Cell growth state detection method and system based on image processing

Abstract

The invention relates to the technical field of cell image processing, in particular to a cell growth state detection method and system based on image processing. The method comprises the following steps: acquiring a suspected cell region degree of each pixel point in a grayscale image of a cell image; clustering the pixel points to obtain expanded boundary pixel points of class clusters, and then obtaining the confidence coefficient of the pixel points of each class cluster; multiplying the suspected cell region degree by the confidence coefficient, and normalizing to obtain the importance degree; analyzing the gray scale conditions of the pixel points in eight neighborhoods of one pixel point to obtain the gray scale change degree of the pixel point; acquiring adjacent pixel points of one pixel point, and analyzing to obtain the continuity degree of the pixel point; the enhancement coefficient of each pixel point is optimized according to the importance degree, the gray level change degree and the continuity degree of each pixel point, and then image enhancement is carried out; and carrying out cell growth state detection by utilizing the enhanced cell image. According to the invention, the accuracy of cell growth state detection can be improved.

Description

Technical Field

[0001] This invention relates to the field of cell image processing technology, and specifically to a method and system for detecting cell growth status based on image processing. Background Technology

[0002] Accurate detection of cell growth status is crucial for research and clinical applications during cell culture. Traditional methods primarily rely on manual microscopy and manual sampling combined with flow cytometry. Manual microscopy depends on the observer's experience and judgment, is highly subjective, prone to errors, and can lead to fatigue over extended periods, further affecting the accuracy of results. Manual sampling and counting can interfere with the cell culture environment, increasing the risk of contamination, and is also cumbersome and inefficient. With the development of computer vision technology, image processing techniques can now be used to detect cell growth status. This involves enhancing acquired cell images to preserve cell region details, acquiring and training a large number of cells at different growth stages to analyze their morphology, texture, and other features, and utilizing machine learning to detect cell growth status. During cell culture, as cells grow, cell stacking may occur in the culture dish due to the increasing number of cells. Because high-powered microscopes have limited depth of field, some structures may become blurred when cells are not on the same focal plane. Therefore, image enhancement processing is necessary when analyzing cell growth images to detect growth status.

[0003] Traditional image enhancement methods (such as linear enhancement and histogram equalization) often adjust the grayscale range of the entire image, which may lead to over-enhancement of low-frequency information (such as the background) while obscuring high-frequency details (such as cell edges). The Laplace operator, however, primarily acts on high-frequency components and has less impact on low-frequency components. Therefore, it can enhance cell edges and details while preserving background information and avoiding interference from background noise. This approach uses the Laplace transform for image enhancement. After obtaining the calculation results for all pixels in the image using the Laplace operator, the Laplace image of the original image is obtained. Then, the Laplace image is superimposed on the original image with a set enhancement coefficient to produce a sharpened image. Traditional methods typically use a fixed enhancement coefficient superimposed on the original image, which can lead to the enhancement of unwanted interference parts (such as impurities in a culture dish), thus interfering with subsequent detection of cell growth status and affecting the judgment of cell growth status. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention aims to provide a method and system for detecting cell growth status based on image processing, the specific technical solution of which is as follows: In a first aspect, one embodiment of the present invention provides a cell growth state detection method based on image processing, the method comprising: The acquired cell images are preprocessed to obtain grayscale images; windows of different scales are established with a pixel as the center, and the degree of suspected cell region of the pixel is obtained based on the grayscale value changes in each window; Cluster the pixels to obtain clusters; expand the boundaries of each cluster to obtain expanded boundary pixels; obtain the confidence score of a cluster's pixels based on its neighboring clusters, the pixels at the boundary, and the expanded boundary pixels; multiply the suspected cell region degree and the confidence score and normalize to obtain the importance score. The degree of grayscale variation of a pixel is obtained by normalizing the absolute value of the difference between the mean grayscale value of the pixels in the eight neighborhoods of a pixel and the grayscale value of the pixel itself. The pixel with the smallest grayscale difference from a pixel within its eight-neighborhood is taken as the neighboring pixel. The continuity of a pixel is obtained based on the difference in grayscale change between a pixel and its neighboring pixels. The enhancement coefficients of each pixel are optimized based on their importance, grayscale variation, and continuity to obtain the final enhancement coefficients, which are then used to enhance the cell image. The enhanced cell image is then used to detect cell growth status.

[0005] Preferably, the degree of suspected cell region of a pixel is obtained based on the grayscale value changes within each window, including: The product of the standard deviation and entropy of the gray values ​​of each pixel within a window corresponding to a given pixel is obtained and the average value is calculated to determine the degree of suspected cell region for that pixel.

[0006] Preferably, the pixels are clustered to obtain clusters, including: The feature vector of a pixel is composed of the degree of suspected cell region and its coordinates. The pixels are then clustered based on the difference in feature vectors between any two pixels to obtain clusters.

[0007] Preferably, the boundaries of various clusters are expanded to obtain expanded boundary pixels, including: Using one pixel as the standard step size, the original boundaries of each type of cluster are expanded outward to obtain the expanded boundaries of each type of cluster. The pixels within the expanded boundaries are the expanded boundary pixels.

[0008] Preferably, obtaining the confidence level of a cluster's pixels based on its neighboring clusters, boundary pixels, and extended boundary pixels includes: The distance between the center of a cluster and the centers of other clusters is obtained and denoted as the first distance. The clusters with the smallest first distance among the other clusters are the neighboring clusters of the given cluster. Obtain the absolute value of the difference in the number of pixels between a cluster and its corresponding neighboring cluster, and the reciprocal of the product of the first distance, as the inter-cluster feature corresponding to the neighboring cluster; calculate the absolute value of the difference between the mean gray value of the pixels at the original boundary of a cluster and the mean gray value of the pixels at the expanded boundary of the cluster, and multiply it by the reciprocal of the standard deviation of the expanded boundary pixels to obtain the boundary feature of the cluster; sum the inter-cluster features corresponding to each neighboring cluster of a cluster and multiply them by the boundary feature to obtain the confidence level of the pixels of the cluster.

[0009] Preferably, obtaining the continuity of a pixel based on the difference in grayscale variation between a pixel and its neighboring pixels includes: The continuity of a pixel is obtained by normalizing the absolute value of the difference between the grayscale changes of a pixel and its neighboring pixels.

[0010] Preferably, the enhancement coefficient of each pixel is optimized according to its importance, grayscale variation, and continuity, including: Set a threshold for grayscale change. If the grayscale change of a pixel is greater than the threshold, it is marked as a pixel with high grayscale change; otherwise, it is marked as a pixel with low grayscale change. For a pixel with high grayscale variation, the enhancement coefficient optimization sensitivity coefficient is multiplied by the degree of grayscale variation of the pixel to obtain the reduction step size of the pixel with high grayscale variation. The difference between the initial enhancement coefficient and the reduction step size is the final enhancement coefficient of the pixel with high grayscale variation. For a pixel with low grayscale variation, the product of the enhancement coefficient optimization sensitivity coefficient, the importance of the pixel, and the continuity of the pixel is the increment step size. The sum of the initial enhancement coefficient and the increment step size is the final enhancement coefficient of the pixel with low grayscale variation.

[0011] Secondly, the present invention also provides a cell growth state detection system based on image processing, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of a cell growth state detection method based on image processing.

[0012] The embodiments of the present invention have at least the following beneficial effects: This application acquires grayscale images of cell images, establishes windows of different scales centered on a single pixel, and analyzes the changes in grayscale values ​​within different windows to obtain the degree of suspected cell regions of the pixel. Using windows of different scales for analysis avoids the problems of windows that are too small to fully capture uneven grayscale features and windows that are too large to contain too many features and obscure the features of the target pixel itself, making the analysis more accurate. Furthermore, the pixels are clustered to obtain expanded boundary pixels of a cluster. Analysis is performed on the boundary features and the features between locally adjacent clusters to obtain the confidence level of the cluster pixels. The degree of suspected cell regions and confidence level are then combined to improve the accuracy of importance analysis. Next, the gray-level change and continuity of pixels are obtained, and the enhancement coefficient of each pixel is optimized based on the importance of each pixel to obtain the final enhancement coefficient of each pixel. When the Laplace operator is used for image enhancement and the Laplace image is superimposed on the original image, the edges and details of the cell regions are appropriately enhanced (weak edges are further enhanced to highlight details; strong edges are appropriately enhanced to eliminate artifacts). In the cell image, the edges and texture details of the cell regions are more prominent, thus enabling a more intuitive analysis and detection of cell growth status. Attached Figure Description

[0013] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart of a cell growth state detection method based on image processing provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating edge enlargement in an image processing-based cell growth state detection method provided in an embodiment of the present invention. Detailed Implementation

[0015] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a cell growth state detection method and system based on image processing proposed in accordance with the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0016] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0017] The following description, in conjunction with the accompanying drawings, details a specific scheme for a cell growth state detection method based on image processing provided by the present invention.

[0018] Example 1: The main application scenario of this invention is: for cell images, the Laplace operator is used to enhance them, improve the image quality, and make the cell region edges and texture details more prominent, so that the cell growth status can be analyzed and detected more intuitively.

[0019] Please see Figure 1 The diagram illustrates a flowchart of a cell growth state detection method based on image processing according to an embodiment of the present invention. The method includes the following steps: Step S1: Preprocess the acquired cell image to obtain a grayscale image; establish windows of different scales centered on a pixel, and obtain the degree of suspected cell region of the pixel based on the grayscale value changes in each window.

[0020] A high-definition camera is installed at a suitable location in the cell preparation device to capture images of the cell growth status under a microscope at various stages. In other words, the cells in the culture device are photographed at regular intervals according to the cell growth cycle to obtain cell images.

[0021] The acquired cell image is first converted to grayscale. Then, to improve the subsequent Laplace transform enhancement effect, Gaussian filtering is applied to eliminate high-frequency noise, resulting in a grayscale image of the cell image, which is then further processed.

[0022] Because organelles such as mitochondria, endoplasmic reticulum, Golgi apparatus, and lysosomes are distributed in different regions of the cell, and the density and composition of these organelles differ from those of the cytoplasm, differences in electronic or optical density occur in local areas, resulting in uneven grayscale in microscopic imaging. Therefore, the degree of suspected cellular region is determined by the uniformity of local grayscale distribution of each pixel.

[0023] After acquiring the grayscale image, due to the varying sizes of organelles within cells, a window that is too small cannot fully capture the uneven grayscale features, while a window that is too large contains too many features and will obscure the features of the target pixel itself. Therefore, windows of different scales are established with any pixel in the image as the center. In this invention, three scale windows are established, namely the first scale window, the second scale window, and the third scale window, with a window scale of n*n, where n=3, 5, and 7. The implementer can determine the window scale according to the actual situation, acquire the grayscale value of each pixel within the window, and then calculate the corresponding grayscale distribution uniformity as the degree of suspected cell region for each pixel.

[0024] Furthermore, the degree of suspected cell region of a pixel is obtained based on the gray value changes within each window. Specifically, the product of the standard deviation and entropy of the gray value of each pixel within each window corresponding to a pixel is obtained and the average value is calculated to obtain the degree of suspected cell region of that pixel.

[0025] The specific calculation model is as follows: , Where Yi represents the degree of the suspected cell region at the i-th pixel. The standard deviation of the gray value in the k-th window of the i-th pixel is the degree of fluctuation. The greater the degree of gray value fluctuation, the greater the gray value fluctuation in the window, that is, the worse the local uniformity, and the more it conforms to the gray value distribution characteristics in the cell region. is the entropy value of the gray value of the pixel in the k-th window of the i-th pixel. The larger the gray entropy, the more complex the gray information contained in the window, that is, the worse the local gray consistency, and the more it conforms to the characteristic of poor gray distribution uniformity in the cell region. This represents the number of windows of different scales corresponding to a single pixel; in this embodiment, it is 3. Yi represents the degree of suspected cell region for the i-th pixel, which is the average uniformity of grayscale distribution across different scale windows.

[0026] Step S2: Cluster the pixels to obtain clusters; expand the boundaries of each cluster to obtain expanded boundary pixels; obtain the confidence score of a cluster's pixels based on its neighboring clusters, the pixels at the boundary, and the expanded boundary pixels; multiply the suspected cell region degree and the confidence score and normalize to obtain the importance.

[0027] Step S1 obtains the degree of suspected cell region for each pixel. During the growth process, cells will form sediment in the nutrient solution for various reasons, and the sediment area will also have uneven gray level. If it is over-enhanced, it will interfere with cell texture detection. Therefore, it is also necessary to combine the possibility that each pixel belongs to the cell region to obtain its importance, and then adjust the enhancement coefficient when using the Laplace operator for image enhancement.

[0028] Because cells may naturally shed debris during growth, and experimental drugs (such as chemotherapy drugs) or fixatives (such as paraformaldehyde) may form precipitates, grayscale images of cells may contain not only cells but also impurities. The characteristics of impurity regions are similar to those of intracellular regions, both exhibiting localized grayscale unevenness. Therefore, it is necessary to combine the local distribution characteristics of each pixel to obtain its confidence level in belonging to a cell region. The degree of suspected cell region and coordinates of each pixel are used as feature vectors. Based on the difference in feature vectors between every two pixels, the pixels are clustered to obtain several clusters. Cell regions, due to the presence of cell membranes or cell walls, have clear boundaries, i.e., they are clustered, and the cluster size is relatively uniform. Impurity regions, however, lack constraints, so their clusters vary in size and shape. Therefore, combining the degree of suspected cell region with the impurity region's significance, the importance of each pixel is obtained.

[0029] For the obtained clusters, since the cellular regions are similar in features to some interference regions, it is necessary to combine the local distribution features of each pixel in the cluster to obtain the confidence level of the suspected cellular region of each pixel.

[0030] First, because cell boundaries distinguish individual cells, the grayscale difference between pixels at the cell boundaries and those within the cell regions is relatively large. Using one pixel as the standard step size for expansion, the original boundaries of each cluster are expanded outwards to obtain the expanded boundaries of each cluster. These expanded boundaries are then used as the cell membrane (wall), and the pixels within the expanded boundaries are designated as the expanded boundary pixels. Figure 2 As shown in the figure, the green area is the expanded boundary, the dark yellow and light yellow parts are a cluster, and the light yellow part is the original boundary of the cluster.

[0031] Furthermore, because cells are bound by cell membranes and culture dishes typically contain the same type of cells, they are generally distributed in clusters (with cells close to each other) and have similar areas (similar number of pixels within each cluster). Therefore, it is necessary to obtain the neighboring clusters for each cluster. Specifically, the distance between the center of one cluster and the centers of other clusters is obtained and denoted as the first distance. Among other clusters, the clusters with the smallest first distance are considered the neighboring clusters of that cluster. The preset number is 5, which can be adjusted by the implementer according to the actual situation. The smaller the difference between the area and distance between each neighboring cluster and its corresponding cluster, the more closely the cluster conforms to the morphology and distribution characteristics of the cell region.

[0032] Finally, the confidence level of a cluster's pixels is obtained based on its neighboring clusters, boundary pixels, and extended boundary pixels. Specifically, the absolute value of the difference in the number of pixels between a cluster and its corresponding neighboring cluster, multiplied by the reciprocal of the product of the first distance, is used as the inter-cluster feature corresponding to the neighboring cluster. The absolute value of the difference between the mean gray value of the original boundary pixels of a cluster and the mean gray value of the extended boundary pixels of the corresponding cluster is calculated and multiplied by the reciprocal of the standard deviation of the extended boundary pixels to obtain the boundary feature of the cluster. The inter-cluster features corresponding to each neighboring cluster of a cluster are summed and multiplied by the boundary feature to obtain the confidence level of the cluster's pixels.

[0033] The specific calculation model is as follows: , Where Zi is the confidence level of the i-th pixel, which is the confidence level of the pixels in the cluster to which the i-th pixel belongs; si is the number of pixels in the cluster to which the i-th pixel belongs; and sj represents the number of pixels in the j-th neighboring cluster to which the i-th pixel belongs. It represents the first distance (the distance between the centers of the two clusters) between the cluster to which the i-th pixel belongs and the j-th neighboring cluster to which the i-th pixel belongs. The cluster features of the j-th neighboring clusters of the i-th pixel are represented. The more similar the area (pixel data) and the smaller the first distance, the denser the distribution, and the more it matches the distribution characteristics of each cell in the cell region. The cluster of the i-th pixel is more likely to be a cell region. and Let represent the average gray value of the pixels at the expanded boundary of the cluster to which the i-th pixel belongs, and the average gray value of the pixels at the original boundary of the cluster to which the i-th pixel belongs, respectively. Expanding the standard deviation of grayscale values ​​of boundary pixels reflects the consistency of grayscale values ​​in that region. Representing boundary features, it is the product of the grayscale difference between the original boundary and the expanded boundary pixels of the cluster and the grayscale consistency of the expanded boundary. The larger the grayscale difference and the higher the grayscale consistency, the more it conforms to the characteristics of the cell region boundary, and the more likely the cluster to be a cell region is.

[0034] After obtaining the confidence score of each pixel, the importance of each pixel can be obtained by combining it with the degree of suspected cell region. The importance is obtained by multiplying the degree of suspected cell region and the confidence score and normalizing the result. The specific calculation model is as follows: Ci represents the importance of the i-th pixel; Yi represents the degree of suspected cell region of the i-th pixel; and Zi represents the confidence level of the i-th pixel. This allows us to obtain the importance of each pixel.

[0035] Step S3: Normalize the absolute value of the difference between the average gray value of the pixels in the eight neighborhoods of a pixel and the gray value of the pixel to obtain the degree of gray change of the pixel.

[0036] After obtaining the importance of each pixel, since the edge strength of each pixel is different, the corresponding gray level change is also different. The Laplace operator responds more strongly to areas with obvious gray level changes in the image. Therefore, when optimizing the enhancement coefficient of each pixel, it is necessary to combine its local gray level change, continuity and importance; weak edges are further enhanced to highlight details; strong edges are appropriately enhanced to eliminate artifacts.

[0037] Since the Laplace operator operates on the eight-neighborhood of each pixel, when analyzing the local grayscale changes of each pixel, the pixels within the eight-neighborhood are analyzed. The absolute value of the difference between the mean grayscale value of the pixels within the eight-neighborhood and the grayscale value of the pixel itself is normalized to obtain the degree of grayscale change of that pixel. The specific calculation model is as follows: , Where Bi represents the grayscale change of the i-th pixel, Hj represents the grayscale value of the j-th neighboring pixel within the eight-neighborhood of the i-th pixel, and norm represents normalization. This represents the difference between the average grayscale value of the i-th pixel and its neighboring pixels. The greater the local grayscale variation of the i-th pixel, the greater the difference between its grayscale value and that of its neighboring pixels. In other words, the higher the degree of grayscale variation, the stronger the local Laplace operator response, and the more likely it is to be over-enhanced. This allows us to obtain the continuity of each pixel.

[0038] Step S4: The pixel with the smallest grayscale difference from a pixel within its eight-neighborhood is taken as the neighboring pixel. The continuity of a pixel is obtained based on the difference in grayscale change between a pixel and its neighboring pixels.

[0039] After obtaining the grayscale change level of each pixel in step S3, it is necessary to analyze the continuity of the pixels. Specifically, the neighboring pixels of a pixel are obtained. The pixel with the smallest grayscale difference from the pixel in the eight-neighborhood is taken as the neighboring pixel. When the grayscale change level of the pixel in the eight-neighborhood is closer to the grayscale change level of the neighboring pixel in the eight-neighborhood, the continuity of the pixel with its neighboring pixels is better, and it is less likely to be an isolated noise or interfering pixel.

[0040] The continuity of a pixel is obtained based on the difference in grayscale variation between a pixel and its neighboring pixels. Specifically, the continuity of a pixel is obtained by normalizing the absolute value of the difference in grayscale variation between a pixel and its neighboring pixels. The specific calculation model is as follows: , Where Li represents the continuity of the i-th pixel; Bi represents the grayscale change of the i-th pixel; The grayscale change of the i-th pixel is calculated using the same method as that of the i-th pixel. The closer the grayscale change of the i-th pixel and its neighboring pixels are, the better the continuity between the i-th pixel and its neighboring pixels, and the less likely it is to be an isolated noise or interfering pixel. When its local Laplace response is weak, its enhancement coefficient should be increased.

[0041] Step S5: Optimize the enhancement coefficient of each pixel according to its importance, grayscale change and continuity to obtain the final enhancement coefficient and enhance the cell image; use the enhanced cell image to detect cell growth status.

[0042] After obtaining the importance, grayscale variation, and continuity of each pixel, for pixels with high grayscale variation, in order to ensure the overall clarity of the image, it is only necessary to adjust them according to their grayscale variation; while for pixels with low grayscale variation, regardless of the magnitude of their grayscale variation, it is only necessary to optimize their enhancement coefficients according to their importance and continuity.

[0043] The enhancement coefficients of each pixel are optimized based on its importance, grayscale variation, and continuity to obtain the final enhancement coefficients and enhance the cell image. Specifically, a grayscale variation threshold is set. Pixels with a grayscale variation greater than the threshold are marked as high-grayscale-variance pixels, and vice versa. Preferably, the grayscale variation threshold is 0.6, which can be set according to user needs. The setting range is generally [0.5, 0.9]. The larger the value, the higher the overall image sharpening intensity, and vice versa.

[0044] For a pixel with high grayscale variation, the enhancement coefficient optimization sensitivity coefficient is multiplied by the degree of grayscale variation of the pixel to obtain the reduction step size of the pixel with high grayscale variation. The difference between the initial enhancement coefficient and the reduction step size is the final enhancement coefficient of the pixel with high grayscale variation. For a pixel with low grayscale variation, the product of the enhancement coefficient optimization sensitivity coefficient, the importance of the pixel, and the continuity of the pixel is the increment step size. The sum of the initial enhancement coefficient and the increment step size is the final enhancement coefficient of the pixel with low grayscale variation.

[0045] The specific calculation model is as follows: , in, Xi represents the final enhancement coefficient of the pixel, which is the initial enhancement coefficient, also known as the fixed enhancement coefficient used by the Laplace operator in this application for image enhancement. Bi represents the degree of change of the i-th pixel, 0.6 is the gray-level change threshold, and 0.5 is the enhancement coefficient optimization sensitivity coefficient. The larger this coefficient is, the stronger the enhancement coefficient optimization and the stronger the image sharpening. It is necessary to set an appropriate sensitivity coefficient to prevent the image from being over-sharpened and distorted. -0.5Bi represents reducing the step size. When the gray-level change of the i-th pixel is high, in order to ensure the clarity of the entire image, its enhancement coefficient needs to be reduced to ensure that the area with strong gray-level change is not over-enhanced. Therefore, it is not necessary to consider whether the pixel is located in the cell region. Li represents the continuity of the i-th pixel. Ci represents the importance of the i-th pixel. 0.5Li*Ci represents increasing the step size. When the gray-level change of the i-th pixel is low, it needs to be enhanced first, that is, its corresponding enhancement coefficient is increased. The basis for enhancement is its degree of belonging to the cell region and its continuity, so as to exclude the situation where interfering pixels are enhanced and interfere with detection.

[0046] After obtaining the optimized enhancement coefficients for each pixel, the entire cell image is enhanced by Laplace transform to obtain the enhanced image, which is then processed in the next step.

[0047] The cell growth states include six stages: cell cycle (G1, S, G2, M), apoptosis, necrosis, etc. Further, historical enhanced cell images are segmented using thresholding (foreground and background segmentation). The segmented images are then manually labeled based on morphological and textural features of cells in different growth states. These labeled images are then used to train an image recognition model (e.g., CNN). Finally, the current enhanced cell image is segmented and input into the trained model to determine the number of cells in different growth states. The detection results are output to the cell growth device's display screen, showing the real-time cell growth state. This helps operators understand the cell growth status and adjust the equipment accordingly for better cell growth.

[0048] Example 2: This embodiment provides a cell growth state detection system based on image processing, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the steps of a cell growth state detection method based on image processing. Since Embodiment 1 has already described a cell growth state detection method based on image processing in detail, it will not be described further here.

[0049] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0050] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0051] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting cell growth state based on image processing, characterized in that, The method includes: The acquired cell images are preprocessed to obtain grayscale images; windows of different scales are established with a pixel as the center, and the degree of suspected cell region of the pixel is obtained based on the grayscale value changes in each window; Cluster the pixels to obtain clusters; expand the boundaries of each cluster to obtain expanded boundary pixels; obtain the confidence score of a cluster's pixels based on its neighboring clusters, the pixels at the boundary, and the expanded boundary pixels; multiply the suspected cell region degree and the confidence score and normalize to obtain the importance score. The degree of grayscale variation of a pixel is obtained by normalizing the absolute value of the difference between the mean grayscale value of the pixels in the eight neighborhoods of a pixel and the grayscale value of the pixel itself. The pixel with the smallest grayscale difference from a pixel within its eight-neighborhood is taken as the neighboring pixel. The continuity of a pixel is obtained based on the difference in grayscale change between a pixel and its neighboring pixels. The enhancement coefficients of each pixel are optimized based on their importance, grayscale variation, and continuity to obtain the final enhancement coefficients, which are then used to enhance the cell image. The enhanced cell image is then used to detect cell growth status.

2. The cell growth state detection method based on image processing according to claim 1, characterized in that, The process of determining the degree of suspected cell region for a pixel based on the grayscale value changes within each window includes: The product of the standard deviation and entropy of the gray values ​​of each pixel within a window corresponding to a given pixel is obtained and the average value is calculated to determine the degree of suspected cell region for that pixel.

3. The cell growth state detection method based on image processing according to claim 1, characterized in that, The process of clustering pixels to obtain clusters includes: The feature vector of a pixel is composed of the degree of suspected cell region and its coordinates. The pixels are then clustered based on the difference in feature vectors between any two pixels to obtain clusters.

4. The cell growth state detection method based on image processing according to claim 1, characterized in that, The process of expanding the boundaries of various clusters to obtain expanded boundary pixels includes: Using one pixel as the standard step size, the original boundaries of each type of cluster are expanded outward to obtain the expanded boundaries of each type of cluster. The pixels within the expanded boundaries are the expanded boundary pixels.

5. The cell growth state detection method based on image processing according to claim 1, characterized in that, The step of obtaining the confidence level of a cluster's pixels based on its neighboring clusters, boundary pixels, and extended boundary pixels includes: The distance between the center of a cluster and the centers of other clusters is obtained and denoted as the first distance. The clusters with the smallest first distance among the other clusters are the neighboring clusters of the given cluster. Obtain the absolute value of the difference in the number of pixels between a cluster and its corresponding neighboring cluster, and the reciprocal of the product of the first distance, as the inter-cluster feature corresponding to the neighboring cluster; calculate the absolute value of the difference between the mean gray value of the pixels at the original boundary of a cluster and the mean gray value of the pixels at the expanded boundary of the cluster, and multiply it by the reciprocal of the standard deviation of the expanded boundary pixels to obtain the boundary feature of the cluster; sum the inter-cluster features corresponding to each neighboring cluster of a cluster and multiply them by the boundary feature to obtain the confidence level of the pixels of the cluster.

6. The cell growth state detection method based on image processing according to claim 1, characterized in that, The method of obtaining the continuity of a pixel based on the difference in grayscale variation between a pixel and its neighboring pixels includes: The continuity of a pixel is obtained by normalizing the absolute value of the difference between the grayscale changes of a pixel and its neighboring pixels.

7. The cell growth state detection method based on image processing according to claim 1, characterized in that, The enhancement coefficient of each pixel is optimized based on its importance, grayscale variation, and continuity, including: Set a threshold for grayscale change. If the grayscale change of a pixel is greater than the threshold, it is marked as a pixel with high grayscale change; otherwise, it is marked as a pixel with low grayscale change. For a pixel with high grayscale variation, the enhancement coefficient optimization sensitivity coefficient is multiplied by the degree of grayscale variation of the pixel to obtain the reduction step size of the pixel with high grayscale variation. The difference between the initial enhancement coefficient and the reduction step size is the final enhancement coefficient of the pixel with high grayscale variation. For a pixel with low grayscale variation, the product of the enhancement coefficient optimization sensitivity coefficient, the importance of the pixel, and the continuity of the pixel is the increment step size. The sum of the initial enhancement coefficient and the increment step size is the final enhancement coefficient of the pixel with low grayscale variation.

8. A cell growth state detection system based on image processing, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the image processing-based cell growth state detection method as described in any one of claims 1-7.

Images