Umbilical cord mesenchymal stem cell preparation process monitoring method and system

Abstract

The present invention relates to the field of image analysis technology, specifically to a method and system for monitoring the preparation process of umbilical cord mesenchymal stem cells, comprising the following steps: based on the umbilical cord mesenchymal stem cell image, image size standardization, noise removal, cell region cropping, grayscale value comparison to remove grayscale value regions below a set threshold, and filtering to eliminate background noise, remove noise points, and extract cell edge information to generate a denoised cell image. In the present invention, by introducing a residual network and convolution operation, cell morphological changes are accurately captured, providing more accurate proliferation state analysis, cell boundary extraction and cycle classification improvements, effectively avoiding the misjudgment of cell state in traditional methods, and deep learning training optimizes image segmentation, making cell cycle division more accurate, ensuring the efficiency and accuracy of the entire monitoring process, and ensuring the standardization and efficiency of the stem cell preparation process.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a method and system for monitoring the preparation process of umbilical cord mesenchymal stem cells. Background Art

[0002] The field of image analysis technology encompasses techniques for extracting useful information from images through image acquisition, processing, and analysis. Core areas within this field include digital image processing, pattern recognition, image segmentation, image enhancement, and feature extraction. Image analysis technology is widely used in a variety of fields, including medical imaging, industrial inspection, machine vision, and security monitoring. Image analysis enables the identification and classification of different objects, forms, and behaviors within images, enabling automated data extraction and processing to support various decision-making processes. In medicine, image analysis is particularly useful for pathological diagnosis, surgical assistance, and disease detection, becoming a crucial tool for assisting diagnosis and treatment.

[0003] Among them, the umbilical cord mesenchymal stem cell preparation process monitoring method refers to a method for real-time monitoring of the umbilical cord mesenchymal stem cell preparation process through image analysis technology. The subject of this patent involves the image acquisition and processing of umbilical cord mesenchymal stem cells during the preparation process, mainly through a microscope or other imaging equipment to obtain image data during the cell culture process, and process and analyze these data. This technology solves the problems of how to accurately monitor cell growth, division, state changes, etc. during the stem cell preparation process. Its means include image acquisition, image processing and analysis, mainly through automated image recognition technology to track and record cell status in real time, thereby ensuring the standardization and efficiency of stem cell preparation.

[0004] The existing technology lacks detailed monitoring and real-time feedback of cell morphological changes during the preparation of umbilical cord mesenchymal stem cells, especially in the identification of cell proliferation status and morphological changes, where large errors often occur. The existing technology relies on simple image processing methods and may not be able to effectively handle the dynamic changes of complex cell morphology, resulting in inaccurate distinction between cell division and quiescence. In addition, the existing technology may be affected by background noise and interference factors during cell cycle classification and image segmentation, and may not be able to accurately divide the cycle stages, thereby affecting the accuracy and standardization of the stem cell preparation process. For example, traditional image processing methods may have fuzzy phenomena when processing cell edges, resulting in unclear identification of cell boundaries, which in turn affects subsequent analysis and judgment. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a method and system for monitoring the preparation process of umbilical cord mesenchymal stem cells.

[0006] In order to achieve the above-mentioned object, the present invention adopts the following technical solution: a method for monitoring the preparation process of umbilical cord mesenchymal stem cells, comprising the following steps:

[0007] S1: Based on the umbilical cord mesenchymal stem cell image, image size normalization, noise removal, cell area cropping, grayscale value comparison to remove grayscale value areas below the set threshold, and filtering methods to eliminate background noise, remove noise points and extract cell edge information to generate a denoised cell image;

[0008] S2: Based on the denoised cell image, the cell image is input into the residual network, the output features of each layer are weighted and processed by convolution operation, the convolution kernel size is adjusted to match the cell morphological changes by applying a nonlinear activation function, the influence of the cell proliferation state on the morphological features is calculated, and the cell edge feature map is generated;

[0009] S3: Based on the cell edge feature map, edge change and grayscale comparison are performed to identify and extract the boundaries of the umbilical cord mesenchymal stem cells, perform region division and boundary extraction, and combine the cell morphology monitoring information to perform boundary value comparison analysis to generate cell boundary interval values;

[0010] S4: Based on the cell boundary interval value, cell cycle stage classification is performed, periodic analysis is performed through cell boundary changes, standards are set to divide the division phase and the resting phase, cell morphological changes within the cycle are calculated, and a cell cycle stage boundary map is generated;

[0011] S5: Based on the cell cycle phase boundary map, perform deep learning training, perform cycle state segmentation and optimization, adjust the neural network layer depth and learning rate, optimize the image segmentation process, adjust the learning rate and weight of each layer, and obtain the optimized segmentation result map.

[0012] The denoised cell image specifically includes a cell image, a noise removal area, and cell edge information. The cell edge feature map includes convolution features, activation function adjustment, and morphological change effects. The cell boundary interval value includes edge changes, grayscale contrast, boundary extraction, and morphological monitoring. The cell cycle stage boundary map includes the division phase, the resting phase, and cell morphological changes. The optimized segmentation result map specifically includes cycle state optimization, network hierarchy adjustment, and segmentation results.

[0013] As a further solution of the present invention, the step of acquiring the denoised cell image is specifically as follows:

[0014] S101: Based on the input umbilical cord mesenchymal stem cell images, all images are resized by a unified scaling ratio. At the same time, the grayscale value of each pixel is compared with the threshold according to a set threshold, and grayscale values ​​less than the threshold are filtered out to generate a standardized image.

[0015] S102: filtering the standardized image using a high-pass or low-pass filter to distinguish the differentiated noise frequencies, adjusting the filter intensity using filter design parameters, and separating the cell region from other regions through threshold determination to establish a denoised cell region;

[0016] S103: Based on the denoised cell area, the edge position is located by calculating the grayscale difference between each pixel in the image and its neighboring pixels, and the edge strength is compared using a preset threshold, using the formula:

[0017]

[0018] Calculate cell edge information and generate denoised cell images;

[0019] Among them, E represents the amount of edge information, I i represents the grayscale value of the i-th pixel in the image, α is the edge clarity coefficient, β is the edge smoothness adjustment coefficient, γ is the detail enhancement coefficient, and N is the total number of pixels in the image.

[0020] As a further embodiment of the present invention, the step of obtaining the cell edge feature map is specifically as follows:

[0021] S201: Based on the denoised cell image, the cell image is input into the residual network, the feature map of each layer is weighted by a convolution operation, the image is filtered multiple times by a filter, the output features of the differentiation layer are weighted, the convolution kernel size of the differentiation layer is adjusted, and the convolution output of each layer is calculated to obtain a convolution result image;

[0022] S202: applying a nonlinear activation function to adjust the convolution result image, adjusting the nonlinear effect of the output feature according to the changes in cell morphology and the differentiated morphology of the cells, and obtaining a feature map that matches the cell morphology by adjusting the size of the convolution kernel;

[0023] S203: Based on the feature map that matches the cell morphology, and according to the relationship between the proliferation state and the cell morphology, analyze the changes in the cell morphological characteristics under the differential proliferation state. By calculating the proliferation rate and the amplitude of the morphological change, and combining the effect of cell proliferation on the image characteristics, adjust the feature map to reflect the changes in the cell proliferation state, using the formula:

[0024]

[0025] Calculate the effect of cell proliferation on morphological characteristics and generate cell edge feature maps;

[0026] Among them, C represents the change value of the morphological characteristic map, F irepresents the gray value of the i-th pixel in the image, α is the coefficient of morphological change, β is the reference coefficient of morphological change, γ is the influence coefficient of proliferation rate, δ is the adjustment factor, and M is the total number of pixels in the image.

[0027] As a further solution of the present invention, the step of obtaining the cell boundary interval value is specifically as follows:

[0028] S301: extracting the grayscale gradient value of the cell edge change based on the cell edge feature map, and screening the pixel points that meet the cell boundary characteristics by comparing the grayscale gradient value with the target threshold to generate a cell edge pixel feature set;

[0029] S302: calling the cell edge pixel feature set, calculating the continuity value of the edge change based on the directionality of the edge change and the grayscale distribution uniformity index, performing region division, and generating a region boundary feature set;

[0030] S303: Call the region boundary feature set, calculate the difference between the boundary feature parameters, and use the formula:

[0031]

[0032] Calculate the change intensity of the region boundary, compare and extract the maximum change intensity value, and generate the cell boundary interval value;

[0033] Among them, B i Indicates the boundary interval value, G j Represents the grayscale value of the j-th boundary pixel, G j+1 Indicates the gray value of the adjacent pixel, D j Represents the distance between pixels, is the average value of all distances, α is the adjustment coefficient of grayscale change, β is the smoothing parameter of the region boundary feature set, and n is the number of boundary pixels.

[0034] As a further embodiment of the present invention, the steps for obtaining the cell cycle phase boundary map are specifically as follows:

[0035] S401: Based on the cell boundary interval value, by cell cycle stage classification, periodic analysis of cell boundary changes is performed, difference values ​​of boundary changes within the cell cycle are calculated, standard divisions between the division phase and the resting phase are set, and a cell cycle stage boundary map is generated;

[0036] S402: calling the cell cycle phase boundary map, combining the temporal characteristics of boundary changes, calculating the cell morphological changes in the differentiation phase, analyzing the magnitude of changes in multiple phases within the cycle, and generating a cell boundary distribution map for the differentiation phase within the cycle;

[0037] S403: In the cell boundary distribution map at the differentiated stage within the cycle, the formula:

[0038]

[0039] Calculate the boundary change intensity of multiple cell cycle stages, optimize the division of multiple stages within the cycle, and generate a cell cycle stage boundary map;

[0040] Among them, C s Indicates the intensity of the boundary change of the cycle stage, P k represents the boundary value of the k-th cell stage, Y k represents the boundary position of the kth stage, Y k+1 represents the boundary position between adjacent stages, is the mean of the boundary position, β is the adjustment coefficient, γ t is the time adjustment coefficient, and m is the total number of stages in the cell cycle.

[0041] As a further solution of the present invention, the step of obtaining the optimized segmentation result map is specifically as follows:

[0042] S501: Based on the cell cycle phase boundary map, performing cycle state segmentation and optimization through deep learning training, combining image data and neural network model, adjusting the neural network layer depth and learning rate, and generating an optimized neural network model;

[0043] S502: calling the optimized neural network model, adjusting the learning rate and weight of each layer, performing back propagation and optimizing the image segmentation process, calculating the gradient value of the weight of each layer, and generating an optimized segmentation strategy;

[0044] S503: In the application of the optimized segmentation strategy, the formula is used:

[0045]

[0046] Calculate the error function of each neural network layer and adjust the parameters to generate the optimized segmentation result map;

[0047] Among them, L represents the loss function value, Z i represents the true value of the i-th image pixel, represents the predicted value, λ is the regularization coefficient, α j is the learning rate adjustment coefficient of layer j, W j represents the weight value of the jth layer, n is the total number of pixels, and m is the number of network layers.

[0048] An umbilical cord mesenchymal stem cell production process monitoring system is used to implement the above-mentioned umbilical cord mesenchymal stem cell production process monitoring method, and the system includes:

[0049] The image processing module normalizes the image size, adjusts the aspect ratio of the image, performs noise removal, removes background noise, crops the cell area, removes areas with grayscale values ​​below a set threshold, eliminates noise in the image through grayscale value contrast operations, extracts cell edge information, and generates a denoised cell image.

[0050] The feature extraction module is based on the denoised cell image, inputs the image into the residual network, processes the output features through convolution operations, adjusts the convolution kernel size to match the cell morphology changes, calculates the impact of the proliferation state on the cell morphology, extracts the features of the cell edge, and generates a cell edge feature map;

[0051] The boundary recognition module performs edge change and grayscale comparison based on the cell edge feature map, identifies the boundary area between the cell and the background, extracts the edge position of the cell, performs region division and boundary extraction, and combines the cell morphology information to perform boundary value comparison and analysis to obtain the cell boundary interval value;

[0052] The cycle analysis module classifies the cell cycle stages based on the cell boundary interval values, analyzes the periodicity through cell boundary changes, determines whether the cell is in the division phase or the resting phase, calculates the changes in cell morphology within the cycle, and obtains a cell cycle stage boundary map;

[0053] The deep learning optimization module performs deep learning training based on the cell cycle phase boundary map, optimizes the layer depth and learning rate of the neural network, adjusts the learning rate and weight of each layer, optimizes the image segmentation process, and generates an optimized cell segmentation result map.

[0054] Compared with the prior art, the advantages and positive effects of the present invention are:

[0055] In the present invention, image standardization and denoising processing are used to provide more reliable basic data for subsequent analysis, reducing the impact of external interference on the analysis results. The introduction of residual networks accurately models the dynamic changes of cell morphology through deep convolution and nonlinear activation functions, ensuring that the changes in cell morphology at different stages can be captured in a timely and accurate manner. The precise classification of cell cycle stages, through periodic analysis, not only improves the precision of monitoring, but also avoids misdiagnosis or missed diagnosis problems caused by inaccurate periodic classification. Ultimately, the optimization of deep learning training improves the accuracy of image segmentation, making the judgment of cell status more scientific, and providing a more efficient and accurate monitoring method for the stem cell preparation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0057] Figure 2 Flowchart of the steps for acquiring denoised cell images of the present invention;

[0058] Figure 3 Flowchart of the steps for obtaining the cell edge feature map of the present invention;

[0059] Figure 4 Flowchart of the steps for obtaining the cell boundary interval value of the present invention;

[0060] Figure 5 A flow chart of the steps for obtaining a cell cycle phase boundary map according to the present invention;

[0061] Figure 6 This is a flow chart of the steps for obtaining the optimized segmentation result map of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0063] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0064] Example 1

[0065] See also Figure 1 The present invention provides a technical solution: a method for monitoring the preparation process of umbilical cord mesenchymal stem cells, comprising the following steps:

[0066] S1: Based on the umbilical cord mesenchymal stem cell image, image size normalization, noise removal, cell area cropping, grayscale value comparison to remove grayscale value areas below the set threshold, and filtering methods to eliminate background noise, remove noise points and extract cell edge information to generate a denoised cell image;

[0067] S2: Based on the denoised cell image, the cell image is input into the residual network, the output features of each layer are weighted and processed by convolution operation, the convolution kernel size is adjusted to match the cell morphological changes by applying a nonlinear activation function, the influence of the cell proliferation state on the morphological features is calculated, and the cell edge feature map is generated;

[0068] S3: Based on the cell edge feature map, edge change and grayscale comparison are performed to identify and extract the boundaries of the umbilical cord mesenchymal stem cells, perform region division and boundary extraction, and combine the cell morphology monitoring information to perform boundary value comparison analysis to generate cell boundary interval values;

[0069] S4: Based on the cell boundary interval value, cell cycle stage classification is performed, periodic analysis is performed through cell boundary changes, standards are set to divide the division phase and the resting phase, cell morphological changes within the cycle are calculated, and a cell cycle stage boundary map is generated;

[0070] S5: Based on the cell cycle phase boundary map, perform deep learning training, perform cycle state segmentation and optimization, adjust the neural network layer depth and learning rate, optimize the image segmentation process, adjust the learning rate and weight of each layer, and obtain the optimized segmentation result map.

[0071] The denoised cell image specifically includes a cell image, a noise removal area, and cell edge information. The cell edge feature map includes convolution features, activation function adjustment, and morphological change effects. The cell boundary interval value includes edge changes, grayscale contrast, boundary extraction, and morphological monitoring. The cell cycle stage boundary map includes the division phase, the resting phase, and cell morphological changes. The optimized segmentation result map specifically includes cycle state optimization, network hierarchy adjustment, and segmentation results.

[0072] See also Figure 2 , the steps of acquiring the denoised cell image are specifically as follows:

[0073] S101: Based on the input umbilical cord mesenchymal stem cell images, all images are resized by a unified scaling ratio. At the same time, the grayscale value of each pixel is compared with the threshold according to a set threshold, and grayscale values ​​less than the threshold are filtered out to generate a standardized image.

[0074] First, all images must be resized using a uniform scaling ratio. This scaling ratio is calculated from the ratio of the input image size to the desired output size: scaling ratio = output size / input size. In practical applications, the input image size may be arbitrary, while the output size can be set based on experimental requirements, such as setting the output to 512 × 512 pixels. Therefore, calculating the scaling ratio by the ratio of the input and output image sizes is crucial for ensuring that all input images are uniformly sized. Next, the grayscale value of each pixel in the image is compared against a set grayscale threshold, and those with grayscale values ​​below the threshold are removed. This process aims to remove pixels with low grayscale values, thereby reducing irrelevant background noise and enhancing the saliency of cellular regions. Comparing the grayscale value to the threshold can be accomplished through simple numerical calculations. A threshold T is set. If the pixel grayscale value I is less than T, the pixel value is set to 0, indicating that the pixel is filtered out; otherwise, the pixel value is retained. In practice, the selected threshold value T is determined by statistically analyzing the grayscale distribution of pixels in the image. For example, an appropriate threshold range is set based on the average or median grayscale values ​​of all pixels in the input image to ensure that the majority of valid cell regions are retained while removing irrelevant regions with lower grayscale values. When analyzing an image, if the threshold value T is set to 80, pixels with grayscale values ​​less than 80 in the image are filtered out, retaining pixels with grayscale values ​​greater than or equal to 80, thereby generating a standardized image. This process effectively filters the image, resulting in a standardized image that removes background noise.

[0075] S102: filtering the standardized image using a high-pass or low-pass filter to distinguish the differentiated noise frequencies, adjusting the filter intensity using filter design parameters, and separating the cell region from other regions through threshold determination to establish a denoised cell region;

[0076] Image filtering involves using either a high-pass or low-pass filter. The choice of filter depends on the frequency characteristics of the noise in the image. If the image contains low-frequency noise, a low-pass filter is used to remove it; conversely, if the noise is high-frequency, a high-pass filter is used. Filter selection and design must be tailored to the characteristics of the image noise. For example, a low-pass filter can smooth high-frequency noise, while a high-pass filter can highlight edge features. Filter design parameters, such as the cutoff frequency, are obtained using spectrum analysis tools based on the noise characteristics. Spectral analysis can help identify the primary frequency components in the image and determine appropriate filter parameters. This process can be performed using image processing software such as MATLAB or OpenCV in Python for analysis and filtering. The filtering strength is controlled by selecting appropriate design parameters (such as the filter's cutoff frequency or attenuation coefficient). In practice, when setting filter parameters, a series of experiments are performed to determine the optimal filtering strength. For example, suppose a low-pass filter is selected for image processing with a cutoff frequency of 0.3. This means that the filter removes signals with frequencies above 0.3 while retaining low-frequency information. In this way, high-frequency noise unrelated to cells in the image can be effectively removed, retaining the main information of the cell area. After the filtering operation is completed, a threshold is set to determine which areas belong to the cell area. This process is completed by comparing the pixels in the image with the set threshold. If the pixel value is higher than the threshold, the area is considered to be a cell area, otherwise it is excluded. The resulting denoised cell area image is obtained by combining high-pass or low-pass filtering and threshold determination.

[0077] S103: Based on the denoised cell area, the edge position is located by calculating the grayscale difference between each pixel in the image and its neighboring pixels, and the edge strength is compared using a preset threshold, using the formula:

[0078]

[0079] Calculate cell edge information and generate denoised cell images;

[0080] Among them, E represents the amount of edge information, I i represents the grayscale value of the i-th pixel in the image, α is the edge clarity coefficient, β is the edge smoothness adjustment coefficient, γ is the detail enhancement coefficient, and N is the total number of pixels in the image.

[0081] formula:

[0082]

[0083] The benefit of the formula is that by introducing grayscale value differences, edge clarity coefficients, smoothness adjustment coefficients, and detail enhancement coefficients, it can perform precise calculations on the cell edges in the image, especially remove the effects of noise, retain the clear edges of the cell area, and effectively improve the quality and details of the image.

[0084] Detailed explanation of the formula and the process of formula calculation and derivation:

[0085] E stands for edge information, which indicates the clarity and strength of the edges in the image.

[0086] I i Represents the grayscale value of the i-th pixel in the image.

[0087] α is the edge definition coefficient, which is used to adjust edge definition and is obtained from experimental data. The reasonable range is α = 1.0 to 2.0.

[0088] β is the edge smoothness adjustment coefficient, which is used to smooth the transition of the edge area. It is obtained through experimental analysis and the reasonable value range is β = 0.1 to 1.0.

[0089] γ is a detail enhancement coefficient used to enhance details in an image. Its value is set to γ ​​= 1.0 to 2.0, and is used to adjust the degree of detail prominence in an image.

[0090] N represents the total number of pixels in the image and is used to indicate the overall size of the image.

[0091] Assume that in actual operation, the grayscale value and edge features of the image have been obtained through the aforementioned normalization and filtering process, and the parameters are set as follows:

[0092] α = 1.5 (edge ​​sharpness coefficient);

[0093] β = 0.5 (edge ​​smoothness adjustment coefficient);

[0094] γ = 1.2 (detail enhancement coefficient);

[0095] Assume that the gray value difference in the image between neighboring pixels is |I i -I i-1 |=20.

[0096] Total number of pixels N = 512 × 512 = 262144 (assuming the image size is 512 × 512);

[0097] Calculate according to the formula:

[0098]

[0099] Assuming the grayscale difference of each pixel is 20, the formula is calculated as follows:

[0100]

[0101] Calculation yields:

[0102]

[0103] The calculated value of edge information E is a constant. This result shows that the edge clarity and details have been optimized through the adjustment of the formula, the edges of the image cell area have been captured more accurately, and the influence of noise on the image has been effectively suppressed.

[0104] See also Figure 3 , the steps of obtaining the cell edge feature map are specifically as follows:

[0105] S201: Based on the denoised cell image, the cell image is input into the residual network, the feature map of each layer is weighted by a convolution operation, the image is filtered multiple times by a filter, the output features of the differentiation layer are weighted, the convolution kernel size of the differentiation layer is adjusted, and the convolution output of each layer is calculated to obtain a convolution result image;

[0106] First, a denoised cell image is fed into a residual network. The input image has a fixed size, such as 512x512 pixels. After preprocessing to remove noise, the image is passed as input to the residual network. The residual network processes the image through multiple convolutional layers, where each convolution operation extracts features from the image. The key to the convolution operation is the use of a weighted convolution kernel to extract local features. For example, a 3x3 convolution kernel can be used to extract edge features, while a 5x5 convolution kernel can capture a wider range of image information. Combining different convolution kernels enhances the network's ability to process diverse details in the cell image. After the convolution operation, the network performs a weighted fusion of the feature maps generated by each convolution layer through weighted adjustment to highlight important features while suppressing unnecessary background information. During this process, the size of the convolution kernel can be adjusted based on the characteristics of the cell image. Smaller cell morphologies may require smaller convolution kernels, while larger cell morphologies are more suitable for larger convolution kernels. Through multiple filtering operations, local features of the image are gradually enhanced, while the output features of the differentiation layer are further processed to ensure the extraction of key information about the cell morphology.

[0107] S202: applying a nonlinear activation function to adjust the convolution result image, adjusting the nonlinear effect of the output feature according to the changes in cell morphology and the differentiated morphology of the cells, and obtaining a feature map that matches the cell morphology by adjusting the size of the convolution kernel;

[0108] Nonlinear activation functions are used to introduce nonlinear effects to image features to enhance the expressive power of neural networks. Commonly used nonlinear activation functions include ReLU, Sigmoid, and Tanh. ReLU is preferred due to its high computational efficiency and ability to effectively mitigate the vanishing gradient problem. After the feature map output by the convolutional layer is processed by the activation function, the image features are further extracted and enhanced. In cell image analysis, cell morphological characteristics (such as size and shape) influence the output of the convolutional layer. The effect of nonlinear activation functions varies depending on cell morphology. For example, smaller cells experience less morphological changes, resulting in a weaker nonlinear effect. However, larger cells experience more pronounced morphological changes, resulting in a more pronounced nonlinear activation effect. Based on the differentiated cell morphology, the convolution kernel size is further adjusted. The convolution kernel size adjustment process involves analyzing different cell morphologies to select an appropriate convolution kernel size to suit the characteristics of each cell. Larger cells may require a larger convolution kernel to extract internal details, while smaller cells may require a smaller convolution kernel to extract outline features. This adjustment process relies on experimental data and analysis of cell image features, testing different convolution kernel combinations to find the most optimal size configuration. In this way, the network can obtain feature maps that match the cell morphology, further improving the accuracy and effectiveness of cell image analysis.

[0109] S203: Based on the feature map that matches the cell morphology, and according to the relationship between the proliferation state and the cell morphology, analyze the changes in the cell morphological characteristics under the differential proliferation state. By calculating the proliferation rate and the amplitude of the morphological change, and combining the effect of cell proliferation on the image characteristics, adjust the feature map to reflect the changes in the cell proliferation state, using the formula:

[0110]

[0111] Calculate the effect of cell proliferation on morphological characteristics and generate cell edge feature maps;

[0112] Among them, C represents the change value of the morphological characteristic map, F i represents the gray value of the i-th pixel in the image, α is the coefficient of morphological change, β is the reference coefficient of morphological change, γ is the influence coefficient of proliferation rate, δ is the adjustment factor, and M is the total number of pixels in the image.

[0113] formula:

[0114]

[0115] The benefit of the formula is that by introducing multiple parameters such as the morphological change coefficient, reference coefficient, proliferation rate influence coefficient and adjustment factor, it can accurately quantify the impact of cell proliferation on morphological characteristics, and combine the changes in the morphological characteristic diagram to improve the accuracy of cell proliferation status analysis.

[0116] Detailed explanation of the formula and the process of formula calculation and derivation:

[0117] C represents the change value of the morphological characteristic map.

[0118] F i Represents the grayscale value of the i-th pixel in the image, F i α Represents the grayscale value of the morphological change of the i-th pixel, Represents the reference coefficient grayscale value of the i-1th pixel.

[0119] α is the coefficient of morphological change, β is the reference coefficient of morphological change, γ is the influence coefficient of proliferation rate, and δ is the adjustment factor.

[0120] M is the total number of pixels in the image. The formula calculates the difference between the grayscale change of each pixel and the grayscale reference value of the previous pixel and adjusts this difference to reflect the impact of proliferation rate and morphological changes on image features. To obtain actual parameter values, measurements based on experimental data are required:

[0121] F i and F i-1 The value is obtained through image acquisition and pixel grayscale analysis. The cell image is grayscaled using image processing software or programming tools (such as OpenCV) and the grayscale value of each pixel is extracted.

[0122] The morphological change coefficient α can be obtained by analyzing the relationship between cell morphological changes and proliferation status, for example, by comparing changes in cell images at different time points and fitting using statistical methods (such as the least squares method).

[0123] The proliferation rate coefficient γ and adjustment factor δ need to be calculated based on the proliferation experimental data and the relationship between cell division rate and image feature changes. The proliferation rate can be obtained through time series analysis, for example, by recording the cell division interval and calculating the average proliferation rate. Assume that the following specific values ​​are obtained through experiment:

[0124] F i =120,F i-1 =100(pixel grayscale value);

[0125] α=1.2, β=0.8, γ=0.5, δ=2;

[0126] The total number of image pixels M = 1000;

[0127] Calculation derivation:

[0128]

[0129] 120 1.2 ≈177.83,100 0.8 ≈63.10,|177.83-63.10|=114.73,|120| 0.5 =10.95, substituting into the formula we get:

[0130]

[0131] C≈1000×58.97=58970;

[0132] The results showed that the change value of the cell morphology characteristic map was 58970, reflecting the significance of cell proliferation status and morphological changes.

[0133] See also Figure 4 , the steps for obtaining the cell boundary interval value are specifically as follows:

[0134] S301: extracting the grayscale gradient value of the cell edge change based on the cell edge feature map, and screening the pixel points that meet the cell boundary characteristics by comparing the grayscale gradient value with the target threshold to generate a cell edge pixel feature set;

[0135] Based on the cell edge feature map, the grayscale gradient value of the edge change is extracted by analyzing the grayscale change characteristics in the image, the grayscale distribution information of specific pixels in the image is called, the fluctuation range of the grayscale change is compared with the set threshold, and the pixels whose grayscale change gradient exceeds the threshold are screened out for each area. The screened pixels are normalized according to the grayscale gradient value, and the grayscale distribution mean of the overall edge and the distribution of local outliers are extracted through statistical analysis. The above processing results are combined to form a pixel point set of grayscale changes at the cell boundary and the pixels are marked and stored to generate a cell edge pixel feature set.

[0136] S302: calling the cell edge pixel feature set, calculating the continuity value of the edge change based on the directionality of the edge change and the grayscale distribution uniformity index, performing region division, and generating a region boundary feature set;

[0137] The cell edge pixel feature set is called, the directional characteristics of grayscale changes in the pixel point set are analyzed, the grayscale change trend between pixels is calculated using a differential operator, the main boundary direction is extracted based on the directional characteristics of grayscale changes, the uniformity of grayscale changes in the boundary direction is analyzed, and the image is divided into regions based on the grayscale distribution mean and directional characteristics. By calculating the continuity of the boundary characteristics of each divided region, the change intensity of the boundary pixel points is associated with the grayscale distribution to form boundary parameters with directional and regional characteristics. The boundary direction and grayscale change characteristic information are stored to generate a regional boundary feature set.

[0138] S303: Call the region boundary feature set, calculate the difference between the boundary feature parameters, and use the formula:

[0139]

[0140] Calculate the change intensity of the region boundary, compare and extract the maximum change intensity value, and generate the cell boundary interval value;

[0141] Among them, B i Indicates the boundary interval value, G j Represents the grayscale value of the j-th boundary pixel, G j+1 Indicates the gray value of the adjacent pixel, D j Represents the distance between pixels, is the average value of all distances, α is the adjustment coefficient of grayscale change, β is the smoothing parameter of the region boundary feature set, and n is the number of boundary pixels.

[0142] formula:

[0143]

[0144] The benefit of the formula is that by introducing the cumulative amount of boundary grayscale difference and the standard deviation of pixel distance, combined with smoothing parameters and adjustment coefficients, the sensitivity and robustness to boundary change intensity are enhanced.

[0145] Detailed explanation of the formula and the process of formula calculation and derivation:

[0146] 1. Set the number of pixels in the region boundary to n = 5, and the grayscale value sequence to G1 = 50, G2 =

[0147] 60, G3=55, G4=70, G5=65, the distance sequence is D1=2, D2=3, D3=2.5, D4=

[0148] 3.5,D5=2.8.

[0149] 2. Calculate the absolute value accumulation of grayscale difference:

[0150]

[0151] 3. Calculate the distance standard deviation:

[0152]

[0153] 4. Substitute the adjustment coefficient α = 1.2 and the smoothing parameter β = 0.8 for calculation:

[0154]

[0155] The results show that the change intensity value of the region boundary is 6.19, which is related to the grayscale gradient value and pixel distance characteristics. This result is used to determine the most significant cell boundary in the region and generate the cell boundary interval value.

[0156] See also Figure 5 The steps for obtaining the cell cycle phase boundary map are specifically as follows:

[0157] S401: Based on the cell boundary interval value, by cell cycle stage classification, periodic analysis of cell boundary changes is performed, difference values ​​of boundary changes within the cell cycle are calculated, standard divisions between the division phase and the resting phase are set, and a cell cycle stage boundary map is generated;

[0158] Cell boundary interval values ​​are obtained by analyzing edge features in cell images. By calculating pixel grayscale values ​​and edge variations in the image, boundary change values ​​are generated and region demarcation is performed. First, a region of interest is selected and the cell image is preprocessed, such as by removing noise and smoothing. Cell boundaries are then identified using edge detection algorithms such as the Canny algorithm. Cell boundary information is obtained by analyzing grayscale gradient changes. The image is then divided into regions, and the entropy of each interval is calculated to ultimately determine the cell boundary values ​​for each interval, further demarcating the cell boundary intervals. This boundary value demarcation provides an important data foundation for further analysis of the cell cycle.

[0159] S402: calling the cell cycle phase boundary map, combining the temporal characteristics of boundary changes, calculating the cell morphological changes in the differentiation phase, analyzing the magnitude of changes in multiple phases within the cycle, and generating a cell boundary distribution map for the differentiation phase within the cycle;

[0160] The cell cycle phase boundary map reflects the morphological changes of cells by analyzing the changes in cell boundaries at different cycle stages. First, the cell boundary interval values ​​are used for periodic analysis, and combined with image time series data, the changes in cell boundaries at different cycle stages are extracted. To ensure the accuracy of the cycle division, time series analysis methods such as sliding window analysis and weighted average method are used to further calculate the boundary values ​​and morphological changes of cells at different stages. The changes in each stage can be quantified by analyzing characteristics such as cell aspect ratio, area, and boundary contour. Using these data, a cell cycle phase boundary map is finally generated, which can visualize the morphological changes of cells at each stage, thereby supporting periodic analysis and other biological research.

[0161] S403: In the cell boundary distribution map at the differentiated stage within the cycle, the formula:

[0162]

[0163] Calculate the boundary change intensity of multiple cell cycle stages, optimize the division of multiple stages within the cycle, and generate a cell cycle stage boundary map;

[0164] Among them, C s Indicates the intensity of the boundary change of the cycle stage, P k represents the boundary value of the k-th cell stage, Y k represents the boundary position of the kth stage, Y k+1 represents the boundary position between adjacent stages, is the mean of the boundary position, β is the adjustment coefficient, γ t is the time adjustment coefficient, and m is the total number of stages in the cell cycle.

[0165] formula:

[0166]

[0167] The benefit of the formula is that by introducing the boundary variation strength (P k ), time adjustment coefficient (γ t ) and the mean of the boundary positions The calculation of the cell cycle phase boundaries can more accurately identify the intensity of changes and optimize the division of cell cycle phases.

[0168] Detailed explanation of the formula and the process of formula calculation and derivation:

[0169] C in the formula s It indicates the intensity of the boundary change of the cycle phase, reflecting the amplitude of the boundary change of each phase in the cell cycle. k represents the boundary value of the kth cell cycle stage, Y k represents the boundary position of the kth stage, Yk+1 represents the boundary position between adjacent stages, is the mean of the boundary positions of all phases in the cell cycle, γ t is the time adjustment coefficient, β is the adjustment coefficient, and m is the total number of stages in the cell cycle.

[0170] Assume that there are five stages in the cell cycle, that is, m = 5, the boundary positions Y1, Y2, Y3, Y4, and Y5 of each stage are: 0.5, 0.6, 0.7, 0.8, and 0.9 respectively, and the boundary values ​​P1, P2, P3, P4, and P5 are: 1.0, 1.2, 1.5, 1.3, and 1.1 respectively. The time adjustment coefficient γ t =0.02, adjustment coefficient β=0.1.

[0171] First, calculate the absolute value of the boundary position change:

[0172] |Y1-Y2|=|0.5-0.6|=0.1, |Y2-Y3|=|0.6-0.7=0.1, |Y3-Y4|=|0.7-0.8|=0.1, Y4-Y5|=|0.8-0.9|=0.1;

[0173] Then, the weighted sum of each stage is calculated:

[0174]

[0175] Compute the mean at the boundary locations:

[0176]

[0177] Next, calculate the difference between each boundary and the mean:

[0178]

[0179] Finally, enter the formula for calculation:

[0180]

[0181] The results show that the intensity of cycle stage boundary change is 1.525, indicating that the amplitude of cell boundary change within this cycle is moderate, which can help further optimize the cycle stage division and support the accuracy of cell cycle analysis.

[0182] See also Figure 6 , the steps for obtaining the optimized segmentation result map are specifically as follows:

[0183] S501: Based on the cell cycle phase boundary map, performing cycle state segmentation and optimization through deep learning training, combining image data and neural network model, adjusting the neural network layer depth and learning rate, and generating an optimized neural network model;

[0184] When deep learning training and cycle state segmentation based on cell cycle phase boundary maps is performed, the input image data is first preprocessed and resized to the network's required input size. Cell images are typically grayscale normalized to ensure pixel values ​​are within a certain range, remove noise, and enhance image quality. Next, the preprocessed image data is fed into the initial layer of the neural network, where image features are extracted through convolutional layers. Each layer in the network is responsible for extracting features at different levels, such as cell edges and shapes. During training, labeled cell cycle phase data is used as the training set. Through iterative optimization, the neural network learns to identify and classify different cell cycle phases from images. To improve network training efficiency, batch normalization is often used to accelerate network convergence and reduce the risk of overfitting. After training, validation is performed on a validation set, and model hyperparameters, such as the learning rate and weight initialization method, are adjusted to improve generalization and accuracy. After training, the neural network is able to generate an optimized neural network model and is ready to enter the next stage of cell cycle classification and image segmentation tasks.

[0185] S502: calling the optimized neural network model, adjusting the learning rate and weight of each layer, performing back propagation and optimizing the image segmentation process, calculating the gradient value of the weight of each layer, and generating an optimized segmentation strategy;

[0186] The neural network model is optimized by adjusting the learning rate and weights of each layer. First, an appropriate learning rate adjustment strategy is selected based on the current model's performance on the validation set. A high learning rate may cause model oscillation or failure to converge, in which case the learning rate should be reduced. A low learning rate, on the other hand, will slow training and prevent effective model optimization. Next, the weights of each layer are updated using the backpropagation algorithm. Backpropagation calculates the error gradient of each layer, propagating the error from the output layer back to the input layer, and then calculates the gradient of each layer to update the weights of each layer. During this process, the learning rate must be adjusted to prevent exploding or vanishing gradients from occurring during the optimization process. Momentum is often used to further accelerate training. As neural network training progresses, the weights of different layers are continuously adjusted after each iteration to further improve image segmentation performance. Each training result is used for validation and fine-tuning to improve model accuracy until optimal performance is achieved. This generates an optimized segmentation strategy for subsequent cell cycle image segmentation tasks.

[0187] S503: In the application of the optimized segmentation strategy, the formula is used:

[0188]

[0189] Calculate the error function of each neural network layer and adjust the parameters to generate the optimized segmentation result map;

[0190] Among them, L represents the loss function value, Z i represents the true value of the i-th image pixel, represents the predicted value, λ is the regularization coefficient, α j is the learning rate adjustment coefficient of layer j, W j represents the weight value of the jth layer, n is the total number of pixels, and m is the number of network layers.

[0191] In the application of optimized segmentation strategy, the formula is used:

[0192]

[0193] The benefit of the formula is that it controls the scale of network weights by introducing regularization terms, avoids overfitting, and considers the impact of weight values ​​on the loss function when updating weights through backpropagation. Detailed explanation of the formula and the derivation process of the formula calculation:

[0194] 1. L represents the loss function value, which evaluates the accuracy of the network output by calculating the difference between the predicted value and the true value (i.e., the sum of squared errors);

[0195] 2.Z i represents the true value of the i-th image pixel, is the pixel value predicted by the model, is the prediction error of the i-th pixel, and the total error is obtained by taking the square and summing it up.

[0196] 3.λ is the regularization coefficient, which is used to balance the weight between the error term and the weight regularization term.

[0197] 4.α j is the learning rate adjustment coefficient of the jth layer, which controls the learning speed of each layer, W j represents the weight value of the jth layer, Represents the square term of the weight, reflecting the contribution of the weight to the loss function.

[0198] 5. This is the regularization of the weight term. By adding this term, the model can avoid excessive weights during update and avoid overfitting problems.

[0199] Specific calculation example:

[0200] Assume there are 3 pixels, Z1=3, Z2=4, Z3=5, and the corresponding prediction values ​​are And with the regularization coefficient λ = 0.1, the learning rate adjustment coefficients α1 = 0.5 and α2 = 0.3, the first layer weight value W1 = 0.2 and the second layer weight value W2 = 0.1, we can calculate the loss function value.

[0201] Calculate the sum of squared errors:

[0202]

[0203] So the sum of squared errors is:

[0204]

[0205] Calculate the regularization part:

[0206]

[0207] So the regularization part is:

[0208]

[0209] Comprehensive loss function:

[0210] L=0.3+0.1·0.023=0.3+0.0023=0.3023;

[0211] The result shows that the loss function value is 0.3023, indicating that under the influence of the error of the current model and the regularization term, the fitting effect of the current model is relatively ideal, but the model weights still need to be further optimized.

[0212] An umbilical cord mesenchymal stem cell production process monitoring system is used to implement the above-mentioned umbilical cord mesenchymal stem cell production process monitoring method, and the system includes:

[0213] The image processing module normalizes the image size, adjusts the aspect ratio of the image, performs noise removal, removes background noise, crops the cell area, removes areas with grayscale values ​​below a set threshold, eliminates noise in the image through grayscale value contrast operations, extracts cell edge information, and generates a denoised cell image.

[0214] The feature extraction module is based on the denoised cell image, inputs the image into the residual network, processes the output features through convolution operations, adjusts the convolution kernel size to match the cell morphology changes, calculates the impact of the proliferation state on the cell morphology, extracts the features of the cell edge, and generates a cell edge feature map;

[0215] The boundary recognition module performs edge change and grayscale comparison based on the cell edge feature map, identifies the boundary area between the cell and the background, extracts the edge position of the cell, performs region division and boundary extraction, and combines the cell morphology information to perform boundary value comparison and analysis to obtain the cell boundary interval value;

[0216] The cycle analysis module classifies the cell cycle stages based on the cell boundary interval values, analyzes the periodicity through cell boundary changes, determines whether the cell is in the division phase or the resting phase, calculates the changes in cell morphology within the cycle, and obtains a cell cycle stage boundary map;

[0217] The deep learning optimization module performs deep learning training based on the cell cycle phase boundary map, optimizes the layer depth and learning rate of the neural network, adjusts the learning rate and weight of each layer, optimizes the image segmentation process, and generates an optimized cell segmentation result map.

[0218] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for monitoring the preparation process of umbilical cord mesenchymal stem cells, characterized in that: The following steps are involved: S1: Based on the umbilical cord mesenchymal stem cell image, image size normalization, noise removal, cell area cropping, grayscale value comparison to remove grayscale value areas below the set threshold, and filtering methods to eliminate background noise, remove noise points and extract cell edge information to generate a denoised cell image; S2: Based on the denoised cell image, the cell image is input into the residual network, the output features of each layer are weighted and processed by convolution operation, the convolution kernel size is adjusted to match the cell morphological changes by applying a nonlinear activation function, the influence of the cell proliferation state on the morphological features is calculated, and the cell edge feature map is generated; S3: Based on the cell edge feature map, edge change and grayscale comparison are performed to identify and extract the boundaries of the umbilical cord mesenchymal stem cells, perform region division and boundary extraction, and combine the cell morphology monitoring information to perform boundary value comparison analysis to generate cell boundary interval values; S4: Based on the cell boundary interval value, cell cycle stage classification is performed, periodic analysis is performed through cell boundary changes, standards are set to divide the division phase and the resting phase, cell morphological changes within the cycle are calculated, and a cell cycle stage boundary map is generated; S5: Based on the cell cycle phase boundary map, perform deep learning training, perform cycle state segmentation and optimization, adjust the neural network layer depth and learning rate, optimize the image segmentation process, adjust the learning rate and weight of each layer, and obtain an optimized segmentation result map; The steps for obtaining the cell boundary interval value are specifically as follows: S301: extracting the grayscale gradient value of the cell edge change based on the cell edge feature map, and screening the pixel points that meet the cell boundary characteristics by comparing the grayscale gradient value with the target threshold to generate a cell edge pixel feature set; S302: calling the cell edge pixel feature set, calculating the continuity value of the edge change based on the directionality of the edge change and the grayscale distribution uniformity index, performing region division, and generating a region boundary feature set; S303: Call the region boundary feature set, calculate the difference between the boundary feature parameters, and use the formula: ； Calculate the change intensity of the region boundary, compare and extract the maximum change intensity value, and generate the cell boundary interval value; in, Indicates the boundary interval value, Indicates the The grayscale value of the boundary pixel, Represents the grayscale value of the adjacent pixel. Represents the distance between pixels, is the average of all distances, is the adjustment coefficient of grayscale change, is the smoothing parameter of the region boundary feature set, is the number of boundary pixels; The steps for obtaining the cell cycle phase boundary map are specifically as follows: S401: Based on the cell boundary interval value, by cell cycle stage classification, periodic analysis of cell boundary changes is performed, difference values ​​of boundary changes within the cell cycle are calculated, standard divisions between the division phase and the resting phase are set, and a cell cycle stage boundary map is generated; S402: calling the cell cycle phase boundary map, combining the temporal characteristics of boundary changes, calculating the cell morphological changes in the differentiation phase, analyzing the magnitude of changes in multiple phases within the cycle, and generating a cell boundary distribution map for the differentiation phase within the cycle; S403: In the cell boundary distribution map at the differentiation stage within the cycle, the formula is used: ； Calculate the boundary change intensity of multiple cell cycle stages, optimize the division of multiple stages within the cycle, and generate a cell cycle stage boundary map; in, represents the intensity of the boundary change of the cycle stage, Indicates the The boundary value of the cell stage, Indicates the The boundary position of each stage, represents the boundary position between adjacent stages, is the mean value at the boundary position, is the adjustment coefficient, is the time adjustment coefficient, is the total number of stages in the cell cycle.

2. The method for monitoring the preparation process of umbilical cord mesenchymal stem cells according to claim 1, characterized in that: The denoised cell image specifically includes a cell image, a noise removal area, and cell edge information. The cell edge feature map includes convolution features, activation function adjustment, and morphological change effects. The cell boundary interval value includes edge changes, grayscale contrast, boundary extraction, and morphological monitoring. The cell cycle stage boundary map includes the division phase, the resting phase, and cell morphological changes. The optimized segmentation result map specifically includes cycle state optimization, network hierarchy adjustment, and segmentation results.

3. The method for monitoring the preparation process of umbilical cord mesenchymal stem cells according to claim 2, characterized in that: The steps for acquiring the denoised cell image are specifically as follows: S101: Based on the input umbilical cord mesenchymal stem cell images, all images are resized by a unified scaling ratio. At the same time, the grayscale value of each pixel is compared with the threshold according to a set threshold, and grayscale values ​​less than the threshold are filtered out to generate a standardized image. S102: filtering the standardized image using a high-pass or low-pass filter to distinguish the differentiated noise frequencies, adjusting the filter intensity using filter design parameters, and separating the cell region from other regions through threshold determination to establish a denoised cell region; S103: Based on the denoised cell area, the edge position is located by calculating the grayscale difference between each pixel in the image and its neighboring pixels, and the edge strength is compared using a preset threshold, using the formula: ； Calculate cell edge information and generate denoised cell images; in, represents the amount of edge information, Represents the image The grayscale value of the pixel, is the edge definition coefficient, is the edge smoothness adjustment coefficient, is the detail enhancement coefficient, is the total number of pixels in the image.

4. The method for monitoring the preparation process of umbilical cord mesenchymal stem cells according to claim 3, characterized in that: The steps for obtaining the cell edge feature map are specifically as follows: S201: Based on the denoised cell image, the cell image is input into the residual network, the feature map of each layer is weighted by a convolution operation, the image is filtered multiple times by a filter, the output features of the differentiation layer are weighted, the convolution kernel size of the differentiation layer is adjusted, and the convolution output of each layer is calculated to obtain a convolution result image; S202: applying a nonlinear activation function to adjust the convolution result image, adjusting the nonlinear effect of the output feature according to the changes in cell morphology and the differentiated morphology of the cells, and obtaining a feature map that matches the cell morphology by adjusting the size of the convolution kernel; S203: Based on the feature map that matches the cell morphology, and according to the relationship between the proliferation state and the cell morphology, analyze the changes in the cell morphological characteristics under the differential proliferation state. By calculating the proliferation rate and the amplitude of the morphological change, and combining the effect of cell proliferation on the image characteristics, adjust the feature map to reflect the changes in the cell proliferation state, using the formula: ； Calculate the effect of cell proliferation on morphological characteristics and generate cell edge feature maps; in, Represents the change value of the morphological feature map, Represents the image The grayscale value of the pixel, is the coefficient of morphological change, is the reference coefficient of morphological change, is the influence coefficient of the proliferation rate, is the adjustment factor, is the total number of pixels in the image.

5. The method for monitoring the preparation process of umbilical cord mesenchymal stem cells according to claim 4, characterized in that: The steps for obtaining the optimized segmentation result map are specifically as follows: S501: Based on the cell cycle phase boundary map, performing cycle state segmentation and optimization through deep learning training, combining image data and neural network model, adjusting the neural network layer depth and learning rate, and generating an optimized neural network model; S502: calling the optimized neural network model, adjusting the learning rate and weight of each layer, performing back propagation and optimizing the image segmentation process, calculating the gradient value of the weight of each layer, and generating an optimized segmentation strategy; S503: In the application of the optimized segmentation strategy, the formula is used: ； Calculate the error function of each neural network layer and adjust the parameters to generate the optimized segmentation result map; in, represents the loss function value, Indicates the The true value of the image pixels, represents the predicted value, is the regularization coefficient, For layer The learning rate adjustment coefficient, Indicates the The weight value of the layer, is the total number of pixels, is the number of network layers.

6. Umbilical cord mesenchymal stem cell preparation process monitoring system, characterized in that: The method for monitoring the preparation process of umbilical cord mesenchymal stem cells according to any one of claims 1 to 5, wherein the system comprises: The image processing module normalizes the image size, adjusts the aspect ratio of the image, performs noise removal, removes background noise, crops the cell area, removes areas with grayscale values ​​below a set threshold, eliminates noise in the image through grayscale value contrast operations, extracts cell edge information, and generates a denoised cell image. The feature extraction module is based on the denoised cell image, inputs the image into the residual network, processes the output features through convolution operations, adjusts the convolution kernel size to match the cell morphology changes, calculates the impact of the proliferation state on the cell morphology, extracts the features of the cell edge, and generates a cell edge feature map; The boundary recognition module performs edge change and grayscale comparison based on the cell edge feature map, identifies the boundary area between the cell and the background, extracts the edge position of the cell, performs region division and boundary extraction, and combines the cell morphology information to perform boundary value comparison and analysis to obtain the cell boundary interval value; The cycle analysis module classifies the cell cycle stages based on the cell boundary interval values, analyzes the periodicity through cell boundary changes, determines whether the cell is in the division phase or the resting phase, calculates the changes in cell morphology within the cycle, and obtains a cell cycle stage boundary map; The deep learning optimization module performs deep learning training based on the cell cycle phase boundary map, optimizes the layer depth and learning rate of the neural network, adjusts the learning rate and weight of each layer, optimizes the image segmentation process, and generates an optimized cell segmentation result map.

Images