A Cell Senescence Recognition Method and System Based on Image Processing

Abstract

This application provides a cell senescence recognition method and system based on image processing, applicable to the field of image processing technology. The method includes: sampling, enhancing, and extracting features from current and historical cell image information to obtain current and historical cell image feature information; obtaining target cell image feature classification parameters and target cell senescence recognition calculation parameters based on historical cell image feature information, initial cell image feature classification parameters, initial cell senescence recognition calculation parameters, and adjustments to the cell image feature classification parameters and cell senescence recognition calculation parameters; and obtaining the current cell senescence recognition image information based on the current cell image feature information, target cell image feature classification parameters, and target cell senescence recognition calculation parameters. This application accurately captures key features such as cell morphology and texture, improving the accuracy and reliability of cell senescence recognition.

Description

Technical Field

[0001] This application belongs to the field of image processing technology, and in particular relates to a method and system for recognizing cell senescence based on image processing. Background Technology

[0002] In the field of cell image analysis, with the continuous development of microscopic imaging technology and related research, the demand for precise processing and analysis of cell images is becoming increasingly urgent. From cell image acquisition to subsequent analysis processes, continuous innovation is underway to pursue more efficient and accurate results.

[0003] In existing technologies, classic machine learning classifiers, such as support vector machines, are typically used to classify cell images and determine cell state in order to achieve cell senescence recognition.

[0004] However, in existing technologies, support vector machines are difficult to adapt to complex and ever-changing cell image data, which greatly reduces the reliability and accuracy of cell senescence recognition. Summary of the Invention

[0005] In view of this, embodiments of this application provide a cell senescence recognition method and system based on image processing, aiming to solve the problem of inaccurate cell senescence recognition in the prior art.

[0006] The first aspect of this application provides a cell senescence recognition method based on image processing, including:

[0007] Acquire information from multiple current cell images and multiple historical cell images;

[0008] Based on multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors, and preset cell image vertical edge enhancement vectors, sampling and enhancement processing are performed on the multiple current cell image information and multiple historical cell image information to obtain multiple current cell image enhancement information and multiple historical cell image enhancement information.

[0009] Based on multiple preset cell image feature extraction vectors, feature extraction processing is performed on the multiple current cell image enhancement information and multiple historical cell image enhancement information to obtain multiple current cell image feature information and multiple historical cell image feature information;

[0010] Based on the aforementioned historical cell image feature information, randomly generated initial cell image feature classification parameter information, multiple randomly generated initial cell senescence recognition calculation parameter information, preset cell image feature classification parameter adjustment step size, and multiple preset cell senescence recognition calculation parameter adjustment step size, the target cell image feature classification parameter information and multiple target cell senescence recognition calculation parameter information are calculated.

[0011] Based on the multiple current cell image feature information, target cell image feature classification parameter information, and multiple target cell senescence recognition calculation parameter information, the current cell senescence recognition image information is calculated.

[0012] A second aspect of this application provides a cell senescence recognition system based on image processing, comprising:

[0013] The cell image information acquisition module is used to acquire multiple current cell image information and multiple historical cell image information;

[0014] The cell image enhancement information generation module is used to sample and enhance multiple current cell image information and multiple historical cell image information based on multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors and preset cell image vertical edge enhancement vectors, to obtain multiple current cell image enhancement information and multiple historical cell image enhancement information.

[0015] The cell image feature information generation module is used to perform feature extraction processing on the multiple current cell image enhancement information and the multiple historical cell image enhancement information according to multiple preset cell image feature extraction vectors, so as to obtain multiple current cell image feature information and multiple historical cell image feature information.

[0016] The target cell image feature classification parameter information and target cell senescence recognition calculation parameter information determination module is used to calculate the target cell image feature classification parameter information and multiple target cell senescence recognition calculation parameter information based on the multiple historical cell image feature information, randomly generated initial cell image feature classification parameter information, multiple randomly generated initial cell senescence recognition calculation parameter information, preset cell image feature classification parameter adjustment step size and multiple preset cell senescence recognition calculation parameter adjustment step size.

[0017] The current cell senescence recognition image information generation module is used to calculate the current cell senescence recognition image information based on the multiple current cell image feature information, target cell image feature classification parameter information, and multiple target cell senescence recognition calculation parameter information.

[0018] A third aspect of this application provides a terminal device, the terminal device including a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the computer program to implement the steps of the image processing-based cell senescence recognition method described in the first aspect above.

[0019] A fourth aspect of this application provides a computer-readable storage medium, comprising: storing a computer program, wherein when executed by a processor, the computer program implements the steps of the image processing-based cell senescence recognition method described in the first aspect above.

[0020] The beneficial effects of this application embodiment compared with the prior art are: this application effectively highlights the key structural details in cell images, accurately captures key features such as cell morphology and texture, and dynamically optimizes cell image feature classification parameter information and multiple cell aging recognition calculation parameter information, thereby improving the adaptability of cell aging recognition calculation, and thus significantly improving the reliability and accuracy of cell aging recognition. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram illustrating the implementation process of the cell senescence recognition method based on image processing provided in Embodiment 1 of this application;

[0023] Figure 2 This is a schematic diagram illustrating the implementation process of the image processing-based cell senescence recognition method provided in Embodiment 2 of this application;

[0024] Figure 3 This is a schematic diagram illustrating the implementation process of the image processing-based cell senescence recognition method provided in Embodiment 3 of this application;

[0025] Figure 4 This is a schematic diagram illustrating the implementation process of the image processing-based cell senescence recognition method provided in Embodiment 4 of this application;

[0026] Figure 5 This is a schematic diagram illustrating the implementation process of the image processing-based cell senescence recognition method provided in Embodiment 5 of this application;

[0027] Figure 6 This is a schematic diagram illustrating the implementation process of the image processing-based cell senescence recognition method provided in Embodiment Six of this application;

[0028] Figure 7 This is a schematic diagram illustrating the implementation process of the image processing-based cell senescence recognition method provided in Embodiment 7 of this application;

[0029] Figure 8 This is a schematic diagram of the structure of the image processing-based cell senescence recognition system provided in the embodiments of this application;

[0030] Figure 9 This is a schematic diagram of the terminal device provided in the embodiments of this application. Detailed Implementation

[0031] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0032] To illustrate the technical solution described in this application, specific embodiments are provided below.

[0033] Figure 1 The implementation flowchart of the cell senescence recognition method based on image processing provided in Embodiment 1 of this application is shown, and is described in detail below:

[0034] Step S101: Obtain multiple current cell image information and multiple historical cell image information.

[0035] In this embodiment, current cell image information refers to the various data information contained in multiple cell images acquired by imaging devices at the current time point, covering visual features such as cell morphology, texture, and edge contours. Multiple images containing cell details can be obtained by real-time or recent imaging of the target cell sample using microscopic imaging devices such as fluorescence microscopes and scanning electron microscopes. After preliminary format conversion and storage processing, these image data form multiple current cell image information sets available for subsequent analysis. Historical cell image information refers to the various data information contained in multiple cell images acquired by the same or similar imaging devices at different historical periods prior to the current time point. This also includes visual features such as cell morphology, texture, and edge contours at different historical stages. Image data acquired and stored for the same or similar cell samples at different past time points can be retrieved from image databases in laboratories or research institutions. After retrieval, extraction, and organization, these historical image data form multiple historical cell image information sets available for comparative analysis.

[0036] Step S102: Based on multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors, and preset cell image vertical edge enhancement vectors, the multiple current cell image information and multiple historical cell image information are sampled and enhanced to obtain multiple current cell image enhancement information and multiple historical cell image enhancement information.

[0037] In this embodiment, the multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors, and preset cell image vertical edge enhancement vectors can all be manually set. The preset cell image sampling vectors can be constructed by analyzing the pixel distribution patterns of a large number of cell images. This can be achieved by first collecting cell image samples of different types and states, statistically analyzing the differences in pixel value ranges between cell regions and background regions, and determining the typical pixel intervals of the cell core structure. For example, in fluorescent cell images, cell region pixel values ​​are mostly concentrated between 80-200, while background pixel values ​​are mostly between 0-50. Within this typical pixel interval, pixel coordinates that cover the main cell structure are selected, and the weights corresponding to these coordinates are set to 1, while the weights of the background region coordinates are set to 0, forming the basic sampling vector. Then, by adjusting the weight distribution in the vector, the vector can be optimized. When scanning images, pixels in densely populated cell regions are prioritized for preservation, while redundant regions with a background ratio exceeding 70% are removed. This results in the generation of multiple preset cell image sampling vectors, each adapted to cell image sampling requirements at different magnification levels. The preset cell image horizontal edge enhancement vectors can be designed specifically for the horizontal edge features of cells. This can be achieved by first selecting cell images containing clear horizontal edges, such as images with well-defined horizontal outlines of cell nuclei, and extracting the gradient change patterns of edge pixels. The calculation shows that the horizontal grayscale difference between adjacent pixels at the horizontal edge is mostly between 30-50, thus constructing a 3×3 vector structure, where the left and right images of the middle horizontal row are... The pixel positions are assigned enhancement weights, such as a weight of -1 for the left column of the middle row, 0 for the middle column, and 1 for the right column. The vertical upward and downward pixel positions are set to a suppression weight of 0. This ensures that when the vector is applied to the image, it enhances areas with significant horizontal gray-level differences. By testing the effect of different weight combinations on horizontal edge enhancement, it was found that when the horizontal weight difference is 2, edge sharpness is improved by more than 40%. This weight distribution was ultimately determined as the preset horizontal edge enhancement vector for cell images. The preset vertical edge enhancement vector for cell images can be designed based on the vertical edge features of cells, such as selecting sample images with prominent vertical cell edges, like the vertical distribution of cell protrusions. The image was analyzed to determine the grayscale changes of pixels at the vertical edges. The analysis showed that the grayscale difference between adjacent pixels in the vertical direction is usually between 25 and 45. Therefore, a 3×3 vector structure can be designed. Enhancement weights are set for the upper and lower pixels of the middle column in the vertical direction, such as -1 for the upper row of the middle column, 0 for the middle row, and 1 for the lower row. The suppression weights are set to 0 for the left and right columns of pixels in the horizontal direction to ensure that the vector prioritizes the enhancement of areas with obvious grayscale changes in the vertical direction. When the vertical weight difference is 2, the recognition of the vertical edge is improved by more than 35%. This weight distribution is determined as the preset vertical edge enhancement vector for cell images to adapt to the vertical edge enhancement needs of different types of cells.The process can begin by calling multiple preset cell image sampling vectors to perform sliding convolution calculations on the current cell image information and historical cell image information to achieve sampling processing. Then, a preset cell image horizontal edge enhancement vector is introduced and multiplied with the vector after convolution calculation to perform horizontal edge enhancement on the current cell image information and historical cell image information after sampling processing. By strengthening the pixel grayscale differences in the horizontal direction in the image, the edge contours of cells in the horizontal direction are highlighted, such as the horizontal boundary of the cell nucleus and the horizontal texture of the cell membrane. Then, a preset cell image vertical edge enhancement vector is introduced and multiplied with the vector after convolution calculation to perform vertical edge enhancement on the current cell image information and historical cell image information after sampling processing. The vertical edge enhancement vector enhances the pixel grayscale changes in the vertical direction, clearly presenting the edge structure of cells in the vertical direction, such as the vertical edges of cell protrusions and the vertical distribution contours of organelles. Thus, after sampling processing and the horizontal and vertical edge enhancement vectors respectively enhance the edge details in different directions, multiple current cell image enhancement information and multiple historical cell image enhancement information with clear edges and prominent key structures are finally obtained.

[0038] Step S103: Based on multiple preset cell image feature extraction vectors, perform feature extraction processing on the multiple current cell image enhancement information and the multiple historical cell image enhancement information to obtain multiple current cell image feature information and multiple historical cell image feature information.

[0039] In this embodiment, the preset cell image feature extraction vector can be pre-defined. It can be constructed by representing the brightness plane and red, green, and blue color planes of the color cell image as quaternion matrices through two-dimensional quaternion singular spectrum analysis, and constructing corresponding feature extraction vectors to associate the pixel spatial relationships of different color planes. For example, a 4×4 quaternion matrix window can be used to slide through the image, and the correlation features between color planes can be extracted through trajectory matrix decomposition. These correlations are then transformed into the weight distribution of the feature extraction vector, so that the vector can focus on the areas with obvious color changes in the cell image, such as the color difference boundary between the cell nucleus and cytoplasm. Alternatively, the cell image signal can be decomposed by combining empirical mode decomposition to obtain multiple intrinsic mode function components. Components containing cell edges and texture details can be selected from these components, and the feature distribution of these components can be transformed into the dimension parameters of the feature extraction vector. For example, the intrinsic mode function components in the first 6 energy concentrations can be selected through binary programming. These components correspond to the edge contours and texture changes of the cell. The feature extraction vector constructed based on this can effectively capture the detailed features of the cell, such as nuclear membrane folds and cell membrane edges. This can involve calling multiple preset cell image feature extraction vectors, each corresponding to different types of features in the cell image, such as cell morphology, texture, and edge features. Each preset cell image feature extraction vector is then multiplied by both current and historical cell image enhancement information. Information matching the feature type targeted by the vector is then filtered and extracted from these enhancements. For example, an extraction vector for cell morphology features will extract cell size, shape, and contour information; an extraction vector for cell texture features will extract cell surface features. The textural distribution and density of the surface are analyzed, and then the feature information extracted by each feature extraction vector is integrated. The features obtained from the same current cell image enhancement information through different feature extraction vectors are summarized to form one of the multiple current cell image feature information corresponding to the image. Similarly, the features obtained from the same historical cell image enhancement information through different feature extraction vectors are integrated to form one of the multiple historical cell image feature information corresponding to the image. After performing this feature extraction process on all current cell image enhancement information and historical cell image enhancement information, multiple current cell image feature information and multiple historical cell image feature information are obtained.

[0040] Step S104: Based on the multiple historical cell image feature information, randomly generated initial cell image feature classification parameter information, multiple randomly generated initial cell senescence recognition calculation parameter information, preset cell image feature classification parameter adjustment step size, and multiple preset cell senescence recognition calculation parameter adjustment step size, the target cell image feature classification parameter information and multiple target cell senescence recognition calculation parameter information are calculated.

[0041] In this embodiment, the initial cell image feature classification parameter information can be a set of basic parameters generated randomly before cell image feature classification, used to initially divide cell image feature categories, such as the number of young cell feature categories, the number of mid-aged cell feature categories, and the number of deeply aged cell feature categories. The initial cell aging recognition calculation parameter information can refer to a set of multiple basic parameters generated randomly before cell aging recognition calculation, used to quantify the degree of cell aging. These parameters affect the aging recognition model's weight allocation and calculation logic for features, such as feature importance weights, aging index conversion coefficients, and thresholds for different aging stages. Based on the texture features, edge features, morphological features, etc., involved in the cell aging recognition task and the calculation requirements, multiple sets of initial parameters can be generated using a random numerical generation method within a preset parameter value range, such as weight coefficient ranges and threshold intervals. Each set of parameters corresponds to a potential aging recognition calculation logic, ensuring that subsequent parameters that fit the real aging pattern can be screened and optimized based on historical cell image feature information and a preset adjustment step size, providing an initial calculation basis for accurate cell aging recognition. The preset step size for adjusting cell image feature classification parameters can be manually preset. It can be a pre-set step size value used to gradually adjust feature classification parameters during iterative optimization. It can be constructed based on historical cell image feature classification experience data and parameter optimization requirements. For example, it can be constructed by first statistically analyzing the distribution range of different categories of features in historical cell image feature data, such as the difference range between young cells and mid-term senescent cells, and the difference range between mid-term senescent cells and deep senescent cells. The mean or median of these ranges is taken as 1 / 10 to 1 / 20 as the initial adjustment step size reference. If the optimal adjustment range of feature classification parameters in historical data is between 0.01 and 0.1, and multiple experiments show that too large a step size will cause parameter oscillation, while too small a step size will result in low iteration efficiency, then the preset step size for adjusting cell image feature classification parameters is fixed at 0.05. This ensures that the parameter adjustment in each iteration effectively approaches the optimal solution, while avoiding skipping the optimal value due to excessive adjustment range, thus adapting to the gradual optimization process of feature classification parameters from the initial value to the target value.The preset adjustment step size for cell senescence recognition calculation parameters can be manually set, or it can be an adjustment range set separately for different senescence recognition calculation parameters. It can be constructed by combining the sensitivity of different parameters to senescence recognition results and historical optimization experience. For example, for weight parameters that affect feature importance allocation, since they have a significant impact on senescence recognition results, a smaller adjustment step size needs to be set to ensure optimization accuracy. For parameters such as the senescence stage determination threshold, which have a relatively mild impact on the results, a larger adjustment step size can be set to improve iteration efficiency. If the optimal adjustment range of the feature weight parameter is between 0.001 and 0.01, and the convergence effect is best when the step size is 0.005, then the adjustment step size of this parameter is preset to 0.005. For the senescence index threshold parameter, historical data shows that it can effectively distinguish senescence stages when adjusted within the range of 0.1 to 0.5, so its adjustment step size is preset to 0.05. In this way, by setting differentiated adjustment step sizes for different parameters, key parameters can be accurately optimized, and the overall parameter optimization efficiency can be improved, ensuring that multiple cell senescence recognition calculation parameters gradually converge to the target value from the initial value. This can be based on multiple historical cell image feature information as foundational data, combined with randomly generated initial cell image feature classification parameters and multiple randomly generated initial cell senescence recognition calculation parameters. A parameter optimization calculation process is initiated, adjusting the step size according to preset cell image feature classification parameters to iteratively adjust the initial cell image feature classification parameters. In each adjustment process, the adjusted parameters are applied to the classification processing of multiple historical cell image feature information. By comparing the matching degree between the classification results and the actual senescence state of historical cells, the applicability of the current parameters is evaluated, and the direction of parameter adjustment to improve classification accuracy is retained. Simultaneously, according to multiple preset cell senescence recognition... Instead of calculating the step size of parameter adjustment, the algorithm optimizes multiple initial cell senescence recognition calculation parameters one by one. Each adjusted parameter is applied to the senescence recognition calculation of historical cell image feature information. By analyzing the deviation between the calculation results and the actual senescence situation, parameter update values ​​that can reduce recognition errors are selected. The algorithm continues to iterate and optimize until the parameters are stable. Finally, after multiple rounds of parameter adjustment and verification based on historical cell image feature information, the iteration stops when the accuracy of cell image feature classification results and the error of cell senescence recognition calculation results both reach the preset standard. The parameters obtained at this time are the target cell image feature classification parameters and multiple target cell senescence recognition calculation parameters.

[0042] Step S105: Calculate the current cell aging recognition image information based on the multiple current cell image feature information, target cell image feature classification parameter information, and multiple target cell aging recognition calculation parameter information.

[0043] In this embodiment, the target cell image feature classification parameter information is a classification standard optimized based on historical cell image feature information. Multiple current cell image feature information are classified, and each feature information is assigned to a corresponding cell aging feature category, such as young cell feature category, mid-aged cell feature category, or deep-aged cell feature category. Then, multiple target cell aging recognition calculation parameters are introduced. These parameters include feature importance weights, aging index conversion coefficients, etc., which are key bases for quantifying the degree of aging after optimization. Combining the classified current cell image feature information, according to the rules in the target cell aging recognition calculation parameters, corresponding weights are assigned to various features and comprehensive calculations are performed. The feature information is transformed into a quantifiable aging index or indicator to reflect the degree of cell aging. Finally, based on the classification results and the quantified aging index, current cell aging recognition image information that can intuitively display the cell aging state is generated. The current cell aging recognition image information can clearly mark the aging stage of the cell, such as distinguishing young, mid-aged, and deep-aged cell regions by different colors or symbols, to fully present the current cell aging recognition results.

[0044] In this embodiment, multiple current cell image features can be classified according to target cell image feature classification parameters to obtain the aging index or classification label of the current cell. Then, the boundary range of all cell regions in the image is determined to ensure that each cell region is accurately defined, providing a clear spatial range for subsequent labeling. For example, in fluorescent cell images, enhanced image clarity allows for precise identification of the outline of each cell nucleus and the overall cell boundary. Furthermore, based on the feature threshold ranges for young, mid-aged, and deeply aged cells defined in the target cell image feature classification parameters, the aging index of each cell region is compared with these thresholds. For example, if the parameter information sets an aging index below 0.3 for young cells, 0.3 to 0.7 for mid-aged cells, and above 0.7 for deeply aged cells, then the aging index of each cell region is checked one by one to determine its aging stage. Subsequently, colors or labels can be assigned to cell regions at different aging stages according to preset visual labeling rules. For example, green labels are assigned to young cell regions, yellow labels to mid-aged cell regions, and red labels to deeply aged cell regions. In specific operations, this can be done... By traversing every pixel of the image and combining cell region boundary information, pixels belonging to young cell regions are filled with green, intermediate senescent regions with yellow, and deep senescent regions with red. Taking an image containing 100 cells as an example, after calculation, if 30 cells are identified as young cells, their regions are marked with green; if 50 are intermediate senescent cells, their regions are marked with yellow; and if 20 are deep senescent cells, their regions are marked with red. Finally, the marked cell images are integrated into the current cell senescence recognition image information. Through the intuitive differences in color or marking, the senescence stage of each cell region is fully presented to preserve the original structure and enhanced details of the cells, ensuring that the marking does not obscure the key morphological information of the cells.

[0045] The cell senescence recognition method based on image processing provided in this application effectively highlights key structural details in cell images, accurately captures key features such as cell morphology and texture, and dynamically optimizes cell image feature classification parameters and multiple cell senescence recognition calculation parameters, thereby improving the adaptability of cell senescence recognition calculation and significantly enhancing the reliability and accuracy of cell senescence recognition.

[0046] Figure 2 The flowchart illustrating the implementation of the image processing-based cell senescence recognition method provided in Embodiment 2 of this application is shown. The difference between this method and Embodiment 1 described above is that:

[0047] Multiple preset cell image sampling vectors include a preset first-scale cell image sampling vector and a preset second-scale cell image sampling vector;

[0048] Step S102 specifically includes:

[0049] Step S201: Based on the preset first-scale cell image sampling vector, the multiple current cell image information and multiple historical cell image information are sampled and processed to obtain multiple first-scale current cell image sampling information and multiple first-scale historical cell image sampling information.

[0050] In this embodiment, the preset first-scale cell image sampling vector can be pre-set manually. It can be achieved by first collecting a large number of cell image samples under low magnification, statistically analyzing the pixel distribution patterns of cell regions and background regions in the image, and finding that the effective global cell structure, such as cell cluster distribution and overall morphology, is mainly concentrated in the middle 60% of the image, while the edge 40% is mostly redundant background. Therefore, the window size of the sampling vector is set to 32×32 pixels, and the weight distribution within the window is as follows: the pixel weight in the image center region is set to 1, and the pixel weight in the edge region is set to 0.3, in order to prioritize the preservation of global structural information. By testing the sampling effect of different window sizes, the sampling efficiency is the highest when the window is 32×32 pixels, with a global structure retention rate of 92% and a background redundancy removal rate of 75%. Finally, the 32×32 pixel size, the pixel weight in the image center region set to 1, and the pixel weight in the edge region set to 0.3 are determined as the preset first-scale cell image sampling vector.

[0051] In this embodiment, sliding sampling can be performed using a preset first-scale cell image sampling vector and multiple current cell image information and multiple historical cell image information. Priority is given to preserving local detail areas with high pixel density in the image. For example, in fluorescent cell images, cell nucleus texture information with a diameter of 5-10 micrometers can be accurately extracted, ultimately resulting in multiple first-scale current cell image sampling information and multiple first-scale historical cell image sampling information that focus on microscopic details.

[0052] Step S202: Based on the preset second-scale cell image sampling vector, the multiple current cell image information and multiple historical cell image information are sampled and processed to obtain multiple second-scale current cell image sampling information and multiple second-scale historical cell image sampling information.

[0053] In this embodiment, the preset second-scale cell image sampling vector can be manually set and can be used to extract local detail features of cells. Cell image samples under high magnification can be selected to analyze the pixel distribution of local cell details such as nuclear membrane folds and cell membrane texture. Detail features are mainly concentrated in local areas where pixel grayscale changes drastically, such as areas where the grayscale difference between adjacent pixels is greater than 30. Therefore, the window size of the sampling vector can be set to 16×16 pixels, and the weight distribution within the window is as follows: the pixel weight of the detail area with drastic grayscale changes is set to 1.2, and the pixel weight of the smooth area is set to 0.5 to enhance the sampling accuracy of local details. Therefore, when the window is 16×16 pixels and the detail area weight is 1.2, it is determined as the preset second-scale cell image sampling vector.

[0054] In this embodiment, sliding sampling can be performed using a preset second-scale cell image sampling vector and multiple current cell image information and multiple historical cell image information. The focus is on preserving the overall morphology and spatial distribution information of the cell, extracting the overall outline of the cell and the positional relationship of adjacent cells, and finally obtaining multiple second-scale current cell image sampling information and multiple second-scale historical cell image sampling information that focus on the macroscopic morphology.

[0055] Step S203: The multiple first-scale current cell image sampling information and the multiple second-scale current cell image sampling information are stitched together to generate multiple current cell image sampling information.

[0056] In this embodiment, the stitching process can be achieved by aligning the spatial coordinates of the images. This can be achieved by superimposing and fusing the microscopic detail features extracted from the current cell image sampling information at the first scale with the macroscopic morphological features from the current cell image sampling information at the second scale according to pixel position. For example, the 200×200 pixel cell nucleus texture detail sampling information and the 500×500 pixel cell outline sampling information can be stitched together at the same coordinate position, so that the stitched current cell image sampling information contains both the cell nucleus texture with a diameter of 5-10 micrometers and the overall cell morphology of 20-50 micrometers, forming multiple current cell image sampling information that have both detailed and global information.

[0057] Step S204: The multiple first-scale historical cell image sampling information and the multiple second-scale historical cell image sampling information are stitched together to generate multiple historical cell image sampling information.

[0058] In this embodiment, the microscopic details in the first-scale historical cell image sampling information can be fused with the macroscopic morphology in the second-scale historical cell image sampling information. For example, the 5-10 micrometer texture details and 20-50 micrometer contour information of 30 young cells in the historical fluorescent cell image can be stitched together so that each historical cell image sampling information contains features at different scales at the same time.

[0059] Step S205: Based on the preset horizontal edge enhancement vector and the preset vertical edge enhancement vector of the cell image, the multiple current cell image sampling information and the multiple historical cell image sampling information are enhanced to obtain multiple current cell image edge enhancement information and multiple historical cell image edge enhancement information.

[0060] In this embodiment, the preset horizontal edge enhancement vector for cell images can be constructed based on the gray-level variation pattern of horizontal edges in cell images. First, cell sample images containing obvious horizontal edges can be selected, such as fluorescent cell images with clear horizontal outlines of cell nuclei or horizontal texture images of cells under a scanning electron microscope. The gray-level difference range of adjacent pixels at the horizontal edges in these images is statistically analyzed. Analysis reveals that the gray-level difference of pixels at the horizontal edges is mostly concentrated between 30 and 50 in the horizontal direction. Therefore, a 3×3 vector structure is artificially designed to focus on enhancing the gray-level difference in the horizontal direction: a suppression weight is set in the left column of the middle row of the vector, a neutral weight is set in the middle column, and an enhancement weight is set in the right column. Suppression weights are set in the vertical up and down rows of pixels to reduce interference. By testing the effect of different weight combinations on the horizontal edge enhancement effect, when the horizontal weight difference is 2 (e.g., left column weight -1, middle column weight 0, right column weight 1), the clarity of the horizontal edges is improved by more than 40%, and no obvious artifacts are produced. Therefore, this weight distribution is determined as the preset horizontal edge enhancement vector for cell images.

[0061] The preset vertical edge enhancement vector for cell images can be manually set. This can be achieved by first selecting sample images with prominent vertical cell edges, such as fluorescence images showing vertically distributed cell protrusions or scanning electron microscope images of vertical textures on material surfaces. By analyzing the grayscale changes of pixels at the vertical edges, it is determined that the grayscale difference between adjacent pixels in the vertical direction is typically between 25 and 45. Therefore, a 3×3 vector structure can be manually designed to emphasize the grayscale difference in the vertical direction: suppression weights are set at the top of the middle column, neutral weights at the middle row, and enhancement weights at the bottom. Suppression weights are also set at the left and right columns of pixels in the horizontal direction to avoid interference from horizontal features. Experiments have shown that when the vertical weight difference is 2 (e.g., top row weight -1, middle row weight 0, bottom row weight 1), the recognition of the vertical edge is improved by more than 35%, clearly presenting the vertical contour of cell protrusions. Therefore, 3×3, top row weight -1, middle row weight 0, and bottom row weight 1 can be determined as the preset vertical edge enhancement vector for cell images. First, a preset horizontal edge enhancement vector for cell images can be applied to current and historical cell image sampling information. This can be achieved by multiplying the preset horizontal edge enhancement vector with multiple current and historical cell image sampling information to enhance the horizontal pixel grayscale differences. For example, this could increase the grayscale difference of the horizontal boundary of the cell nucleus from 30 to 50, improving clarity by 40%. Then, a preset vertical edge enhancement vector for cell images can be used to enhance vertical features. This can be achieved by multiplying the preset vertical edge enhancement vector with multiple current and historical cell image sampling information, increasing the grayscale difference of the vertical edge of cell protrusions from 25 to 45, improving recognition by 35%. After enhancement, both horizontal and vertical edge details in the current and historical cell image edge enhancement information are highlighted, such as the horizontal texture of the nuclear membrane and the vertical protrusions of the cell membrane, which are clearly distinguishable.

[0062] Step S206: Based on the multiple current cell image edge enhancement information, multiple historical cell image edge enhancement information, multiple current cell image information, and multiple historical cell image information, obtain multiple current cell image enhancement information and multiple historical cell image enhancement information.

[0063] In this embodiment, the prominent edge details in the current cell image edge enhancement information are superimposed with the basic pixel values ​​in the original current cell image information. For example, the pixel values ​​in the edge region are weighted by 10-20%, so that the enhanced image retains the integrity of the original structure while strengthening key edge features. The historical cell image enhancement information is fused using the same method. Taking an image containing 100 cells as an example, the edge clarity of the green area of ​​young cells is improved by 40% after enhancement, the texture recognition of the yellow area of ​​mid-aged cells is improved by 35%, and the contour integrity of the red area of ​​deeply aged cells is improved by 50%. Finally, multiple current cell image enhancement information and multiple historical cell image enhancement information with sharp edges and rich details are obtained.

[0064] The cell senescence recognition method based on image processing provided in this application integrates microscopic details and macroscopic morphological information through multi-scale sampling, and strengthens key structural features by combining edge enhancement technology. This provides more comprehensive cell image data support for subsequent feature extraction and dynamic parameter optimization, thereby comprehensively improving the accuracy and reliability of cell senescence recognition.

[0065] Figure 3 The flowchart illustrating the implementation of the image processing-based cell senescence recognition method provided in Embodiment 3 of this application is shown. The difference between this method and Embodiment 1 is that step S103 specifically includes:

[0066] Step S301: Pixel extraction is performed on the multiple current cell image enhancement information and the multiple historical cell image enhancement information to obtain multiple current cell image pixel information and multiple historical cell image pixel information.

[0067] In this embodiment, pixel extraction can be achieved by traversing each spatial coordinate of the image. For the enhanced current and historical cell images, the grayscale or color value of each pixel is extracted sequentially by row and column. For example, for a 256×256 pixel fluorescent cell enhanced image, the pixel value at coordinates (x, y) is extracted point by point. The pixel values ​​in the cell region are mostly in the range of 80-200, while those in the background region are mostly in the range of 0-50. During the extraction process, the spatial location information of the pixels is preserved to ensure that each pixel value corresponds one-to-one with the coordinates in the original image. Finally, multiple current cell image pixel information and multiple historical cell image pixel information containing all pixel values ​​and coordinate information are obtained.

[0068] Step S302: Extract color information from the multiple current cell image pixel information and the multiple historical cell image pixel information to obtain multiple current cell image pixel color information and multiple historical cell image pixel color information.

[0069] In this embodiment, color information extraction can be based on the RGB color model of the color cell image, separating the color values ​​of the red, green, and blue channels from each pixel. For example, for each pixel in the current cell image, the red channel value (R), green channel value (G), and blue channel value (B) are extracted. The red channel value in the cell nucleus region may be higher (e.g., 150-200), while the green channel value in the cytoplasm region may be more significant (e.g., 100-160). By statistically analyzing the color channel distribution of different cell regions, the three-channel color composition of each pixel is obtained, forming multiple current cell image pixel color information and multiple historical cell image pixel color information.

[0070] Step S303: Based on the multiple current cell image pixel color information and the multiple historical cell image pixel color information, perform matching processing on the multiple current cell image pixel information and the multiple historical cell image pixel information to obtain multiple current cell color matching pixel information and multiple historical cell color matching pixel information.

[0071] In this embodiment, the matching process is achieved by comparing pixel color similarity. A color matching threshold is set, such as the difference between the three channel color values ​​being less than 20. The pixel color information of the current cell image is compared with the pixel color information of the same aging stage in historical cell images. For example, the pixel color (R:180, G:120, B:90) of the current young cell region is matched with the typical color (R:175-185, G:115-125, B:85-95) of historical young cells, and pixels with differences within the threshold range are retained. The same matching method is used for intermediate and deep aging cell regions. Finally, multiple current cell color matching pixel information and multiple historical cell color matching pixel information with consistent color characteristics are obtained.

[0072] Step S304: Generate multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors based on the multiple current cell color matching pixel information and the multiple historical cell color matching pixel information.

[0073] In this embodiment, vector generation is achieved by arranging the color values ​​of matching pixels in spatial order, forming a one-dimensional vector by arranging the color matching pixels within each cell region in row-major order. For example, for a young cell region with a diameter of 20 micrometers, containing 500 color matching pixels, the RGB three-channel values ​​of these pixels are arranged sequentially to form a current cell color matching pixel vector with a length of 1500. Vectors for historical cell regions are generated according to the same rules to ensure that the vector dimension is adapted to the size of the cell region. For example, the vector lengths of 30 historical young cells are all in the range of 1200-1800, ensuring consistency in subsequent feature extraction.

[0074] Step S305: Based on multiple preset cell image feature extraction vectors, perform feature extraction processing on the multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors to obtain multiple current cell image feature information and multiple historical cell image feature information.

[0075] In this embodiment, the preset cell image feature extraction vector can be manually preset, constructed based on two-dimensional quaternion singular spectrum analysis, or it can be a selected enhanced color cell image, whose brightness plane and red, green, and blue color planes are represented as quaternion matrices. A 4×4 quaternion matrix window is used to slide through the image, and the spatial relationship between pixels in different color planes is extracted through trajectory matrix decomposition. For example, in a fluorescent cell image, there is a significant color difference between the red channel in the cell nucleus region and the green channel in the cytoplasm. By analyzing the distribution pattern of this difference, the weights of pixels with close color association are set to 1, and the weights of irrelevant pixels are set to 0, forming a vector that can capture color boundary features. Then, it can be combined with empirical modalities. Decomposing and optimizing the vector dimension can be achieved by performing empirical mode decomposition on the cell image signal to obtain multiple intrinsic mode function components. The first six components with concentrated energy are selected through binary programming, and the feature distribution of these components is transformed into the dimension parameters of the vector. For example, the weights of the high-frequency components of nuclear membrane folds and the mid-frequency components of cell membrane texture are increased so that the vector can focus on areas with drastic gray-level changes. Finally, the weights of the corresponding vector positions for the vertical edge regions of cell protrusions can be set to 0.4-0.5 to ensure accurate extraction of texture changes at vertical edges. This allows the cell image feature extraction vector to not only associate with the multi-planar information of the color image but also capture subtle edge and texture features, adapting to the feature extraction needs of cells at different stages of aging.

[0076] This can be achieved by multiplying a preset cell image feature extraction vector with multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors, respectively, to filter out morphological features, texture features, and edge features. These features are then integrated to form multiple current cell image feature information and multiple historical cell image feature information. For example, each current cell feature information contains 20 feature dimensions such as color mean, texture complexity, and edge clarity.

[0077] The cell senescence recognition method based on image processing provided in this application strengthens the correlation between cell color features and senescence state by introducing pixel-level color matching and multi-scale feature extraction vectors for calculation, providing more accurate feature data for subsequent parameter optimization, thereby improving the reliability and accuracy of cell senescence recognition.

[0078] Figure 4 The flowchart illustrating the implementation of the image processing-based cell senescence recognition method provided in Embodiment 4 of this application is shown. Its difference from Embodiment 3 described above lies in:

[0079] Multiple preset cell image feature extraction vectors include preset cell image locking feature extraction vectors, preset cell image association feature extraction vectors, and preset cell image convergence feature extraction vectors;

[0080] Step S305 specifically includes:

[0081] Step S401: Based on the multiple current cell color matching pixel vectors, multiple historical cell color matching pixel vectors, and the preset cell image locking feature extraction vector, obtain multiple current cell image locking feature information and multiple historical cell image locking feature information.

[0082] In this embodiment, the preset cell image locking feature extraction vector can be manually preset and can be used to accurately capture fixed features in cells that are strongly correlated with the aging state, such as cell nucleus morphology and nuclear membrane edge. During construction, cell images after super-resolution reconstruction and enhancement can be selected first, and the distribution of core fixed features of cells at different aging stages can be statistically analyzed. For example, the nucleus of young cells is mostly round and the nuclear membrane edge is smooth, while the nucleus of deeply aging cells has an irregular shape and obvious nuclear membrane edge folds. Then, based on two-dimensional quaternion singular spectrum analysis, a 4×4 matrix window can be used to slide through the cell nucleus region, setting the weight of the nuclear membrane edge pixels to 1.2, the weight of the pixels inside the cell nucleus to 0.8, and the weight of the background region to 0, so that the vector responds preferentially to key fixed features such as the nuclear membrane. Experimental tests show that when the nuclear membrane edge weight is 1.2, the accuracy of nuclear membrane feature recognition of young cells and aging cells reaches 92%, which is significantly higher than other weight distributions. Therefore, the vector weight can be fine-tuned to adapt to the enhanced details, ensuring that the enhanced nuclear membrane folds, cell nucleus contours and other fixed features can be locked. Finally, this weight distribution is determined as the preset cell image locking feature extraction vector. Multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors can be multiplied by a preset cell image locking feature extraction vector, and the multiplication result can be used as multiple current cell image locking feature information and multiple historical cell image locking feature information.

[0083] Step S402: Based on the multiple current cell color matching pixel vectors, multiple historical cell color matching pixel vectors, and preset cell image association feature extraction vectors, multiple current cell image association feature information and multiple historical cell image association feature information are obtained.

[0084] In this embodiment, the preset cell image association feature extraction vector can be used to capture the intrinsic correlation between cell features. This can be achieved by first analyzing the correlation patterns between features in the enhanced cell image. For example, the grayscale difference between the nucleus and cytoplasm texture of young cells is concentrated between 40-60, while the difference decreases to 20-30 in mid-aged cells, and their spatial positions show a center-periphery correspondence. Then, based on empirical mode decomposition, the top four intrinsic mode function components containing feature association information can be selected. The pixel region weights corresponding to these components are set to 1.0, and irrelevant regions are set to 0.3, allowing the vector to focus on regions with close feature associations. This is then adapted to the multi-plane characteristics of color cell images, integrating the correlation relationships of RGB color channels through a quaternion matrix to ensure the vector can capture color-dimensional feature associations. Finally, this structure is determined as the preset cell image association feature extraction vector. Alternatively, multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors can be multiplied by the preset cell image association feature extraction vector, and the multiplication results can be used as multiple current cell image association feature information and multiple historical cell image association feature information.

[0085] Step S403: Based on the multiple current cell color matching pixel vectors, multiple historical cell color matching pixel vectors, and the preset cell image convergence feature extraction vector, obtain multiple current cell image convergence feature information and multiple historical cell image convergence feature information.

[0086] In this embodiment, the preset cell image convergence feature extraction vector can be manually set. It can be used to integrate global features of the cell, such as overall morphology, texture distribution, and edge integrity. It can be determined by statistically analyzing the energy distribution of various features in the cell image, where edge features are concentrated in high-frequency components, morphological features are concentrated in low-frequency components, and texture features are distributed in mid-to-high-frequency components. The vector weights can then be designed to cover the entire image region, with a weight of 1.0 for high-frequency edge regions, 0.9 for low-frequency morphological regions, and 0.8 for mid-to-high-frequency texture regions, ensuring that all types of features are effectively converged. Finally, the weights can be fine-tuned to enhance the convergence of enhanced detail features, for example, increasing the weight of the enhanced cell membrane texture region from 0.8 to 0.9. This weight distribution is then used as the preset cell image convergence feature extraction vector. Alternatively, multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors can be multiplied by the preset cell image convergence feature extraction vector, and the multiplication results serve as multiple current cell image convergence feature information and multiple historical cell image convergence feature information.

[0087] Step S404: Based on the multiple current cell image locking feature information and the multiple current cell image association feature information, obtain multiple current cell image locking association feature information.

[0088] In this embodiment, the dot product of multiple current cell image locking feature information and multiple current cell image associated feature information can be calculated, and the dot product calculation result can be used as multiple current cell image locking associated feature information.

[0089] Step S405: Based on the multiple historical cell image locking feature information and the multiple historical cell image association feature information, obtain multiple historical cell image locking association feature information.

[0090] In this embodiment, the dot product of multiple historical cell image locking feature information and multiple historical cell image association feature information can be calculated, and the dot product calculation result can be used as multiple historical cell image locking association feature information.

[0091] Step S406: Based on the multiple current cell image locking association feature information and the multiple current cell image convergence feature information, multiple current cell image feature information is obtained.

[0092] In this embodiment, multiple current cell image locking associated feature information and multiple current cell image converged feature information can be multiplied together, and the multiplication result can be used as multiple current cell image feature information.

[0093] Step S407: Based on the multiple historical cell image locking association feature information and the multiple historical cell image convergence feature information, multiple historical cell image feature information is obtained.

[0094] In this embodiment, multiple historical cell image locking association feature information and multiple historical cell image convergence feature information can be multiplied together, and the multiplication result can be used as multiple historical cell image feature information.

[0095] The cell senescence recognition method based on image processing provided in this application provides more comprehensive and accurate feature data for cell senescence recognition by dividing and fusing multiple types of feature extraction vectors, focusing on key features while also taking into account correlation patterns and global states, thereby comprehensively improving the reliability and accuracy of cell senescence recognition results.

[0096] Figure 5 The flowchart illustrating the implementation of the image processing-based cell senescence recognition method provided in Embodiment 5 of this application is shown. The difference between this method and Embodiment 1 is that step S104 specifically includes:

[0097] Step S501: Based on the randomly generated initial cell image feature classification parameter information, randomly extract the feature information of the multiple historical cell images to obtain multiple extracted historical cell image feature information.

[0098] In this embodiment, the randomly generated initial cell image feature classification parameters include the number of categories, such as three categories: young, mid-aging, and deep aging, and the extraction ratio, for example, 20%. Based on these parameters, samples are randomly extracted from historical cell image feature information, for example, 200 are extracted from 1000 historical feature information as the extracted historical cell image feature information, ensuring that the extracted samples cover different aging stages, with young cell features accounting for 30%, mid-aging for 50%, and deep aging for 20%, providing initial reference samples for subsequent classification.

[0099] Step S502: Based on the multiple historical cell image feature information and the multiple extracted historical cell image feature information, multiple remaining historical cell image feature information are obtained.

[0100] In this embodiment, the extracted samples can be removed from all historical cell image feature information. For example, 200 extracted features can be removed from 1000 historical features to obtain 800 remaining historical cell image feature information. The remaining features need to retain the original data distribution pattern, and the ratio of young, mid-aging, and deep aging features should be consistent with the original data (3:5:2) to ensure the representativeness of subsequent classification.

[0101] Step S503: Calculate the Euclidean distance between the multiple extracted historical cell image feature information and the multiple remaining historical cell image feature information to obtain multiple historical cell image feature distance information.

[0102] In this embodiment, Euclidean distance calculation is used to measure the similarity between features. For example, the distance between each remaining feature and the same aging stage feature in the extracted features is calculated. The distance between young cell features is mostly concentrated in 0.2-0.5, the distance between mid-aging features is 0.3-0.6, and the distance between deep aging features is 0.4-0.7. The smaller the distance, the more similar the features are. Finally, the distance value between each remaining feature and the extracted feature is obtained as the historical cell image feature distance information.

[0103] Step S504: Based on the distance information of the multiple historical cell image features and the feature information of the multiple extracted historical cell images, classify the feature information of the multiple remaining historical cell images to obtain multiple initial historical cell image feature category information.

[0104] In this embodiment, a nearest neighbor classification rule can be used to assign each remaining feature to the category of the nearest extracted feature. For example, if a remaining feature is 0.3 meters away from the extracted feature of young cells, it is classified into the young cell feature category. After classification, three initial historical cell image feature categories are formed: young, mid-aged, and deep-aged, with each category accounting for 30%, 50%, and 20% of the features, respectively.

[0105] Step S505: Calculate the median and mean of the feature category information of the multiple initial historical cell images to obtain the median and mean of the multiple initial historical cell image feature categories.

[0106] In this embodiment, the median and mean of each initial feature class are calculated: the mean of the initial historical cell image feature class for young cell features is 0.35, and the median of the initial historical cell image feature class is 0.3; the mean of the initial historical cell image feature class for mid-aged cell features is 0.5, and the median of the initial historical cell image feature class is 0.5; the mean of the initial historical cell image feature class for deep aging cell features is 0.65, and the median of the initial historical cell image feature class is 0.6.

[0107] Step S506: Determine the central information of multiple initial historical cell image feature categories based on the number of features in multiple initial historical cell image feature categories, the mean of multiple initial historical cell image feature categories, and a preset threshold for the difference in cell image feature category parameters.

[0108] In this embodiment, the preset threshold for the difference in cell image feature category parameters can be set manually, and can be 0.1. It can be understood that when the difference between the mean of the initial historical cell image feature category and the median of the initial historical cell image feature category is less than 0.1, the mean is taken as the central information of the initial historical cell image feature category; when the difference between the mean of the initial historical cell image feature category and the median of the initial historical cell image feature category is greater than or equal to 0.1, the median is taken as the central information of the initial historical cell image feature category.

[0109] Step S507: Determine whether the central information of the multiple initial historical cell image feature categories is the same as the feature information of the multiple extracted historical cell images; if yes, proceed to step S508; if no, use the central information of the multiple initial historical cell image feature categories as the feature information of the multiple extracted historical cell images, and return to step S502.

[0110] In this embodiment, the central information of the feature category of the initial historical cell image is compared with the feature information of the extracted historical cell image. If they are consistent, the classification is stable; if they are inconsistent, the central information is used as the new core of the extracted feature until the central information and the core of the extracted feature are consistent, thus ensuring classification convergence.

[0111] Step S508: The multiple initial historical cell image feature category information is used as multiple target historical cell image feature category information.

[0112] In this embodiment, after the classification converges, the initial historical cell image feature category information has stably reflected the distribution of historical features. The initial historical cell image feature category information is determined as the target historical cell image feature category information. The central values ​​of the young, mid-aged, and deep-aged features are 0.35, 0.5, and 0.65, respectively, providing a classification standard for subsequent parameter optimization.

[0113] Step S509: Based on the feature category information of the multiple target historical cell images and the calculation parameters of multiple randomly generated initial cell senescence recognition, multiple historical cell senescence recognition images are obtained.

[0114] In this embodiment, the initial cell senescence recognition calculation parameters include feature weights and senescence index thresholds. These parameters can be combined with target category information to quantify historical features and generate historical cell senescence recognition image information that marks young cells as green areas, mid-senescent cells as yellow areas, and deeply senescent cells as red areas, corresponding to historical real senescence state images.

[0115] Step S510: Calculate the accuracy information of multiple historical cell senescence recognition images and a preset cell senescence recognition information database.

[0116] In this embodiment, the preset cell senescence recognition information database can be constructed based on a large number of labeled historical cell senescence images, which may contain 10,000 cell images of known senescence stages. By comparing the labeled regions of the recognized image with the corresponding images in the database, the accuracy of each image is calculated. For example, if the recognition accuracy of a certain historical image is 92%, that is, 92% of the region labels are consistent with the actual senescence stage, multiple historical cell senescence recognition accuracy information are obtained.

[0117] Step S511: Calculate the average of the multiple historical cell senescence recognition accuracy information to obtain the average historical cell senescence recognition accuracy.

[0118] In this embodiment, the accuracy of 100 historical recognition images is averaged. If 80 of them have an accuracy of 90%-95% and 20 have an accuracy of 85%-90%, then the average is (80×92.5%+20×87.5%) / 100=91.5%, which is taken as the average accuracy of historical cell senescence recognition.

[0119] Step S512: Determine whether the average historical cell senescence recognition accuracy is greater than or equal to the preset historical cell senescence recognition accuracy threshold; if yes, proceed to step S513; if no, proceed to step S514.

[0120] In this embodiment, the preset historical cell senescence recognition accuracy threshold is constructed based on clinical application needs and is set to 90% (when the accuracy is ≥90%, it meets practical requirements). If the mean is 91.5% ≥90%, the parameter is valid; if it is 88% <90%, the parameter needs to be adjusted.

[0121] Step S513: Use the randomly generated initial cell image feature classification parameter information as the target cell image feature classification parameter information, and use the multiple randomly generated initial cell senescence recognition calculation parameter information as multiple target cell senescence recognition calculation parameter information.

[0122] In this embodiment, when the accuracy reaches the target, the initial parameters have been adapted to the historical data and are determined as the target parameters. For example, the classification parameters are 3 categories and the extraction ratio is 20%, and the aging recognition parameters are nuclear membrane weight of 0.6 and threshold of 0.3 / 0.7, which provide the final parameters for the current cell recognition.

[0123] Step S514: Adjust the initial cell image feature classification parameter information and the initial cell senescence recognition calculation parameter information according to the preset cell image feature classification parameter adjustment step size and multiple preset cell senescence recognition calculation parameter adjustment step size to obtain intermediate cell image feature classification parameter information and intermediate cell senescence recognition calculation parameter information.

[0124] In this embodiment, the preset cell image feature classification parameter adjustment step size is 0.05 (the extraction ratio is adjusted by ±5% each time), and the weight and threshold of the aging recognition parameter adjustment step size are both 0.05. If the initial extraction ratio of 20% results in insufficient accuracy, it is adjusted to 25%; the nuclear membrane weight is increased from 0.6 to 0.65, and the threshold is decreased from 0.3 to 0.25 to obtain intermediate parameters.

[0125] Step S515: Use the intermediate cell image feature classification parameter information as the randomly generated initial cell image feature classification parameter information, use the multiple intermediate cell senescence recognition calculation parameter information as multiple randomly generated initial cell senescence recognition calculation parameter information, and return to step S501.

[0126] In this embodiment, intermediate parameters are used instead of initial parameters, and steps S501-S512 are repeated until the accuracy reaches the target. For example, after three adjustments, with an extraction ratio of 30%, a nuclear membrane weight of 0.7, and a threshold of 0.2 / 0.75, the average accuracy reaches 92%, and these parameters are finally determined as the target parameters.

[0127] The image processing-based cell senescence recognition method provided in this application improves the stability and accuracy of cell senescence recognition by iteratively optimizing classification parameters and senescence recognition parameters, combined with statistical analysis and accuracy verification of historical data, ensuring that the parameters are adapted to the distribution of cell senescence characteristics.

[0128] Figure 6 The flowchart illustrating the implementation of the image processing-based cell senescence recognition method provided in Embodiment Six of this application is shown. The difference between this method and Embodiment Five is that step S504 specifically includes:

[0129] Step S601: Arrange the multiple historical cell image feature distance information corresponding to each of the remaining historical cell image feature information in ascending order to obtain multiple historical cell image feature distance sequences.

[0130] In this embodiment, for each remaining historical cell image feature, its distance values ​​with all extracted historical cell image features are collected and arranged in ascending order to form a distance sequence. For example, if the distance values ​​between a certain remaining young cell feature and the extracted features are 0.3, 0.5, and 0.7, sorting them in ascending order yields a historical cell image feature distance sequence of [0.3, 0.5, 0.7], ensuring that the closest extracted feature is placed first, providing a clear priority reference for subsequent classification.

[0131] Step S602: The first element value of the multiple historical cell image feature distance sequences is used as the multiple historical cell image feature classification and discrimination distance.

[0132] In this embodiment, the first element of each distance sequence is the distance between the remaining feature and the most similar extracted feature, which is determined as the classification discriminant distance. For example, the first element of the above sequence, 0.3, is used as the classification discriminant distance of the remaining feature. The smaller this value, the higher the similarity between the remaining feature and the extracted feature, providing a key basis for classification decisions. Statistics show that the classification discriminant distance for young cell features is mostly between 0.2 and 0.4, for mid-aged features it is between 0.3 and 0.6, and for deep-aged features it is between 0.4 and 0.7, consistent with the feature difference pattern of different aging stages.

[0133] Step S603: Based on the classification and discrimination distance of the multiple historical cell image features and the multiple extracted historical cell image feature information, classify the multiple remaining historical cell image feature information to obtain multiple initial historical cell image feature category information.

[0134] In this embodiment, each remaining feature is assigned to the category of the extracted feature corresponding to its classification discriminant distance. For example, if the classification discriminant distance of a remaining feature is 0.3, corresponding to the extracted young cell feature, it is assigned to the young cell feature category; if the distance is 0.5, corresponding to the extracted mid-aged feature, it is assigned to the mid-aged category. The rationality of the classification can be verified by combining preset classification distance thresholds, such as young cells ≤ 0.4, mid-aged 0.4-0.6, and deep aging ≥ 0.6, ultimately forming three initial historical cell image feature category information, where the proportions of young, mid-aged, and deep aging features are 30%, 50%, and 20%, respectively, consistent with the historical data distribution.

[0135] The cell senescence recognition method based on image processing provided in this application enhances the accuracy of feature classification by sorting distance information and filtering the discrimination distance, ensuring the consistency of the category of the remaining features with the most similar extracted features, providing a more reliable classification basis for subsequent parameter optimization, and significantly improving the stability and accuracy of cell senescence recognition.

[0136] Figure 7 The flowchart illustrating the implementation of the image processing-based cell senescence recognition method provided in Embodiment Seven of this application is shown. The difference between this method and Embodiment Five is that step S506 specifically includes:

[0137] Step S701: Calculate the difference between the number of the multiple initial historical cell image feature categories and the mean of the multiple initial historical cell image feature categories to obtain the difference of multiple cell image feature category parameters.

[0138] In this embodiment, for each type of initial historical cell image feature category information, the absolute difference between the number of features in the initial historical cell image feature category and the mean of the initial historical cell image feature category is calculated. For example, the number of features in the initial historical cell image feature category for the young cell feature category is 0.3, and the mean of the initial historical cell image feature category is 0.35, with a difference of 0.05; the number of features in the initial historical cell image feature category for the mid-term senescent cell feature category is 0.5, and the mean of the initial historical cell image feature category is 0.5, with a difference of 0; the number of features in the initial historical cell image feature category for the deep senescent cell feature category is 0.6, and the mean of the initial historical cell image feature category is 0.65, with a difference of 0.05. These differences are the differences of multiple cell image feature category parameters, used to reflect the degree of concentration of the distribution of each type of feature data.

[0139] Step S702: Determine whether the difference in the cell image feature category parameters is greater than a preset threshold for the difference in cell image feature category parameters; if yes, proceed to step S703; if no, proceed to step S704.

[0140] In this embodiment, the preset threshold for the difference in cell image feature category parameters is manually set. Its construction is based on stability analysis of historical feature data and is set to 0.1. This threshold is used to determine whether the difference between the mean and median affects the representativeness of the category center. Each cell image feature category parameter difference is compared with this threshold. For example, the difference for young cell features is 0.05 ≤ 0.1, the difference for mid-aged cell features is 0 ≤ 0.1, and the difference for deep-aged cell features is 0.05 ≤ 0.1, all of which are not greater than the threshold. If the difference for a certain feature is 0.12, it is determined to be greater than the threshold.

[0141] Step S703: Determine the number of features in the initial historical cell image feature category as the central information of the initial historical cell image feature category.

[0142] In this embodiment, when the difference in cell image feature category parameters exceeds a preset threshold, it indicates that the feature data distribution may be skewed. The median is more representative of the central tendency of the data. Therefore, the initial historical cell image feature category median is used as the central information of that category. For example, if the initial historical cell image feature category median of a certain feature is 0.4, the initial historical cell image feature category mean is 0.55, and the difference 0.15 > 0.1, then 0.4 is used as the initial historical cell image feature category central information of that category to ensure that the central information is not disturbed by extreme values.

[0143] Step S704: The mean value of the initial historical cell image feature categories is determined as the central information of the initial historical cell image feature categories.

[0144] In this embodiment, when the difference in cell image feature category parameters is less than or equal to a preset threshold, it indicates that the feature data distribution is relatively symmetrical, and the mean effectively reflects the central tendency of the data. Therefore, the mean of the initial historical cell image feature categories is determined as the central information of the initial historical cell image feature categories. For example, the difference in the feature categories of young cells is 0.05≤0.1, and the mean is 0.35 as the central information; the difference in the feature categories of mid-aged cells is 0≤0.1, and the mean is 0.5 as the central information; the difference in the feature categories of deeply aged cells is 0.05≤0.1, and the mean is 0.65 as the central information, making the central information more consistent with the average level of the data.

[0145] The image processing-based cell senescence recognition method provided in this application dynamically selects central information by quantitatively analyzing the difference between the mean and the median and combining it with a preset threshold. This ensures that the central information can accurately reflect the core of the category with different distribution characteristics, improves the reliability of the classification standard, and provides a more accurate benchmark for subsequent parameter optimization and senescence recognition, thereby enhancing the stability and accuracy of cell senescence recognition.

[0146] Corresponding to the method in the above embodiments, Figure 8 The diagram shows a structural block diagram of an image processing-based cell senescence recognition system provided in an embodiment of this application. For ease of explanation, only the parts related to the embodiments of this application are shown. Figure 8 The example image processing-based cell senescence recognition system can be the execution subject of the image processing-based cell senescence recognition method provided in the aforementioned embodiment 1.

[0147] Reference Figure 8 The image processing-based cell senescence recognition system includes:

[0148] The cell image information acquisition module 810 is used to acquire multiple current cell image information and multiple historical cell image information;

[0149] The cell image enhancement information generation module 820 is used to perform sampling and enhancement processing on the multiple current cell image information and multiple historical cell image information based on multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors and preset cell image vertical edge enhancement vectors, to obtain multiple current cell image enhancement information and multiple historical cell image enhancement information.

[0150] The cell image feature information generation module 830 is used to perform feature extraction processing on the multiple current cell image enhancement information and the multiple historical cell image enhancement information according to multiple preset cell image feature extraction vectors, so as to obtain multiple current cell image feature information and multiple historical cell image feature information.

[0151] The target cell image feature classification parameter information and target cell senescence recognition calculation parameter information determination module 840 is used to calculate the target cell image feature classification parameter information and multiple target cell senescence recognition calculation parameter information based on the multiple historical cell image feature information, randomly generated initial cell image feature classification parameter information, multiple randomly generated initial cell senescence recognition calculation parameter information, preset cell image feature classification parameter adjustment step size and multiple preset cell senescence recognition calculation parameter adjustment step size.

[0152] The current cell senescence recognition image information generation module 850 is used to calculate the current cell senescence recognition image information based on the multiple current cell image feature information, target cell image feature classification parameter information, and multiple target cell senescence recognition calculation parameter information.

[0153] The process by which each module in the image processing-based cell senescence recognition system provided in this application implements its respective function can be found in the foregoing. Figure 1 The description of Embodiment 1 shown will not be repeated here.

[0154] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0155] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0156] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0157] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0158] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance. It should also be understood that although the terms "first," "second," etc., are used in the text to describe various elements in some embodiments of this application, these elements should not be limited by these terms. These terms are merely used to distinguish one element from another. For example, a first table may be named a second table, and similarly, a second table may be named a first table, without departing from the scope of the various described embodiments. Both the first table and the second table are tables, but they are not the same table.

[0159] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0160] The cell senescence recognition method based on image processing provided in this application can be applied to terminal devices such as mobile phones, tablets, wearable devices, in-vehicle devices, augmented reality (AR) / virtual reality (VR) devices, laptops, ultra-mobile personal computers (UMPCs), netbooks, and personal digital assistants (PDAs). This application does not impose any restrictions on the specific type of terminal device.

[0161] For example, the terminal device may be a station (STAION, ST) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA) device, a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, an in-vehicle device, a vehicle networking terminal, a computer, a laptop computer, a handheld communication device, a handheld computing device, a satellite wireless device, a wireless modem card, a set-top box (STB), customer premises equipment (CPE), and / or other devices used for communication over a wireless system, as well as next-generation communication systems, such as mobile terminals in 5G networks or mobile terminals in future evolved Public Land Mobile Network (PLMN) networks.

[0162] As an example and not a limitation, when the terminal device is a wearable device, the term "wearable device" can also refer to any device that utilizes wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices worn directly on the body or integrated into a user's clothing or accessories. Wearable devices are not merely hardware devices; they achieve powerful functions through software support, data interaction, and cloud interaction. Broadly defined, wearable smart devices include those with comprehensive functions, large sizes, and the ability to perform complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those focused on a specific application function that require interaction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0163] Figure 9 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application. For example... Figure 9 As shown, the terminal device 9 of this embodiment includes: at least one processor 90 ( Figure 9 (Only one is shown in the image), a memory 91, which stores a computer program 92 that can run on the processor 90. When the processor 90 executes the computer program 92, it implements the steps in the various image processing-based cell senescence recognition method embodiments described above, for example... Figure 1 Steps S101 to S105 are shown. Alternatively, when the processor 90 executes the computer program 92, it implements the functions of each module / unit in the above system embodiments, for example... Figure 8 The functions of modules 810 to 850 are shown.

[0164] The terminal device 9 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor 90 and a memory 91. Those skilled in the art will understand that... Figure 9 This is merely an example of terminal device 9 and does not constitute a limitation on terminal device 9. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal device may also include input transmission devices, network access devices, buses, etc.

[0165] The processor 90 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0166] In some embodiments, the memory 91 may be an internal storage unit of the terminal device 9, such as a hard disk or memory of the terminal device 9. The memory 91 may also be an external storage device of the terminal device 9, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal device 9. Furthermore, the memory 91 may include both internal and external storage units of the terminal device 9. The memory 91 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory 91 can also be used to temporarily store data that has been sent or will be sent.

[0167] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0168] This application also provides a terminal device, which includes at least one memory, at least one processor, and a computer program stored in the at least one memory and executable on the at least one processor. When the processor executes the computer program, it causes the terminal device to implement the steps in any of the above method embodiments.

[0169] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0170] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0171] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0172] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A cell senescence recognition method based on image processing, characterized in that, include: Acquire information from multiple current cell images and multiple historical cell images; Based on multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors, and preset cell image vertical edge enhancement vectors, sampling and enhancement processing are performed on the multiple current cell image information and multiple historical cell image information to obtain multiple current cell image enhancement information and multiple historical cell image enhancement information. Based on multiple preset cell image feature extraction vectors, feature extraction processing is performed on the multiple current cell image enhancement information and multiple historical cell image enhancement information to obtain multiple current cell image feature information and multiple historical cell image feature information; Based on the aforementioned historical cell image feature information, randomly generated initial cell image feature classification parameter information, multiple randomly generated initial cell senescence recognition calculation parameter information, preset cell image feature classification parameter adjustment step size, and multiple preset cell senescence recognition calculation parameter adjustment step size, the target cell image feature classification parameter information and multiple target cell senescence recognition calculation parameter information are calculated. Based on the multiple current cell image feature information, target cell image feature classification parameter information, and multiple target cell senescence recognition calculation parameter information, the current cell senescence recognition image information is calculated. The step of calculating the target cell image feature classification parameters and multiple target cell aging recognition calculation parameters based on the multiple historical cell image feature information, randomly generated initial cell image feature classification parameters, multiple randomly generated initial cell aging recognition calculation parameters, preset cell image feature classification parameter adjustment step size, and multiple preset cell aging recognition calculation parameter adjustment step size specifically includes: Based on the randomly generated initial cell image feature classification parameter information, the feature information of the multiple historical cell images is randomly extracted to obtain multiple extracted historical cell image feature information. Based on the feature information of the multiple historical cell images and the feature information of the multiple extracted historical cell images, multiple feature information of the remaining historical cell images are obtained; Calculate the Euclidean distance between the multiple extracted historical cell image feature information and the multiple remaining historical cell image feature information to obtain multiple historical cell image feature distance information; Based on the distance information of the multiple historical cell image features and the feature information of the multiple extracted historical cell images, the feature information of the multiple remaining historical cell images is classified to obtain multiple initial historical cell image feature category information. Calculate the median and mean of the feature category information of the multiple initial historical cell images to obtain the median and mean of the feature categories of the multiple initial historical cell images; Based on the number of initial historical cell image feature categories, the mean of multiple initial historical cell image feature categories, and a preset threshold for the difference in cell image feature category parameters, the central information of multiple initial historical cell image feature categories is determined; Determine whether the central information of the feature categories of the multiple initial historical cell images is the same as the feature information of the multiple extracted historical cell images; If so, the multiple initial historical cell image feature category information shall be used as the multiple target historical cell image feature category information; If not, the central information of the multiple initial historical cell image feature categories is used as multiple extracted historical cell image feature information, and the process returns to the step of obtaining multiple remaining historical cell image feature information based on the multiple historical cell image feature information and the multiple extracted historical cell image feature information. Based on the feature category information of the multiple target historical cell images and the calculation parameters of multiple randomly generated initial cell senescence recognition, multiple historical cell senescence recognition image information are obtained; Based on the multiple historical cell senescence recognition image information and the preset cell senescence recognition information database, the accuracy information of multiple historical cell senescence recognition is calculated. The mean of the accuracy information for identifying historical cell senescence is calculated to obtain the mean accuracy of historical cell senescence identification. Determine whether the average historical cell senescence recognition accuracy rate is greater than or equal to a preset historical cell senescence recognition accuracy rate threshold. If so, the randomly generated initial cell image feature classification parameter information is used as the target cell image feature classification parameter information, and the multiple randomly generated initial cell senescence recognition calculation parameter information is used as multiple target cell senescence recognition calculation parameter information; If not, the initial cell image feature classification parameter information and the initial cell senescence recognition calculation parameter information are adjusted according to the preset cell image feature classification parameter adjustment step size and multiple preset cell senescence recognition calculation parameter adjustment step size to obtain intermediate cell image feature classification parameter information and intermediate cell senescence recognition calculation parameter information. The intermediate cell image feature classification parameter information is used as the randomly generated initial cell image feature classification parameter information, and multiple intermediate cell senescence recognition calculation parameter information is used as multiple randomly generated initial cell senescence recognition calculation parameter information. Then, the process is returned to the step of randomly extracting multiple historical cell image feature information based on the randomly generated initial cell image feature classification parameter information to obtain multiple extracted historical cell image feature information.

2. The cell senescence recognition method based on image processing as described in claim 1, characterized in that, Multiple preset cell image sampling vectors include a preset first-scale cell image sampling vector and a preset second-scale cell image sampling vector; The step of sampling and enhancing multiple current cell image information and multiple historical cell image information based on multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors, and preset cell image vertical edge enhancement vectors to obtain multiple current cell image enhancement information and multiple historical cell image enhancement information specifically includes: Based on a preset first-scale cell image sampling vector, the multiple current cell image information and multiple historical cell image information are sampled and processed to obtain multiple first-scale current cell image sampling information and multiple first-scale historical cell image sampling information. Based on a preset second-scale cell image sampling vector, the multiple current cell image information and multiple historical cell image information are sampled and processed to obtain multiple second-scale current cell image sampling information and multiple second-scale historical cell image sampling information. The multiple first-scale current cell image sampling information and the multiple second-scale current cell image sampling information are stitched together to generate multiple current cell image sampling information. The multiple first-scale historical cell image sampling information and the multiple second-scale historical cell image sampling information are stitched together to generate multiple historical cell image sampling information. Based on the preset horizontal edge enhancement vector and the preset vertical edge enhancement vector of the cell image, the multiple current cell image sampling information and the multiple historical cell image sampling information are enhanced to obtain multiple current cell image edge enhancement information and multiple historical cell image edge enhancement information. Based on the multiple current cell image edge enhancement information, multiple historical cell image edge enhancement information, multiple current cell image information, and multiple historical cell image information, multiple current cell image enhancement information and multiple historical cell image enhancement information are obtained.

3. The cell senescence recognition method based on image processing as described in claim 1, characterized in that, The step of performing feature extraction processing on the multiple current cell image enhancement information and the multiple historical cell image enhancement information based on multiple preset cell image feature extraction vectors to obtain multiple current cell image feature information and multiple historical cell image feature information specifically includes: Pixel extraction is performed on the multiple current cell image enhancement information and multiple historical cell image enhancement information to obtain multiple current cell image pixel information and multiple historical cell image pixel information; Color information is extracted from the pixel information of the multiple current cell images and the pixel information of the multiple historical cell images to obtain the color information of the pixels of the multiple current cell images and the color information of the pixels of the multiple historical cell images. Based on the multiple current cell image pixel color information and the multiple historical cell image pixel color information, the multiple current cell image pixel information and the multiple historical cell image pixel information are matched to obtain multiple current cell color matching pixel information and multiple historical cell color matching pixel information. Based on the multiple current cell color matching pixel information and the multiple historical cell color matching pixel information, generate multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors. Based on multiple preset cell image feature extraction vectors, feature extraction processing is performed on the multiple current cell color matching pixel vectors and multiple historical cell color matching pixel vectors to obtain multiple current cell image feature information and multiple historical cell image feature information.

4. The cell senescence recognition method based on image processing as described in claim 3, characterized in that, Multiple preset cell image feature extraction vectors include preset cell image locking feature extraction vectors, preset cell image association feature extraction vectors, and preset cell image convergence feature extraction vectors; The step of performing feature extraction processing on the multiple current cell color matching pixel vectors and the multiple historical cell color matching pixel vectors based on multiple preset cell image feature extraction vectors to obtain multiple current cell image feature information and multiple historical cell image feature information specifically includes: Based on the multiple current cell color matching pixel vectors, multiple historical cell color matching pixel vectors, and the preset cell image locking feature extraction vector, multiple current cell image locking feature information and multiple historical cell image locking feature information are obtained. Based on the multiple current cell color matching pixel vectors, multiple historical cell color matching pixel vectors, and the preset cell image association feature extraction vectors, multiple current cell image association feature information and multiple historical cell image association feature information are obtained. Based on the multiple current cell color matching pixel vectors, multiple historical cell color matching pixel vectors, and the preset cell image convergence feature extraction vector, multiple current cell image convergence feature information and multiple historical cell image convergence feature information are obtained. Based on the multiple current cell image locking feature information and the multiple current cell image association feature information, multiple current cell image locking association feature information is obtained; Based on the locking feature information of the multiple historical cell images and the association feature information of the multiple historical cell images, the locking association feature information of the multiple historical cell images is obtained; Based on the multiple current cell image locking association feature information and the multiple current cell image convergence feature information, multiple current cell image feature information is obtained; Based on the associated feature information of the multiple historical cell images and the converged feature information of the multiple historical cell images, multiple historical cell image feature information is obtained.

5. The cell senescence recognition method based on image processing as described in claim 1, characterized in that, The step of classifying the remaining historical cell image feature information based on the distance information of the multiple historical cell image features and the feature information of the multiple extracted historical cell images to obtain multiple initial historical cell image feature category information specifically includes: Arrange the multiple historical cell image feature distance information corresponding to each of the remaining historical cell image feature information in ascending order to obtain multiple historical cell image feature distance sequences. The first element of the distance sequence of the multiple historical cell image features is used as the classification and discrimination distance of the multiple historical cell image features; Based on the classification and discrimination distance of the multiple historical cell image features and the feature information of the multiple extracted historical cell images, the feature information of the multiple remaining historical cell images is classified to obtain multiple initial historical cell image feature category information.

6. The cell senescence recognition method based on image processing as described in claim 1, characterized in that, The step of determining the central information of multiple initial historical cell image feature categories based on the number of features in multiple initial historical cell image feature categories, the mean of multiple initial historical cell image feature categories, and a preset threshold for the difference in cell image feature category parameters specifically includes: Calculate the difference between the number of feature categories in the multiple initial historical cell images and the mean of the multiple initial historical cell image feature categories to obtain the parameter difference of multiple cell image feature categories; Determine whether the difference in the cell image feature category parameters is greater than a preset threshold for the difference in cell image feature category parameters; If so, the number in the initial historical cell image feature category is determined as the central information of the initial historical cell image feature category; If not, the mean value of the initial historical cell image feature categories is determined as the central information of the initial historical cell image feature categories.

7. A cell senescence recognition system based on image processing, characterized in that, include: The cell image information acquisition module is used to acquire multiple current cell image information and multiple historical cell image information; The cell image enhancement information generation module is used to sample and enhance multiple current cell image information and multiple historical cell image information based on multiple preset cell image sampling vectors, preset cell image horizontal edge enhancement vectors and preset cell image vertical edge enhancement vectors, to obtain multiple current cell image enhancement information and multiple historical cell image enhancement information. The cell image feature information generation module is used to perform feature extraction processing on the multiple current cell image enhancement information and the multiple historical cell image enhancement information according to multiple preset cell image feature extraction vectors, so as to obtain multiple current cell image feature information and multiple historical cell image feature information. The target cell image feature classification parameter information and target cell senescence recognition calculation parameter information determination module is used to calculate the target cell image feature classification parameter information and multiple target cell senescence recognition calculation parameter information based on the multiple historical cell image feature information, randomly generated initial cell image feature classification parameter information, multiple randomly generated initial cell senescence recognition calculation parameter information, preset cell image feature classification parameter adjustment step size and multiple preset cell senescence recognition calculation parameter adjustment step size. The current cell senescence recognition image information generation module is used to calculate the current cell senescence recognition image information based on the multiple current cell image feature information, target cell image feature classification parameter information, and multiple target cell senescence recognition calculation parameter information. The step of calculating the target cell image feature classification parameters and multiple target cell aging recognition calculation parameters based on the multiple historical cell image feature information, randomly generated initial cell image feature classification parameters, multiple randomly generated initial cell aging recognition calculation parameters, preset cell image feature classification parameter adjustment step size, and multiple preset cell aging recognition calculation parameter adjustment step size specifically includes: Based on the randomly generated initial cell image feature classification parameter information, the feature information of the multiple historical cell images is randomly extracted to obtain multiple extracted historical cell image feature information. Based on the feature information of the multiple historical cell images and the feature information of the multiple extracted historical cell images, multiple feature information of the remaining historical cell images are obtained; Calculate the Euclidean distance between the multiple extracted historical cell image feature information and the multiple remaining historical cell image feature information to obtain multiple historical cell image feature distance information; Based on the distance information of the multiple historical cell image features and the feature information of the multiple extracted historical cell images, the feature information of the multiple remaining historical cell images is classified to obtain multiple initial historical cell image feature category information. Calculate the median and mean of the feature category information of the multiple initial historical cell images to obtain the median and mean of the feature categories of the multiple initial historical cell images; Based on the number of initial historical cell image feature categories, the mean of multiple initial historical cell image feature categories, and a preset threshold for the difference in cell image feature category parameters, the central information of multiple initial historical cell image feature categories is determined; Determine whether the central information of the feature categories of the multiple initial historical cell images is the same as the feature information of the multiple extracted historical cell images; If so, the multiple initial historical cell image feature category information shall be used as the multiple target historical cell image feature category information; If not, the central information of the multiple initial historical cell image feature categories is used as multiple extracted historical cell image feature information, and the process returns to the step of obtaining multiple remaining historical cell image feature information based on the multiple historical cell image feature information and the multiple extracted historical cell image feature information. Based on the feature category information of the multiple target historical cell images and the calculation parameters of multiple randomly generated initial cell senescence recognition, multiple historical cell senescence recognition image information are obtained; Based on the multiple historical cell senescence recognition image information and the preset cell senescence recognition information database, the accuracy information of multiple historical cell senescence recognition is calculated. The mean of the accuracy information for identifying historical cell senescence is calculated to obtain the mean accuracy of historical cell senescence identification. Determine whether the average historical cell senescence recognition accuracy rate is greater than or equal to a preset historical cell senescence recognition accuracy rate threshold. If so, the randomly generated initial cell image feature classification parameter information is used as the target cell image feature classification parameter information, and the multiple randomly generated initial cell senescence recognition calculation parameter information is used as multiple target cell senescence recognition calculation parameter information; If not, the initial cell image feature classification parameter information and the initial cell senescence recognition calculation parameter information are adjusted according to the preset cell image feature classification parameter adjustment step size and multiple preset cell senescence recognition calculation parameter adjustment step size to obtain intermediate cell image feature classification parameter information and intermediate cell senescence recognition calculation parameter information. The intermediate cell image feature classification parameter information is used as the randomly generated initial cell image feature classification parameter information, and multiple intermediate cell senescence recognition calculation parameter information is used as multiple randomly generated initial cell senescence recognition calculation parameter information. Then, the process is returned to the step of randomly extracting multiple historical cell image feature information based on the randomly generated initial cell image feature classification parameter information to obtain multiple extracted historical cell image feature information.

8. A terminal device, characterized in that, The terminal device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6.

Images