Optimized Detection Method for CTC Images Based on Fluorescent Probe Detection

Abstract

The present invention relates to the field of cell detection, and particularly to an optimized detection method for CTC images based on fluorescence probe detection. The method includes: acquiring a CTC fluorescence microscopic image detected by a fluorescence probe method, converting it into a grayscale image, and performing edge detection; evaluating the boundary degree of edge pixel points through the gradient information of edge pixel points in the microscopic grayscale image; optimizing the boundary fitting process through the boundary degree of edge pixel points; optimizing the evaluation range of the fuzzy factor according to the boundary fitting result; optimizing the objective function of fuzzy clustering according to the optimized evaluation of the fuzzy factor, and completing the segmentation of the CTC fluorescence microscopic grayscale image based on fuzzy clustering to obtain a preliminary segmentation result; performing cell image processing according to the preliminary segmentation result of the CTC fluorescence microscopic image to obtain a CTC cell image detection result, thereby solving the problem of inaccurate CTC image detection caused by uneven fluorescence intensity.

Description

Technical Field

[0001] The present invention relates to the technical field of cell detection, and particularly to an optimized detection method for CTC images based on fluorescence probe detection. Background Art

[0002] Circulating tumor cells (CTCs) are cells shed from solid tumors and enter the peripheral blood circulation, which have important clinical significance. The detection and analysis of CTCs can provide key evidence for the early diagnosis, efficacy evaluation, recurrence monitoring, and personalized treatment of cancer. In the process of processing and detecting CTC fluorescence microscopic images, the commonly used method for segmenting CTC images is usually to perform preliminary segmentation on CTC fluorescence microscopic images based on fuzzy clustering, and complete the detection of tumor cells in CTC fluorescence microscopic images through cell region filling and morphological processing according to the preliminary segmentation results. However, since the fluorescence signal intensity is easily affected by factors such as the concentration of the fluorescence probe, the intensity of the excitation light, and the cell morphology, resulting in blurred boundaries of target cells or loss of some pixel points, affecting the accuracy of segmentation and detection. Therefore, how to solve the inaccurate segmentation of CTC fluorescence microscopic images caused by blurred cell boundaries and partial pixel loss due to uneven fluorescence signal intensity in the process of fuzzy clustering segmentation of CTC images based on fluorescence probe detection has become an urgent problem to be solved. Summary of the Invention

[0003] In view of this, an embodiment of the present invention provides an optimized detection method for CTC images based on fluorescence probe detection to solve the problem of inaccurate CTC detection caused by blurred cell boundaries and partial pixel loss due to uneven fluorescence signal intensity.

[0004] An embodiment of the present invention provides an optimized detection method for CTC images based on fluorescence probe detection. The optimized detection method for CTC images based on fluorescence probe detection includes the following steps:

[0005] Obtain the CTC fluorescence microscopic image detected by the fluorescence probe method and convert it into a CTC fluorescence microscopic grayscale image; perform edge detection on the CTC fluorescence microscopic grayscale image to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image; evaluate the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale image through the gradient information of the edge pixel points in the CTC fluorescence microscopic grayscale image to obtain the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale image; optimize the cell boundary fitting process of the edge pixel points in the CTC fluorescence microscopic grayscale image through the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale image to obtain the boundary fitting result of the CTC fluorescence microscopic grayscale image; optimize the evaluation range of the fuzzy factor of the pixel points in the CTC fluorescence microscopic grayscale image according to the boundary fitting result of the CTC fluorescence microscopic grayscale image to obtain the optimized fuzzy factor evaluation; evaluate the objective function of the pixel points in the fuzzy clustering process according to the optimized fuzzy factor evaluation of the pixel points in the CTC fluorescence microscopic grayscale image, and complete the segmentation of the CTC fluorescence microscopic grayscale image based on fuzzy clustering according to the optimized objective function to obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale image; perform cell image processing according to the preliminary segmentation result of the CTC fluorescence microscopic image to obtain the CTC cell image detection result.

[0006] Preferably, the performing edge detection on the CTC fluorescence microscopic grayscale image to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image includes:

[0007] Obtain the double thresholds of the CTC fluorescence microscopic grayscale image and the edge detection Canny algorithm, and perform edge detection on the CTC fluorescence microscopic grayscale image through the Canny algorithm to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image.

[0008] Preferably, the evaluating the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale image through the gradient information of the edge pixel points in the CTC fluorescence microscopic grayscale image to obtain the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale image includes:

[0009] Obtain the gradient direction and gradient magnitude of all edge pixels in the CTC fluorescence microscopic grayscale image, obtain the preset number of neighboring pixels of the edge pixels, and obtain the neighboring set of any edge pixel among all edge pixels in the CTC fluorescence microscopic grayscale image according to the number of neighboring pixels of the edge pixels; for any target edge pixel in the CTC fluorescence microscopic grayscale image, use the cosine function mapping result of the gradient direction difference between the target edge pixel and any edge pixel in the neighboring set of the target edge pixel as the first gradient direction difference of the target edge pixel; use the calculation result of subtracting the gradient magnitude of the target edge pixel from the average value of all gradient magnitudes in the neighboring set of the target edge pixel as the first gradient magnitude difference of the target edge pixel; use the average value of the spatial distance differences between the target edge pixel and all edge pixels in the neighboring set of the target edge pixel as the first average distance of the target edge pixel;

[0010] Use the calculation result of multiplying the first gradient direction difference, the first gradient magnitude difference, and the first average distance of the target edge pixel as the first boundary evaluation, use the normalization calculation result of the average value of the first boundary evaluations of all edge pixels corresponding to the target edge pixel and the neighboring set of the target edge pixel as the second boundary evaluation, and use the calculation result of subtracting the second boundary evaluation from the constant 1 as the boundary degree of the target edge pixel.

[0011] Preferably, optimizing the cell boundary fitting process of the edge pixels in the CTC fluorescence microscopic grayscale image through the boundary degree of the edge pixels in the CTC fluorescence microscopic grayscale image to obtain the boundary fitting result of the CTC fluorescence microscopic grayscale image includes:

[0012] Obtain all edge pixels in the CTC fluorescence microscopic grayscale image, obtain the corresponding boundary degrees of all the edge pixels, and obtain the preset cell boundary ellipse fitting parameters, where the cell boundary ellipse fitting parameters include the ellipse fitting center, the ellipse fitting major axis length, and the ellipse fitting minor axis length; the boundary fitting objective function is:

[0013] ;

[0014] Wherein, represents the boundary fitting objective function corresponding to the th edge pixel; represents the preset number of neighboring pixels of the edge pixels; represents the th edge pixel in the neighboring set corresponding to the th edge pixel; represents the The abscissa of the th edge pixel in the neighborhood set corresponding to a certain edge pixel in the image; Represents the abscissa of the center of the ellipse fitting preset in the boundary fitting process; Represents the th edge pixel in the neighborhood set corresponding to a certain edge pixel; The ordinate of the th edge pixel in the image; Represents the ordinate of the center of the ellipse fitting preset in the boundary fitting process; Represents the length of the major axis of the ellipse preset in the boundary fitting process; Represents the length of the minor axis of the ellipse preset in the boundary fitting process;

[0015] For the boundary fitting objective function, the gradient descent method is used to jointly optimize the ellipse center coordinates and the major and minor axis length parameters, and the fitting process is completed to obtain all the boundary fitting results in the CTC fluorescence microscopic grayscale image.

[0016] Preferably, the optimization of the evaluation range of the fuzzy factor of the pixel points in the CTC fluorescence microscopic grayscale image is carried out according to the boundary fitting results of the CTC fluorescence microscopic grayscale image, and the optimized fuzzy factor evaluation is obtained, including:

[0017] For the boundary fitting results of the CTC fluorescence microscopic grayscale image, the elliptical range formed by each boundary fitting result is used as a cell region, and the pixel points in the cell region are used as a whole local range. For any pixel point in the CTC fluorescence microscopic grayscale image, it is judged whether it is in any cell region. If the pixel point is in any cell region, the cell region where the pixel point is located is used as the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the fuzzy clustering process. If the pixel point is not in any cell region, the eight-neighborhood range of the pixel point itself is used as the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the fuzzy clustering process.

[0018] For any pixel point in the CTC fluorescence microscopic grayscale image, the optimized fuzzy factor evaluation is obtained through the overall difference between the pixel points in the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the CTC fluorescence microscopic grayscale image and the center point of the target cluster class in the clustering process.

[0019] Preferably, the evaluation of the objective function of the pixel points in the fuzzy clustering process is carried out according to the optimized fuzzy factor evaluation of the pixel points in the CTC fluorescence microscopic grayscale image, and the CTC fluorescence microscopic grayscale image segmentation based on fuzzy clustering is completed according to the optimized objective function, and the preliminary segmentation result of the CTC fluorescence microscopic grayscale image is obtained, including:

[0020] The calculation formula of the objective function of the pixel points in the fuzzy clustering process is as follows:

[0021] ;

[0022] Wherein, represents the objective function of the clustering process of the CTC fluorescence microscopic gray-scale image; represents the number of pixel points in the CTC fluorescence microscopic gray-scale image; represents the preset number of cluster classes in the clustering process of the CTC fluorescence microscopic gray-scale image; represents the th pixel point in the CTC fluorescence microscopic gray-scale image to the th cluster class membership degree; represents the fuzzy weighting index in the fuzzy clustering process; represents the Euclidean distance of the gray value between the th pixel point in the CTC fluorescence microscopic gray-scale image and the cluster center point of the th cluster class in the clustering process; represents the optimized fuzzy factor evaluation of the th pixel point in the CTC fluorescence microscopic gray-scale image to the th cluster class in the fuzzy clustering process;

[0023] Based on the optimized objective function, complete the segmentation of the CTC fluorescence microscopic gray-scale image based on fuzzy clustering, and obtain the preliminary segmentation result of the CTC fluorescence microscopic gray-scale image.

[0024] Preferably, performing cell image processing according to the preliminary segmentation result of the CTC fluorescence microscopic image to obtain the CTC cell image detection result, including:

[0025] Obtain the preliminary segmentation result of the CTC fluorescence microscopic gray-scale image, and according to the segmentation results corresponding to each cluster class in the preliminary segmentation result, obtain the cluster class corresponding to the CTC cell; perform flood filling on the cluster class corresponding to the CTC cell, and perform morphological operation processing to obtain the final segmentation result, and obtain the detection result of the CTC cell image.

[0026] Preferably, the optimized fuzzy factor evaluation is obtained by the overall difference between the pixel points in the optimized evaluation range of the fuzzy factor corresponding to the pixel points of the CTC fluorescence microscopic gray-scale image and the target cluster center point in the clustering process, including:

[0027] When any pixel point in the CTC fluorescence microscopic gray-scale image is in any cell region, the calculation formula of the fuzzy factor of any pixel point in the CTC fluorescence microscopic gray-scale image is as follows:

[0028] ;

[0029] Among them, represents the fuzzy factor of the th pixel point in the CTC fluorescence microscopic grayscale image with respect to the th cluster class; represents the optimized evaluation range of the th pixel point in the CTC fluorescence microscopic grayscale image; represents the spatial Euclidean distance between the th pixel point in the CTC fluorescence microscopic grayscale image and the th pixel point in the optimized evaluation range of this pixel point; represents the membership degree of the th pixel point in the optimized evaluation range of the CTC fluorescence microscopic grayscale image with respect to the th cluster class; represents the fuzzy weighting index in the fuzzy clustering process; represents the Euclidean distance of the gray value between the th pixel point in the optimized evaluation range of the CTC fluorescence microscopic grayscale image and the th pixel point and the cluster center point of the th cluster class in the clustering process. th cluster class.

[0030] The beneficial effects of the embodiments of the present invention compared with the prior art are:

[0031] The present invention acquires CTC fluorescence microscopic images detected by a fluorescence probe method and converts them into CTC fluorescence microscopic grayscale images; performs edge detection on the CTC fluorescence microscopic grayscale images to obtain all edge pixel points in the CTC fluorescence microscopic grayscale images; evaluates the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images through the gradient information of the edge pixel points in the CTC fluorescence microscopic grayscale images to obtain the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images; optimizes the cell boundary fitting process of the edge pixel points in the CTC fluorescence microscopic grayscale images through the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images to obtain the boundary fitting result of the CTC fluorescence microscopic grayscale images; optimizes the evaluation range of the fuzzy factor of the pixel points in the CTC fluorescence microscopic grayscale images according to the boundary fitting result of the CTC fluorescence microscopic grayscale images to obtain the optimized fuzzy factor evaluation; evaluates the objective function of the pixel points in the fuzzy clustering process according to the optimized fuzzy factor evaluation of the pixel points in the CTC fluorescence microscopic grayscale images, and completes the segmentation of the CTC fluorescence microscopic grayscale images based on fuzzy clustering according to the optimized objective function to obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale images; performs cell image processing according to the preliminary segmentation result of the CTC fluorescence microscopic images to obtain the CTC cell image detection result. Among them, by evaluating the cell boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images, and then optimizing the cell boundary fitting through the cell boundary degree of the edge pixel points, the accuracy of cell boundary fitting is improved, and the local range for evaluating the fuzzy factor of the pixel points in the fuzzy clustering process is obtained according to the optimized fitting result. When performing fuzzy clustering on the CTC fluorescence microscopic grayscale images, the cluster classification of the pixel points in the cell boundary range is optimized to obtain a more accurate preliminary segmentation result of the CTC fluorescence microscopic images, thereby avoiding inaccurate segmentation of the CTC fluorescence microscopic images caused by fuzzy cell boundaries due to uneven fluorescence intensity and loss of some pixels, and improving the detection accuracy of the CTC fluorescence microscopic images. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings, which form a part of the present invention, are used to provide a further understanding of the present invention. The schematic embodiments and descriptions thereof of the present invention are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0033] Figure 1 is a flowchart of the optimized detection method for CTC images based on fluorescence probe detection according to the embodiment of the present invention; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] Embodiments of the present disclosure will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having like or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0035] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned accompanying drawings are used to distinguish similar objects and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used for this toothpaste can be interchanged under appropriate circumstances so that the embodiments of the present disclosure described here can be implemented in an order other than those illustrated or described here. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0036] In order to illustrate the technical solution of the present invention, it will be described below through specific embodiments.

[0037] The specific scenario targeted by the present invention is: during the process of detecting and identifying tumor cells in CTC fluorescence microscopic images, image optimization processing is performed.

[0038] In addition, in the embodiments of the present invention, the CTC mentioned refers to circulating tumor cells.

[0039] See Figure 1 , which is a flowchart of the CTC image optimization detection method based on fluorescence probe detection provided in the first embodiment of the present invention. As Figure 1 shown, the CTC image optimization detection method based on fluorescence probe detection may include:

[0040] Step S101, obtain a CTC fluorescence microscopic image detected by a fluorescence probe method and convert it into a CTC fluorescence microscopic grayscale image; perform edge detection on the CTC fluorescence microscopic grayscale image to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image.

[0041] Collect a CTC fluorescence microscopic image detected by a fluorescence probe method. The resolution of the collected image is a fixed resolution. In this embodiment, pixels are selected. Then, perform color space conversion on the CTC fluorescence microscopic image. Specifically, use a grayscale processing method to convert the collected CTC fluorescence microscopic image into a grayscale image for subsequent CTC cell detection. Among them, the grayscale processing technology is a prior art and will not be elaborated here.

[0042] After obtaining the CTC fluorescence microscopic grayscale image, perform edge detection on the CTC fluorescence microscopic grayscale image to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image, including:

[0043] Obtain the double thresholds of the CTC fluorescence microscopic grayscale image and the Canny edge detection algorithm, and perform edge detection on the CTC fluorescence microscopic grayscale image through the Canny algorithm to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image.

[0044] It should be noted that in one embodiment, the double thresholds of the Canny edge detection algorithm are selected as , and the double thresholds can be adjusted according to the real-time scenario without requirements. Thus, perform edge detection on the CTC fluorescence microscopic grayscale image to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image.

[0045] Step S102, evaluate the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale image through the gradient information of the edge pixel points in the CTC fluorescence microscopic grayscale image to obtain the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale image.

[0046] After obtaining all edge pixel points in the CTC fluorescence microscopic grayscale image, it is necessary to analyze all edge pixel points in the image to obtain the boundary degree corresponding to each edge pixel point. Obtain the gradient direction and gradient amplitude of all edge pixel points in the CTC fluorescence microscopic grayscale image, obtain the preset number of neighboring pixel points of the edge pixel points, and obtain the neighboring set of any edge pixel point in all edge pixel points in the CTC fluorescence microscopic grayscale image according to the number of neighboring pixel points of the edge pixel points; for any target edge pixel point in the CTC fluorescence microscopic grayscale image, use the cosine function mapping result of the gradient direction difference between the target edge pixel point and any edge pixel point in the neighboring set of the target edge pixel point as the first gradient direction difference of the target edge pixel point; use the calculation result of subtracting the gradient amplitude of the target edge pixel point from the average value of all gradient amplitudes in the neighboring set of the target edge pixel point as the first gradient amplitude difference of the target edge pixel point; use the average value of the spatial distance differences between the target edge pixel point and all edge pixel points in the neighboring set of the target edge pixel point as the first average distance of the target edge pixel point;

[0047] The calculation result of multiplying the first gradient direction difference, the first gradient magnitude difference of the target edge pixel by the first average distance is used as the first boundary evaluation. The normalized calculation result of the average value of the first boundary evaluations of all edge pixels in the target edge pixel and the corresponding neighbor set of the target edge pixel is used as the second boundary evaluation. The calculation result of subtracting the constant 1 from the second boundary evaluation is used as the boundary degree of the target edge pixel.

[0048] In one embodiment, assume that the gradient value of the th edge pixel is , and the gradient direction of the th edge pixel is . Then, the calculation expression corresponding to the boundary degree of the th edge pixel in the CTC fluorescence microscopic grayscale image is:

[0049]

[0050] Where, represents the boundary degree of the th edge pixel in the CTC fluorescence microscopic grayscale image; represents the preset number of neighboring pixels of the edge pixel; represents the set of neighboring pixels of the th edge pixel in the CTC fluorescence microscopic grayscale image; respectively represent the gradient direction of the th edge pixel and the th edge pixel in the neighbor set of the th edge pixel; represents the cosine function; represents the th edge pixel in the neighbor set of the th edge pixel in the CTC fluorescence microscopic grayscale image; represents the average gradient of the set of neighboring edge pixels of the th edge pixel in the CTC fluorescence microscopic grayscale image; represents the spatial Euclidean distance between the th edge pixel and the th edge pixel in the neighbor set of the th edge pixel in the CTC fluorescence microscopic grayscale image; represents the normalization function; represents the constant ; represents the absolute value calculation.

[0051] It should be noted that in the existing fuzzy clustering process based on local pixel information, during the membership degree iteration process of clustering, when evaluating the membership degree of the target pixel, the membership degree of the target data point is optimized by the distance between the neighborhood pixels of the target pixel and the center points of each cluster, so as to ensure that the pixels in the local range of the target pixel are divided into the same cluster during the CTC image segmentation based on clustering. However, because only the neighborhood range of the target pixel in the image is considered for the local information of the target pixel in the fuzzy clustering process, when there is a lack of internal information of a single cell due to uneven fluorescence intensity in the current scenario, the pixels in the cell still cannot be accurately segmented. To solve this problem, in the fuzzy clustering process based on local information, the local information of the pixels within the boundary needs to be obtained according to the boundary pixels of the cell, increasing the range represented by the local in the clustering, so as to divide the pixels in a single cell into the same cluster. Therefore, it is necessary to first evaluate the boundary degree of the pixels in the CTC fluorescence microscopic image, and then optimize the boundary fitting process of the edge pixels in the image according to the boundary degree of the pixels, so as to avoid the incorrect evaluation caused by the influence of non-edge cell pixels during the process of obtaining the cell boundary by fitting the edge pixels alone. That is, according to the gradient direction consistency between the edge pixels in the CTC fluorescence microscopic gray image and other edge pixels in its neighboring set , the gradient value consistency between the edge pixels and other edge pixels in its neighboring set and the spatial distance between the edge pixels and other edge pixels in its neighboring set to evaluate the boundary degree. When the target edge pixel and the pixels in the set of its neighboring edge pixels show similar gradient directions, the boundary degree of the target edge pixel is higher. When the gradient intensity of the neighboring set where the target edge pixel is located is more uniform, that is higher, the boundary degree of this edge pixel is lower. When the overall spatial distance between the target edge pixel and other edge pixels in its neighboring set is higher, that is higher, the boundary degree of this edge pixel is lower. In one embodiment, the preset number of neighboring pixels of the edge pixel is set to , and the setting of this number of neighboring pixels can be adjusted accordingly according to the actual scenario and there is no requirement.

[0052] Step S103, optimize the cell boundary fitting process of the edge pixels in the CTC fluorescence microscopic gray image according to the boundary degree of the edge pixels in the CTC fluorescence microscopic gray image, and obtain the boundary fitting result of the CTC fluorescence microscopic gray image.

[0053] After obtaining the boundary degree of the edge pixel point, the cell boundary fitting process of the edge pixel point in the CTC fluorescence microscopic grayscale image is optimized according to the boundary degree of each edge pixel point, and the boundary fitting result of the CTC fluorescence microscopic grayscale image is obtained. Specifically, all edge pixel points in the CTC fluorescence microscopic grayscale image are obtained, the boundary degrees corresponding to all edge pixel points are obtained, and the preset cell boundary ellipse fitting parameters are obtained, and the cell boundary ellipse fitting parameters include the ellipse fitting center, the ellipse fitting major axis length, and the ellipse fitting minor axis length; the boundary fitting objective function is:

[0054] ;

[0055] in, Indicates The boundary fitting objective function corresponding to the edge pixels; Indicates the number of neighboring pixels of the preset edge pixel; Indicates The first The boundary degree of edge pixels; Indicates The first The horizontal coordinate of the edge pixel in the image; Indicates the horizontal coordinate of the center of the ellipse fitting circle preset in the boundary fitting process; Indicates The first The vertical coordinate of the edge pixel in the image; Indicates the ordinate of the center of the ellipse fitting circle preset in the boundary fitting process; Indicates the length of the major axis of the ellipse preset in the boundary fitting process; Indicates the length of the minor axis of the ellipse preset in the boundary fitting process;

[0056] It should be noted that in the process of cell boundary fitting, each edge pixel is fitted through the edge pixels in its neighbor set. The first The boundary degree of an edge pixel point is the boundary degree of the pixel point obtained in the above-mentioned boundary degree obtaining process.

[0057] For the boundary fitting objective function, the coordinates of the center of the ellipse and the major and minor axis length parameters are jointly optimized by the gradient descent method to complete the fitting process, and all boundary fitting results in the CTC fluorescence microscopy grayscale image are obtained.

[0058] Step S104, optimize the evaluation range of the fuzzy factor of the pixel points in the CTC fluorescence microscopic gray-scale image according to the boundary fitting result of the CTC fluorescence microscopic gray-scale image, and obtain the optimized fuzzy factor evaluation

[0059] After obtaining all the boundary fitting results in the CTC fluorescence microscopic gray-scale image, it is necessary to optimize the evaluation range of the fuzzy factor of the pixel points in the CTC fluorescence microscopic gray-scale image according to the boundary fitting result of the CTC fluorescence microscopic gray-scale image, and obtain the optimized fuzzy factor evaluation. Specifically, for the boundary fitting result of the CTC fluorescence microscopic gray-scale image, the elliptical range formed by each boundary fitting result is used as a cell region, and the pixel points in the cell region are used as an overall local range. For any pixel point in the CTC fluorescence microscopic gray-scale image, it is judged whether it is in any cell region. If the pixel point is in any cell region, the cell region where the pixel point is located is used as the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the fuzzy clustering process. If the pixel point is not in any cell region, the octal neighborhood range of the pixel point itself is used as the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the fuzzy clustering process.

[0060] For any pixel point in the CTC fluorescence microscopic gray-scale image, obtain the optimized fuzzy factor evaluation through the overall difference between the pixel points in the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the CTC fluorescence microscopic gray-scale image and the center point of the target cluster class in the clustering process.

[0061] The optimized fuzzy factor evaluation is obtained through the overall difference between the pixel points in the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the CTC fluorescence microscopic gray-scale image and the center point of the target cluster class in the clustering process. Specifically, in one embodiment, the th pixel point in the CTC fluorescence microscopic gray-scale image for the th cluster class, the calculation formula of the fuzzy factor is as follows:

[0062] ;

[0063] Wherein, represents the fuzzy factor of the th pixel point in the CTC fluorescence microscopic gray-scale image for the th cluster class; represents the optimized evaluation range of the th pixel point in the CTC fluorescence microscopic gray-scale image; represents the spatial Euclidean distance between the th pixel point in the CTC fluorescence microscopic gray-scale image and the th pixel point in the optimized evaluation range of this pixel point; represents the membership degree of the th pixel point in the optimized evaluation range of the th pixel point in the CTC fluorescence microscopic grayscale image to the th cluster class; represents the fuzzy weighting index in the fuzzy clustering process; represents the th pixel point in the optimized evaluation range of the th pixel point in the CTC fluorescence microscopic grayscale image and the Euclidean distance of the grayscale value between the th pixel point and the cluster center point of the

[0064] It should be noted that when the position of the th pixel point in the CTC fluorescence microscopic grayscale image is within any boundary fitting result in the image, then the optimized evaluation range of the th pixel point in the CTC fluorescence microscopic grayscale image is the boundary range corresponding to the boundary fitting result corresponding to this pixel point. When the position of the th pixel point in the CTC fluorescence microscopic grayscale image is not within any boundary fitting result in the image, then the optimized evaluation range of the th pixel point in the CTC fluorescence microscopic grayscale image is the eight-neighborhood range of this pixel point.

[0065] Step S105, evaluate the objective function of the pixel points in the fuzzy clustering process according to the optimized fuzzy factor evaluation of the pixel points in the CTC fluorescence microscopic grayscale image, and complete the segmentation of the CTC fluorescence microscopic grayscale image based on fuzzy clustering according to the optimized objective function to obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale image; perform cell image processing according to the preliminary segmentation result of the CTC fluorescence microscopic image to obtain the CTC cell image detection result.

[0066] After obtaining the optimized fuzzy factor evaluation, it is necessary to evaluate the objective function of the pixel points in the fuzzy clustering process according to the optimized fuzzy factor evaluation of the pixel points in the CTC fluorescence microscopic grayscale image.

[0067] In one embodiment, the calculation formula of the objective function of the pixel points in the fuzzy clustering process is:

[0068]

[0069] Among them, represents the objective function of the clustering process of the CTC fluorescence microscopic grayscale image; represents the number of pixel points in the CTC fluorescence microscopic grayscale image; represents the preset number of cluster classes in the clustering process of the CTC fluorescence microscopic grayscale image; represents the th pixel point in the CTC fluorescence microscopic grayscale image to the th cluster class membership degree; represents the fuzzy weighting exponent in the fuzzy clustering process; represents the th pixel point in the CTC fluorescence microscopic grayscale image and the th cluster center point in the clustering process, the Euclidean distance of the grayscale values between them; represents the th pixel point in the CTC fluorescence microscopic grayscale image to the th optimized fuzzy factor evaluation of the cluster class in the fuzzy clustering process;

[0070] It should be noted that in the objective function of the pixel points in the fuzzy clustering process, the fuzzy weighting exponent is a preset parameter. In this embodiment, the fuzzy weighting exponent is set to , and this parameter setting can be adjusted according to the actual scenario without specific requirements.

[0071] Complete the segmentation of the CTC fluorescence microscopic grayscale image based on fuzzy clustering according to the optimized objective function, and obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale image.

[0072] After obtaining the preliminary segmentation result of the CTC fluorescence microscopic grayscale image, cell image processing can be performed according to the preliminary segmentation result of the CTC fluorescence microscopic image to obtain the CTC cell image detection result. Specifically, obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale image, and according to the segmentation results corresponding to each cluster class in the preliminary segmentation result, obtain the cluster class corresponding to the CTC cell; perform flood filling on the cluster class corresponding to the CTC cell, and perform morphological operation processing to obtain the final segmentation result, and obtain the detection result of the CTC cell image.

[0073] It should be noted that flood filling and morphological operation processing are existing technologies, so they will not be elaborated here.

[0074] In summary, the embodiments of the present invention acquire CTC fluorescence microscopic images detected by a fluorescence probe method and convert them into CTC fluorescence microscopic grayscale images; perform edge detection on the CTC fluorescence microscopic grayscale images to obtain all edge pixel points in the CTC fluorescence microscopic grayscale images; evaluate the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images through the gradient information of the edge pixel points in the CTC fluorescence microscopic grayscale images to obtain the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images; optimize the cell boundary fitting process of the edge pixel points in the CTC fluorescence microscopic grayscale images through the boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images to obtain the boundary fitting result of the CTC fluorescence microscopic grayscale images; optimize the evaluation range of the fuzzy factor of the pixel points in the CTC fluorescence microscopic grayscale images according to the boundary fitting result of the CTC fluorescence microscopic grayscale images to obtain the optimized fuzzy factor evaluation; evaluate the objective function of the pixel points in the fuzzy clustering process according to the optimized fuzzy factor evaluation of the pixel points in the CTC fluorescence microscopic grayscale images, and complete the segmentation of the CTC fluorescence microscopic grayscale images based on fuzzy clustering according to the optimized objective function to obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale images; perform cell image processing according to the preliminary segmentation result of the CTC fluorescence microscopic images to obtain the CTC cell image detection result. Among them, by evaluating the cell boundary degree of the edge pixel points in the CTC fluorescence microscopic grayscale images, and then optimizing the cell boundary fitting through the cell boundary degree of the edge pixel points, the accuracy of cell boundary fitting is improved, and the local range for evaluating the fuzzy factor of the pixel points in the fuzzy clustering process is obtained according to the optimized fitting result. When performing fuzzy clustering on the CTC fluorescence microscopic grayscale images, the cluster classification of the pixel points in the cell boundary range is optimized to obtain a more accurate preliminary segmentation result of the CTC fluorescence microscopic images, thereby avoiding inaccurate segmentation of the CTC fluorescence microscopic images caused by blurred cell boundaries due to uneven fluorescence intensity and loss of some pixels, and improving the detection accuracy of the CTC fluorescence microscopic images.

[0075] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A CTC image optimization detection method based on fluorescent probe detection, characterized in that: The method comprises: The method comprises the following steps: obtaining a CTC fluorescence microscopic image detected by a fluorescent probe method and converting it into a CTC fluorescence microscopic grayscale image; performing edge detection on the CTC fluorescence microscopic grayscale image to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image; evaluating the boundary degree of edge pixel points in the CTC fluorescence microscopic grayscale image by using the gradient information of edge pixel points in the CTC fluorescence microscopic grayscale image to obtain the boundary degree of edge pixel points in the CTC fluorescence microscopic grayscale image; optimizing the cell boundary fitting process of edge pixel points in the CTC fluorescence microscopic grayscale image by using the boundary degree of edge pixel points in the CTC fluorescence microscopic grayscale image to obtain the CTC The boundary fitting result of the fluorescence microscopic grayscale image; according to the boundary fitting result of the CTC fluorescence microscopic grayscale image, the fuzzy factor evaluation range of the pixel points in the CTC fluorescence microscopic grayscale image is optimized to obtain the optimized fuzzy factor evaluation; according to the optimized fuzzy factor evaluation of the pixel points in the CTC fluorescence microscopic grayscale image, the objective function of the pixel points in the fuzzy clustering process is evaluated, and according to the optimized objective function, the CTC fluorescence microscopic grayscale image segmentation based on fuzzy clustering is completed to obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale image; according to the preliminary segmentation result of the CTC fluorescence microscopic image, cell image processing is performed to obtain the CTC cell image detection result.

2. The CTC image optimization detection method based on fluorescent probe detection according to claim 1, characterized in that: The performing edge detection on the CTC fluorescence microscopic grayscale image to obtain all edge pixel points in the CTC fluorescence microscopic grayscale image includes: A double threshold of the CTC fluorescence microscopic grayscale image and the edge detection Canny algorithm is obtained, and edge detection is performed on the CTC fluorescence microscopic grayscale image using the Canny algorithm to obtain all edge pixels in the CTC fluorescence microscopic grayscale image.

3. The CTC image optimization detection method based on fluorescent probe detection according to claim 1, characterized in that, The step of evaluating the boundary degree of edge pixels in the CTC fluorescence microscopic grayscale image by using the gradient information of edge pixels in the CTC fluorescence microscopic grayscale image to obtain the boundary degree of edge pixels in the CTC fluorescence microscopic grayscale image includes: Obtain the gradient direction and gradient amplitude of all edge pixels in the CTC fluorescence microscopic grayscale image, obtain a preset number of neighboring pixels of the edge pixel, and obtain a neighbor set of any edge pixel among all edge pixels in the CTC fluorescence microscopic grayscale image according to the number of neighboring pixels of the edge pixel; for any target edge pixel in the CTC fluorescence microscopic grayscale image, use the cosine function mapping result of the gradient direction difference between the target edge pixel and any edge pixel in the neighbor set of the target edge pixel as the first gradient direction difference of the target edge pixel; use the calculation result of subtracting the gradient amplitude of the target edge pixel from the mean of all gradient amplitudes in the neighbor set of the target edge pixel as the first gradient amplitude difference of the target edge pixel; use the mean of the spatial distance difference between the target edge pixel and all edge pixels in the neighbor set of the target edge pixel as the first average distance of the target edge pixel; The calculation result of multiplying the first gradient direction difference, the first gradient amplitude difference and the first average distance of the target edge pixel point is used as the first boundary evaluation, the normalized calculation result of the first boundary evaluation mean of the target edge pixel point and all edge pixels in the neighboring set corresponding to the target edge pixel point is used as the second boundary evaluation, and the calculation result of subtracting the constant 1 from the second boundary evaluation is used as the boundary degree of the target edge pixel point.

4. The CTC image optimization detection method based on fluorescent probe detection according to claim 1, characterized in that: The step of optimizing the cell boundary fitting process of edge pixels in the CTC fluorescence microscopic grayscale image by the boundary degree of edge pixels in the CTC fluorescence microscopic grayscale image to obtain the boundary fitting result of the CTC fluorescence microscopic grayscale image includes: All edge pixels in the CTC fluorescence microscopic grayscale image are obtained, the boundary degrees corresponding to all edge pixels are obtained, and preset cell boundary ellipse fitting parameters are obtained, wherein the cell boundary ellipse fitting parameters include the ellipse fitting center, the ellipse fitting major axis length, and the ellipse fitting minor axis length; the boundary fitting objective function is: ； in, Indicates The boundary fitting objective function corresponding to the edge pixels; Indicates the number of neighboring pixels of the preset edge pixel; Indicates The first The boundary degree of edge pixels; Indicates The first The horizontal coordinate of the edge pixel in the image; Indicates the horizontal coordinate of the center of the ellipse fitting circle preset in the boundary fitting process; Indicates The first The vertical coordinate of the edge pixel in the image; Indicates the ordinate of the center of the ellipse fitting circle preset in the boundary fitting process; Indicates the length of the major axis of the ellipse preset in the boundary fitting process; Indicates the length of the minor axis of the ellipse preset in the boundary fitting process; For the boundary fitting objective function, the coordinates of the center of the ellipse and the major and minor axis length parameters are jointly optimized by the gradient descent method to complete the fitting process, and all boundary fitting results in the CTC fluorescence microscopy grayscale image are obtained.

5. The CTC image optimization detection method based on fluorescent probe detection according to claim 1, characterized in that, The step of optimizing the fuzzy factor evaluation range of the pixel points in the CTC fluorescence microscopic grayscale image according to the boundary fitting result of the CTC fluorescence microscopic grayscale image to obtain the optimized fuzzy factor evaluation includes: For the boundary fitting results of the CTC fluorescence microscopic grayscale image, the elliptical range formed by each boundary fitting result is taken as a cell region, and the pixel points in the cell region are taken as a whole local range. For any pixel point in the CTC fluorescence microscopic grayscale image, it is judged whether it is in any cell region. If the pixel point is in any cell region, the cell region where the pixel point is located is used as the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the fuzzy clustering process. If the pixel point is not in any cell region, the eight-neighborhood range of the pixel point itself is used as the optimized evaluation range of the fuzzy factor corresponding to the pixel point in the fuzzy clustering process. For any pixel point in the CTC fluorescence microscopic grayscale image, the optimized fuzzy factor evaluation is obtained by the overall difference between the pixel point in the optimized evaluation range of the fuzzy factor corresponding to the pixel point of the CTC fluorescence microscopic grayscale image and the target cluster center point in the clustering process.

6. The CTC image optimization detection method based on fluorescent probe detection according to claim 1, characterized in that: The objective function of the pixels in the fuzzy clustering process is evaluated according to the fuzzy factor evaluation after the pixel in the CTC fluorescence microscopic grayscale image is optimized, and the CTC fluorescence microscopic grayscale image segmentation based on fuzzy clustering is completed according to the optimized objective function to obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale image, including: The calculation formula of the objective function of the pixel point in the fuzzy clustering process is: ； in, represents the objective function of the clustering process of the CTC fluorescence microscopic grayscale image; Represents the number of pixels in the CTC fluorescence microscopy grayscale image; Indicates the preset number of clusters for the clustering process of the CTC fluorescence microscopic grayscale image; Indicates the CTC fluorescence microscopic grayscale image Pixel pair The degree of membership of each cluster; represents the fuzzy weighted index in the fuzzy clustering process; Indicates the CTC fluorescence microscopic grayscale image The pixel points are related to the The Euclidean distance of the gray values ​​between the cluster center points of the clusters; Indicates the CTC fluorescence microscopic grayscale image The pixel points in the fuzzy clustering process are Evaluation of optimized fuzzy factors for clusters; According to the optimized objective function, the CTC fluorescence microscopic grayscale image segmentation based on fuzzy clustering is completed to obtain the preliminary segmentation result of the CTC fluorescence microscopic grayscale image.

7. The CTC image optimization detection method based on fluorescent probe detection according to claim 1, characterized in that: The performing cell image processing according to the preliminary segmentation result of the CTC fluorescence microscopy image to obtain the CTC cell image detection result includes: Obtain a preliminary segmentation result of the CTC fluorescence microscopic grayscale image, and obtain the clusters corresponding to the CTC cells according to the segmentation results corresponding to each cluster in the preliminary segmentation result; perform flood filling on the clusters corresponding to the CTC cells, perform morphological operation processing to obtain the final segmentation result, and obtain the detection result of the CTC cell image.

8. The CTC image optimization detection method based on fluorescent probe detection according to claim 5, characterized in that: The optimized fuzzy factor evaluation is obtained by the overall difference between the pixel points in the optimized evaluation range of the fuzzy factor corresponding to the pixel points of the CTC fluorescence microscopic grayscale image and the target cluster center point in the clustering process, including: When any pixel in the CTC fluorescence microscopic grayscale image is in any cell region, the calculation formula for the fuzzy factor of any pixel in the CTC fluorescence microscopic grayscale image is: ； in, Indicates the CTC fluorescence microscopic grayscale image Pixels for the The fuzzy factor of each cluster; Indicates the CTC fluorescence microscopic grayscale image The optimized evaluation range of pixels; Indicates the CTC fluorescence microscopic grayscale image The pixel point and the optimal evaluation range of the pixel point The spatial Euclidean distance between pixels; Indicates the CTC fluorescence microscopic grayscale image The first pixel in the optimization evaluation range Pixels for the The degree of membership of each cluster; represents the fuzzy weighted index in the fuzzy clustering process; Indicates the CTC fluorescence microscopic grayscale image The first pixel in the optimization evaluation range The pixel points are related to the The Euclidean distance between the grayscale values ​​of the cluster center points of the clusters.

Images