold is set to 0.95; in other embodiments, the implementer may adjust this value according to the actual situation.

[0097] We acquire all EMT index similarity pairs from the static control group that are greater than a preset similarity threshold and are to be analyzed. For each pair of images, we perform differential analysis on the features of all liver cancer cells. The greater the difference in a certain feature, the lower its contribution to the final classification.

[0098] Furthermore, using any feature in the feature descriptor as the target feature, the contribution of the target feature to the EMT index is determined based on the difference in the target features in the feature descriptors of liver cancer cells in two live-cell micrographs of all contributing image pairs to be analyzed.

[0099] Any pair of contributing images to be analyzed is taken as the target pair of contributing images to be analyzed; the average of the absolute values ​​of the differences in the feature descriptors of each pair of liver cancer cells in the two live-cell micrographs of the target pair of contributing images to be analyzed is calculated as the image pair contribution of the target feature; the image pair contribution of the target feature of all contributing image pairs to be analyzed in the static control group is averaged to obtain the contribution of the target feature to the EMT index.

[0100] In this embodiment of the invention, the contribution of the k-th feature in the feature descriptor to the EMT index is... The calculation formula is:

[0101] ;

[0102] Where norm is the normalization function; N is the number of categories of EMT indicators corresponding to the static control group; This represents the number of contributing image pairs to be analyzed in the static control group. The set of liver cancer cells in a live-cell micrograph of the I-th contributing image pair to be analyzed; The set of liver cancer cells in another live-cell micrograph in the I-th contributing image pair to be analyzed; Microscopic images of live cells in the I-th contributing image pair to be analyzed. The value of the k-th feature in the feature descriptor corresponding to the m-th liver cancer cell; Microscopic images of live cells in the I-th contributing image pair to be analyzed. The value of the k-th feature in the feature descriptor corresponding to the nth liver cancer cell.

[0103] In step S300, the liver cancer cells in each live cell microscopic image are classified by combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to the EMT index, resulting in multiple categories.

[0104] In the similarity analysis of liver cancer cells in the static control group and the dynamic observation group, the contribution of each feature to EMT similarity was fused to obtain the image feature matching degree of the fused EMT feature contribution of tumor cells in any live cell microscopic image.

[0105] For any live cell microscopic image, the cell feature deviation between any two liver cancer cells is determined by combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to the EMT index.

[0106] The specific method for obtaining cell characteristic deviation is as follows:

[0107] Any feature in the feature descriptor is used as the target feature; any liver cancer cell in the live cell micrograph is used as the first liver cancer cell, and any liver cancer cell in the same live cell micrograph other than the first liver cancer cell is used as the second liver cancer cell.

[0108] The differences in target features corresponding to the first and second hepatocellular carcinoma cells are identified as cellular feature differences.

[0109] The product of the differences in cell characteristics between the first and second hepatocellular carcinoma cells and the contribution of the target features to the EMT index is used as the initial bias of the target features of the first and second hepatocellular carcinoma cells.

[0110] The arithmetic square root of the sum of the initial deviations of all characteristics of the first and second hepatocellular carcinoma cells is taken as the intermediate deviation.

[0111] By combining the intermediate deviation between the first and second hepatocellular carcinoma cells and the cumulative contribution of all features to the EMT index, the degree of cellular feature deviation between the first and second hepatocellular carcinoma cells was determined.

[0112] In some embodiments, taking the i-th liver cancer cell in a live-cell micrograph as the first liver cancer cell and the j-th liver cancer cell in the live-cell micrograph as the second liver cancer cell, and using the k-th feature in the feature descriptor as the target feature, the cell feature deviation between the i-th and j-th liver cancer cells is defined. The calculation formula is:

[0113] ;

[0114] Where G is the sum of the contributions of all features in the feature descriptor to the EMT index, which is also the cumulative sum of the contributions of all features in the feature descriptor to the EMT index. The contribution of the k-th feature in the feature descriptor to the EMT index; The number of types of features in the feature descriptor; , These are the values ​​of the k-th feature in the feature descriptors corresponding to the i-th and j-th liver cancer cells, respectively. The difference in cellular characteristics between the i-th and j-th liver cancer cells corresponding to the k-th characteristic; The initial deviation between the i-th liver cancer cell and the j-th liver cancer cell corresponding to the k-th feature; This represents the intermediate deviation between the i-th and j-th liver cancer cells.

[0115] Based on the cellular feature deviation between liver cancer cells, liver cancer cells in live cell micrographs are classified into multiple categories. Specifically, when the cellular feature deviation between two liver cancer cells is less than a preset deviation threshold, the two liver cancer cells are classified into the same category. This process is repeated for any pair of liver cancer cells in the live cell micrograph to obtain multiple categories.

[0116] In this embodiment of the invention, the preset deviation threshold value is 0.05. In other embodiments, the implementer may adjust this value according to the actual situation.

[0117] Live cell micrographs of liver cancer cells in the static control group and dynamic observation group were classified, and multiple categories were obtained for each live cell micrograph.

[0118] Step S400: Analyze the liver cancer cells belonging to the same category at the same location in each live cell microscopic image of the dynamic observation group and the static control group, and determine the similarity of EMT index between each live cell microscopic image of the dynamic observation group and the static control group.

[0119] Obtain the percentage of liver cancer cells in each category of live-cell micrographs relative to the total number of liver cancer cells in the live-cell micrographs;

[0120] Using any live cell microscopic image from the static control group as the live cell microscopic image to be analyzed; based on the proportion of liver cancer cells within each category, the categories in the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed are matched to obtain multiple category pairs, specifically:

[0121] Based on the proportion of liver cancer cells within each category, the categories corresponding to the live cell microscopic images of the dynamic observation group are arranged in reverse order to obtain the first category proportion sequence. Similarly, based on the proportion of liver cancer cells within each category, the categories corresponding to the live cell microscopic images to be analyzed are arranged in reverse order to obtain the second category proportion sequence. When the number of elements in the first and second category proportion sequences differs, the sequence with fewer elements is padded with zeros. The lengths of the first and second category proportion sequences are the same after zero-padding. The element values ​​in both sequences represent the proportion of cells. Categories with the same index in both sequences are considered as a category pair.

[0122] More specifically, when the number of elements in the first category proportion sequence and the second category proportion sequence are different, the sequence with fewer elements is padded with zeros as follows: the number of categories in the two live-cell micrographs being compared is converted to the same number; for example, taking the live-cell micrograph with more categories before conversion as the benchmark, the proportion of each category is reversed; zeros are padded to the categories in the live-cell micrograph with more categories before conversion. For example, live-cell micrographs α1 correspond to 4 categories, and after reversing the order, the proportions of the categories are 0.5, 0.3, 0.1, and 0.1; live-cell micrographs α2 correspond to 3 categories, and after reversing the order, the proportions of the categories are 0.6, 0.2, and 0.2; since live-cell micrographs α2 have one fewer category than live-cell micrographs α1, zeros are padded to group β so that the two live-cell micrographs ultimately correspond to the same number of categories. After adding zeros, the α2 live cell micrographs correspond to 4 categories. After reversing the order, the percentages of each category are 0.6, 0.2, 0.2, and 0.

[0123] After obtaining multiple category pairs, for ease of understanding, any category pair is used as the target category pair. The similarity of the feature descriptors of liver cancer cells within the target category pair corresponding to the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed is analyzed to obtain the feature similarity value of the target category pair.

[0124] The difference in the number of liver cancer cells within the target category pair is obtained by analyzing the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed.

[0125] By combining the feature similarity value and cell number difference of the target category pair, a single index similarity of the target category pair is obtained;

[0126] By summing the single index similarity of all category pairs, the EMT index similarity between the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed is obtained, where the EMT index similarity is a normalized value.

[0127] In some embodiments, taking the i-th live cell microscopic image in the static control group as the live cell microscopic image to be analyzed, and the category pair formed by the q-th category as the target category pair, the corresponding EMT index similarity calculation formula is as follows:

[0128] ;

[0129] in, The similarity of the EMT index between the live cell microscopic images in the dynamic observation group and the i-th live cell microscopic image in the static control group is represented by norm(), which is the normalization function; sim() is the cosine similarity function. is the average vector of feature descriptors of liver cancer cells in the q-th category of live-cell microscopic images from the dynamic observation group; is the average vector of the feature descriptors of liver cancer cells in the qth category of the i-th live cell microscopic image in the static control group; This is the maximum of the number of categories corresponding to the live cell microscopic images in the current dynamic observation group and the number of categories corresponding to the i-th live cell microscopic image in the static control group; The number of liver cancer cells in the q-th category in the live-cell micrographs of the dynamic observation group; represents the number of liver cancer cells in the q-th category of the i-th live-cell microscopic image in the static control group; It is an absolute value function; The feature similarity value of the q-th category pair; is the difference in the number of cells for the q-th category pair.

[0130] This allows us to obtain the similarity of EMT indices between the live cell microscopic images of the dynamic observation group and any live cell microscopic image of the static control group.

[0131] Step S500: Using EMT index similarity as weight, adjust the EMT index in the live cell microscopic images of the dynamic observation group to obtain the optimized EMT index.

[0132] When the similarity of features between two images is high, the greater the similarity of the EMT index between the corresponding live cell microscopic images in the dynamic observation group and each live cell microscopic image in the static control group, the more likely the live cell microscopic images in the current dynamic observation group can be predicted by using the similarity of the EMT index between the live cell microscopic images in the dynamic observation group and each live cell microscopic image in the static control group as the weight of the index similarity.

[0133] For the live cell microscopic images to be analyzed, the similarity of the EMT index between the live cell microscopic images of the dynamic observation group and the live cell microscopic images to be analyzed is used as the weight.

[0134] For any EMT index, the similarity of the EMT index of each live cell microscopic image in the static control group is used as the weight to perform a weighted summation of the EMT index of each live cell microscopic image in the static control group, so as to obtain the optimized EMT index of each live cell microscopic image in the dynamic observation group.

[0135] The c-th optimized EMT index of live-cell micrographs in the adjusted dynamic observation group The calculation formula is:

[0136] ;

[0137] in, The number of images in the static control group whose EMT index similarity to the live cell microscopic images in the dynamic observation group is not 0; The similarity of the EMT index between the live cell microscopic images of the dynamic observation group and the i-th live cell microscopic image of the static control group; The c-th EMT index is the i-th live cell micrograph in the static control group.

[0138] The above steps yielded optimized EMT indices for live cell microscopic images at various stages of live cell dynamic imaging. In practical applications, it is necessary to dynamically analyze the changing trends of the optimized EMT indices. Therefore, it is necessary to fit the optimized EMT indices obtained at different stages to obtain dynamic change characteristics. During fitting, the least-squares multiple curve fitting method for a single optimized EMT indices can be used. The detailed steps are: (1) Obtain each optimized EMT indice corresponding to the live cell microscopic images at different sampling times in the dynamic observation group; (2) Perform least-squares curve fitting on each optimized EMT indices, i.e., different EMT indices correspond to different dynamic curves. Thus, the EMT indices conversion curves for evaluating liver cancer cells under corresponding culture conditions, i.e., dynamic EMT characteristics, are obtained.

[0139] Please see Figure 2 , Figure 2 This invention provides a system block diagram of an image sequence dynamic feature analysis system for the EMT process of liver cancer cells. The system includes the following modules:

[0140] The initial data acquisition module is used to acquire liver cancer cells and EMT indices in live cell microscopic images of the dynamic observation group during the liver tumor cell culture process, and to construct a feature descriptor for each liver cancer cell.

[0141] The contribution analysis module is used to compare the correlation between different features in the feature descriptors of liver cancer cells and EMT indices in live cell micrographs of the static control group, and to determine the contribution of different features in the feature descriptors to the EMT indices.

[0142] The cell classification module is used to classify liver cancer cells in each live cell microscopic image by combining the feature descriptors corresponding to liver cancer cells and the contribution of different features to the EMT index, resulting in multiple categories.

[0143] The data analysis module is used to analyze the liver cancer cells belonging to the same category at the same location in each live cell microscopic image of the dynamic observation group and the static control group, and to determine the similarity of the EMT index between each live cell microscopic image of the dynamic observation group and the static control group.

[0144] The optimization index module is used to adjust the EMT index in the live cell microscopic images of the dynamic observation group by using the similarity of EMT index as weight, so as to obtain the optimized EMT index.

[0145] Alternatively, the transmission medium may be a wired link, such as, but not limited to, coaxial cable, fiber optic cable and digital subscriber line, or a wireless link, such as, but not limited to, wireless Fidelity (WIFI), Bluetooth and mobile device networks.

[0146] It should be noted that the device provided in the above embodiments is only an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above.

[0147] This invention provides a schematic diagram of the structure of a computer device. Exemplarily, the computer device includes: a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the computer device can perform any of the aforementioned methods for dynamic feature analysis of image sequences during the EMT process of liver cancer cells.

[0148] Furthermore, embodiments of the present invention also protect an apparatus that may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to perform an image sequence dynamic feature analysis method for the EMT process of liver cancer cells provided in embodiments of the present invention.

[0149] In this embodiment of the invention, the device can be divided into functional modules according to the above method example. For example, each module can correspond to a separate function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and is only a logical functional division. In actual implementation, there may be other division methods.

[0150] When each module is divided according to its function, the device may also include a signal uploading module, a determination module, and an adjustment module. It should be noted that all relevant content of each step involved in the above method embodiments can be referenced from the functional descriptions of the corresponding functional modules, and will not be repeated here.

[0151] It should be understood that the apparatus provided in this embodiment of the invention is used to perform the above-described method for dynamic feature analysis of image sequences during the EMT process of liver cancer cells, and therefore can achieve the same effect as the above-described implementation method.

[0152] When using integrated units, the device may include a processing module and a storage module. When applied to a device, the processing module can be used to control and manage the device's operations. The storage module can be used to support the device in executing program code, etc. The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits as described in this disclosure. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of Digital Signal Processing (DSP) and a microprocessor, etc., and the storage module may be a memory.

[0153] In addition, the device provided in the embodiments of the present invention may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute the image sequence dynamic feature analysis method for the EMT process of liver cancer cells provided in the above embodiments.

[0154] This invention also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the aforementioned method steps to implement the image sequence dynamic feature analysis method for the EMT process of liver cancer cells provided in the above embodiments.

[0155] This invention also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to implement the image sequence dynamic feature analysis method for the EMT process of liver cancer cells provided in the above embodiments.

[0156] In this invention, the apparatus, computer-readable storage medium, computer program product, or chip provided in the embodiments are all used to execute the corresponding methods described above. Therefore, the beneficial effects they achieve can be referred to the beneficial effects in the corresponding methods described above, and will not be repeated here. Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In the embodiments provided by this invention, it should be understood that the disclosed apparatus and method can be implemented in other ways.

[0157] The device embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0158] It should also be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0159] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0160] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0161] The above content is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the protection scope of the present invention.

Claims

1. An image sequence dynamic feature analysis method of the EMT process of liver cancer cells, characterized by, The method comprises the following steps: Obtaining liver cancer cells and EMT indicators in live cell microscopic images of a dynamic observation group in a liver tumor cell culture process, and constructing a feature descriptor of each liver cancer cell; Comparing the correlation between different features in the feature descriptor of the liver cancer cells in the live cell microscopic images of the static control group and the EMT indicators to determine the contribution of the different features in the feature descriptor to the EMT indicators; Combining the feature descriptor of the liver cancer cells and the contribution of the different features to the EMT indicators, classifying the liver cancer cells in each live cell microscopic image, and obtaining multiple categories; Analyzing the liver cancer cells in the same category in the live cell microscopic images of the dynamic observation group and each live cell microscopic image in the static control group to determine the EMT indicator similarity between the live cell microscopic images of the dynamic observation group and each live cell microscopic image in the static control group, including: taking any live cell microscopic image in the static control group as a live cell microscopic image to be analyzed; matching the categories in the live cell microscopic images of the dynamic observation group and the live cell microscopic image to be analyzed based on the proportion of the number of liver cancer cells in the categories, obtaining multiple category pairs; taking any category pair as a target category pair, analyzing the similarity of the feature descriptors of the intra-category liver cancer cells in the target category pair of the live cell microscopic images of the dynamic observation group and the live cell microscopic image to be analyzed, obtaining a feature similarity value of the target category pair; analyzing the number difference of the intra-category liver cancer cells in the target category pair of the live cell microscopic images of the dynamic observation group and the live cell microscopic image to be analyzed, obtaining a cell number difference of the target category pair; combining the feature similarity value and the cell number difference of the target category pair of the category pair, obtaining a single indicator similarity of the target category pair; accumulating the single indicator similarities of all category pairs to obtain the EMT indicator similarity between the live cell microscopic images of the dynamic observation group and the live cell microscopic image to be analyzed, wherein the EMT indicator similarity is a normalized value; Taking the EMT indicator similarity as a weight, adjusting the EMT indicators in the live cell microscopic images of the dynamic observation group to obtain optimized EMT indicators; The cth optimized EMT indicator of the live cell microscopic image of the adjusted dynamic observation group The calculation formula is: ； wherein, is the number of images in the static control group whose EMT index similarity with the live cell microscopic image of the dynamic observation group is not 0; is the EMT index similarity of the live cell microscopic image of the dynamic observation group with the i-th live cell microscopic image in the static control group; is the c-th EMT index of the i-th live cell microscopic image in the static control group.

2. The image sequence dynamic feature analysis method of the EMT process of liver cancer cells according to claim 1, characterized in that, The comparison of the correlation between different features in the feature descriptor of the liver cancer cells in the live cell microscopic images of the static control group and the EMT indicators to determine the contribution of the different features in the feature descriptor to the EMT indicators comprises: According to the similarity of the EMT indicators of any two live cell microscopic images in the static control group, matching the live cell microscopic images to obtain a contribution image pair to be analyzed corresponding to the static control group; Taking any feature in the feature descriptor as a target feature, determining the contribution of the target feature to the EMT indicators according to the difference of the target feature in the feature descriptors of the liver cancer cells in the two live cell microscopic images of all contribution image pairs to be analyzed.

3. The method of claim 2, wherein the method is characterized by, The matching of the live cell microscopic images according to the similarity of the EMT indicators of any two live cell microscopic images in the static control group to obtain a contribution image pair to be analyzed corresponding to the static control group comprises: Combining multiple EMT indicators of each live cell microscopic image into an EMT indicator description vector; For any two live cell microscopic images of the static control group, calculate the negative correlation normalized value of the Euclidean distance between the EMT index description vectors as the EMT index similarity of the two live cell microscopic images; The two live cell microscopic images of the static control group with an EMT index similarity greater than a preset similarity threshold are used as the to-be-analyzed contribution image pair.

4. The image sequence dynamic feature analysis method of the EMT process of liver cancer cells according to claim 2, characterized in that, The contribution degree of the target feature to the EMT index is determined according to the difference of the target feature in the feature descriptor of the hepatocarcinoma cells in the two live cell microscopic images of all to-be-analyzed contribution image pairs, and the contribution degree of the target feature to the EMT index is determined. Any to-be-analyzed contribution image pair is taken as a target to-be-analyzed contribution image pair. The average value of the absolute value of the difference of the target feature in the feature descriptor of the hepatocarcinoma cells in the two live cell microscopic images of the target to-be-analyzed contribution image pair is taken as the image pair contribution degree of the target feature. The image pair contribution degrees of the target feature of all to-be-analyzed contribution image pairs in the static control group are averaged to obtain the contribution degree of the target feature to the EMT index.

5. The image sequence dynamic feature analysis method of the EMT process of liver cancer cells according to claim 1, characterized in that, The hepatocarcinoma cells in each live cell microscopic image are classified according to the feature descriptor corresponding to the hepatocarcinoma cells and the contribution degrees of different features to the EMT index to obtain multiple categories, including: For any one live cell microscopic image, the cell feature deviation degree between any two hepatocarcinoma cells is determined according to the feature descriptor corresponding to the hepatocarcinoma cells and the contribution degrees of different features to the EMT index. The hepatocarcinoma cells in the live cell microscopic image are classified according to the cell feature deviation degree between the hepatocarcinoma cells to obtain multiple categories.

6. The image sequence dynamic feature analysis method of the EMT process of liver cancer cells according to claim 5, characterized in that, For any one live cell microscopic image, the cell feature deviation degree between any two hepatocarcinoma cells is determined according to the feature descriptor corresponding to the hepatocarcinoma cells and the contribution degrees of different features to the EMT index, including: Any feature in the feature descriptor is taken as a target feature, and any hepatocarcinoma cell in the live cell microscopic image is taken as a first hepatocarcinoma cell, and any hepatocarcinoma cell other than the first hepatocarcinoma cell in the same live cell microscopic image is taken as a second hepatocarcinoma cell. The difference of the target feature corresponding to the first hepatocarcinoma cell and the second hepatocarcinoma cell is determined as the cell feature difference. The product value of the cell feature difference of the first hepatocarcinoma cell and the second hepatocarcinoma cell and the contribution degree of the target feature to the EMT index is taken as the initial deviation of the target feature of the first hepatocarcinoma cell and the second hepatocarcinoma cell. The arithmetic square root of the sum of the initial deviations of all features of the first hepatocarcinoma cell and the second hepatocarcinoma cell is taken as an intermediate deviation. The cell feature deviation degree between the first hepatocarcinoma cell and the second hepatocarcinoma cell is determined according to the intermediate deviation of the first hepatocarcinoma cell and the second hepatocarcinoma cell and the cumulative sum of the contribution degrees of all features to the EMT index.

7. The image sequence dynamic feature analysis method of the EMT process of liver cancer cells according to claim 5, characterized in that, The hepatocarcinoma cells in the live cell microscopic image are classified according to the cell feature deviation degree between the hepatocarcinoma cells to obtain multiple categories, including: When the cell feature deviation degree between two hepatocarcinoma cells is less than a preset deviation threshold, the two hepatocarcinoma cells are divided into the same category; any two hepatocarcinoma cells in the live cell microscopic image are traversed to obtain multiple categories. 8.The method of claim 1, wherein, The category in the dynamic observation group and the category in the to-be-analyzed live cell microscopic image are matched based on the proportion of the number of liver cancer cells in the category, and a plurality of category pairs are obtained, including: The categories corresponding to the live cell microscopic images of the dynamic observation group are arranged in descending order based on the proportion of the number of liver cancer cells in the category, and a first category proportion sequence is obtained; the categories corresponding to the to-be-analyzed live cell microscopic images are arranged in descending order based on the proportion of the number of liver cancer cells in the category, and a second category proportion sequence is obtained; wherein, when the number of elements in the first category proportion sequence and the second category proportion sequence is different, the sequence with fewer elements is supplemented with 0, and the length of the first category proportion sequence and the second category proportion sequence after the 0 supplementing processing is the same; the categories with the same serial number in the first category proportion sequence and the second category proportion sequence are taken as a category pair.

Images