Diagnostic method and device for inflammatory bowel disease based on pathological morphological characteristics

By constructing windows and correcting the crypt probability of pixel points in the pathological grayscale image of intestinal biopsy sample slices, crypt images are generated, and the problem of relying on manual annotation and high computing resources in the prior art is solved, and the inflammatory bowel is realized Accurate identification and diagnosis of diseases.

CN118762267BActive Publication Date: 2024-11-26CHANGJIANG CLOUD TRANSIT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411236955.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-05
Publication Date
2024-11-26
Estimated Expiration
2044-09-05

AI Technical Summary

Technical Problem

Existing convolutional neural network models for pathological morphological features of inflammatory bowel disease require a large number of intestinal pathological images and manual annotations, increasing costs and high computing resources demand, resulting in recognition accuracy dependent on image quality and Labeling accuracy.

Method used

By constructing a window in the pathological grayscale image of intestinal biopsy sample slices, calculator probability of pixel points is calculated, and the characteristics of adjacent windows are corrected. Finally, crypt images are generated through grayscale value mapping to achieve inflammatory bowel disease The identification avoids the need for manual annotation and high-performance computing devices.

Benefits of technology

This method does not require manual annotation, and is simple to calculate, saving the cost and computing resources of crypt recognition. The acquired crypt images are more accurate, eliminating matrix interference without coloring, and providing accurate diagnostic support.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118762267B_ABST
    Figure CN118762267B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image processing technology, and in particular to a method and device for diagnosing inflammatory bowel disease based on pathological morphological features. The method includes: determining the crypt probability of a pixel point according to the grayscale distribution in the window corresponding to the pixel point in the pathological grayscale image of the intestinal biopsy sample slice, correcting the crypt probability of the pixel point according to the crypt probability of the pixel point in the adjacent window, screening the approximate window of the pixel point according to the difference between the obtained first corrected probability and the first corrected probability of the pixel point in the adjacent window, determining the second corrected probability of the pixel point according to the first corrected probability of all pixels on the line segment connecting the pixel point and the central pixel point of the approximate window, mapping the grayscale value of the pixel point according to the second corrected probability of the pixel point, thereby obtaining a crypt image, so that the doctor can identify inflammatory bowel disease according to the crypt image. The crypt image obtained by the present invention is accurate and helps doctors diagnose inflammatory bowel disease.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method and device for diagnosing inflammatory bowel disease based on pathological morphological characteristics. Background Art

[0002] Inflammatory bowel disease is an idiopathic intestinal inflammatory disease that affects the ileum, rectum, and colon. It mainly includes two main types: Crohn's Disease (CD) and Ulcerative Colitis (UC). Crohn's disease usually causes destruction of the mucosal crypt structure, such as crypt branching, irregular expansion, shortening, elongation, and loss. Ulcerative colitis usually causes crypt structural disorder, causing cryptitis, crypt abscesses, etc., characterized by crypt enlargement, destruction, and fusion. Therefore, when identifying inflammatory bowel disease, analysis of crypt characteristics is crucial.

[0003] With the development of computer technology, machine learning, deep learning and other technologies have gradually been applied to the field of inflammatory bowel disease to assist doctors in diagnosing inflammatory bowel disease. For example, the patent document with the publication number "CN117351218A" and the name "A method for identifying crypt stretching images with pathological morphological characteristics of inflammatory bowel disease" discloses a method for identifying crypt stretching images with pathological morphological characteristics of inflammatory bowel disease, including: setting labels for historical microscopic images of crypts, the labels include crypt stretching images and non-crypt stretching images, to obtain a data set; establishing a convolutional neural network model; setting a loss function and using the data set to train the convolutional neural network model to obtain an optimal model, and in response to newly captured microscopic images, generating image classification results of crypt stretching according to the optimal model.

[0004] Although the above scheme can assist doctors in identifying crypt stretching, the accuracy of the training results of the convolutional neural network model used depends on the quality of the collected intestinal pathology images and the accuracy of manual annotation. Usually a large number of intestinal pathology images are required for training, which increases the cost of collecting a large number of intestinal pathology images and the cost of manual annotation. At the same time, a large amount of computing resources is required in the training and reasoning process of the convolutional neural network model, including high-performance computing devices and large-capacity memory, which increases the computing cost. Summary of the invention

[0005] In order to solve the problem of high cost of neural network recognition of crypt features, the present invention provides a method and device for diagnosing inflammatory bowel disease based on pathological morphological features.

[0006] In a first aspect, the present invention provides a method for diagnosing inflammatory bowel disease based on pathological morphological characteristics, using the following technical solution:

[0007] A window is constructed with each pixel in the pathological grayscale image of the intestinal biopsy sample slice as the center to determine the crypt probability of the pixel, wherein the crypt probability is positively correlated with the grayscale mean of all pixels in the window corresponding to the pixel, and negatively correlated with the grayscale difference between the pixel and each pixel in its corresponding window; the crypt probability of the pixel is corrected according to the crypt probability of the pixel in the adjacent window of the window corresponding to the pixel to obtain a first corrected probability, wherein the first corrected probability is positively correlated with the crypt probability of the pixel, negatively correlated with the difference between the crypt probability of the pixel and the crypt probability of the central pixel of the adjacent window, and negatively correlated with the variance of the crypt probability of all pixels in the adjacent window; based on An approximate window of a pixel is selected based on the difference between the first corrected probability of the pixel and the first corrected probability of the pixel in an adjacent window; the pixel is connected to the central pixel of the approximate window to obtain a first line segment; the second corrected probability of the pixel is determined based on the first corrected probability of all pixels on the first line segment, the second corrected probability is positively correlated with the mean of the first corrected probabilities of the pixels on the first line segment, and negatively correlated with the difference between the first corrected probabilities of adjacent pixels on the first line segment; the grayscale value of the pixel is mapped based on the second corrected probability of the pixel, and a crypt image is obtained based on the size of the mapped grayscale value, so that doctors can identify inflammatory bowel disease based on the crypt image.

[0008] The effect is that: when the present invention performs crypt identification, there is no need to manually annotate the pathological grayscale image, the calculation method in the present invention is simple, and no high-performance computing equipment and large-capacity memory are required. Compared with the use of neural networks, the labor cost and calculation cost of crypt identification are saved; the present invention analyzes the grayscale distribution in the window corresponding to the pixel point, so that the acquired crypt probability can reflect the possibility that the pixel point is located inside the crypt area, the present invention corrects the crypt probability of the pixel point in combination with the characteristics of the pixel point in the adjacent window of the window corresponding to the pixel point, avoids the interference of the uncolored matrix on the crypt area identification, and makes the crypt image finally obtained more accurate, the present invention obtains the second corrected probability on the basis of the first corrected probability, maps the grayscale value of the pixel point according to the second corrected probability, realizes the enhancement of the contrast between the pixel point in the crypt area and the pixel point in the non-crypt area, and the crypt image obtained based on the mapped grayscale value is more accurate, eliminates the interference of the uncolored matrix, etc., and can help doctors diagnose inflammatory bowel disease and provide important biological evidence for clinicians.

[0009] Preferably, the crypt probability of the pixel point satisfies the expression: ;in, Indicates Crypt probability of each pixel; Indicates The grayscale mean of all pixels in the window corresponding to the pixel; Indicates The gray value of a pixel, Indicates The pixel point corresponds to the window Gray value of pixel; Indicates the window size; represents the hyperbolic tangent function.

[0010] The effect is that the grayscale value of the pixel point inside the crypt area is larger and the color is more uniform, and the grayscale value corresponding to the cell structure outside the crypt area is more different. The present invention analyzes the grayscale distribution in the window corresponding to the pixel point, and the crypt probability obtained can reflect the possibility that the pixel point is located inside the crypt area.

[0011] Preferably, the first modified probability satisfies the expression: ;in, Indicates The first correction probability of pixels; Indicates Crypt probability of each pixel; Indicates The pixel point corresponds to the window Crypt probability of each pixel; Indicates the window size; Indicates The pixel corresponding to the window The crypt probability of the center pixel of each adjacent window; Indicates The pixel corresponding to the window The variance of the crypt probability of all pixels in adjacent windows; Indicates The number of adjacent windows of the window corresponding to the pixel point; represents the hyperbolic tangent function.

[0012] The effect is that the present invention corrects the crypt probability of a pixel point by combining the characteristics of the pixel points in the adjacent window of the window corresponding to the pixel point, thereby avoiding the interference of the uncolored matrix on the crypt area identification, and making the finally obtained crypt image more accurate.

[0013] Preferably, the method of screening the approximate window of a pixel point based on the difference between the first revised probability of the pixel point and the first revised probability of the pixel points in an adjacent window includes: determining the approximation between the window corresponding to the pixel point and the adjacent window based on the first revised probability of the pixel point, the approximation being negatively correlated with the maximum value of the difference between the first revised probability of each pixel point in the window corresponding to the pixel point and the first revised probability of each pixel point in the adjacent window, and negatively correlated with the mean value of the difference between the first revised probabilities of all pixels in the window corresponding to the pixel point; and taking the adjacent window having the maximum approximation to the window corresponding to the pixel point as the approximate window of the pixel point.

[0014] Preferably, the approximation satisfies the expression: ;in, Indicates The window corresponding to the pixel point and its The proximity of adjacent windows; Indicates The pixel point corresponds to the window The first correction probability of pixels; Indicates The pixel corresponding to the window The first The first correction probability of pixels; Indicates the window size; represents the maximum value function; Indicates The mean of the differences in the first corrected probabilities between all the pixels in the window corresponding to the pixels; Indicates The pixel corresponding to the window The mean of the differences in the first corrected probabilities between all pixels in adjacent windows; Represents an exponential function with a natural constant as its base.

[0015] Preferably, the second modified probability satisfies the expression: ;in, Indicates Crypt probability of each pixel; Indicates The first correction probability of pixels; Indicates The mean of the first corrected probabilities of all pixels on the first line segment of pixels; Indicates The first line segment of the pixel The first correction probability of pixels; Indicates The first line segment of the pixel The first correction probability of pixels; Indicates The number of pixels on the first line segment of pixels; Represents the normalization function.

[0016] Preferably, mapping the grayscale value of the pixel point according to the second correction probability of the pixel point includes: for each pixel point in the pathological grayscale image, multiplying the grayscale value of each pixel point by its second correction probability to obtain a mapped grayscale value of each pixel point.

[0017] The effect is that the grayscale value of the pixel in the crypt area is larger, and the second correction probability of the pixel in the crypt area is larger. The second correction probability of the pixel in the non-crypt area is smaller. By multiplying the grayscale value of the pixel with its second correction probability to obtain the mapping grayscale value, the contrast between the pixel in the crypt area and the pixel in the non-crypt area is enhanced.

[0018] Preferably, the method of obtaining the crypt image according to the obtained mapping grayscale value size includes: performing Otsu threshold segmentation on the pathological grayscale image according to the mapping grayscale values ​​of all pixels in the pathological grayscale image to obtain a binary image; and multiplying the binary image with the pathological grayscale image to obtain the crypt image.

[0019] The effect is that the present invention maps the grayscale value of the pixel through the second correction probability of the pixel. The crypt image obtained based on the mapped grayscale value is more accurate, eliminating the interference of uncolored matrix, etc., which can help doctors diagnose inflammatory bowel disease and provide important biological evidence for clinicians.

[0020] In a second aspect, the present invention provides an inflammatory bowel disease diagnostic device based on pathological morphological characteristics, which adopts the following technical solution:

[0021] The inflammatory bowel disease diagnosis device based on pathological morphological characteristics comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned inflammatory bowel disease diagnosis method based on pathological morphological characteristics is implemented.

[0022] By adopting the above technical solution, the above-mentioned inflammatory bowel disease diagnosis method based on pathological morphological characteristics is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0023] The present invention has the following technical effects: the calculation method in the present invention is simple, saving the labor cost and calculation cost of crypt identification; the crypt probability of the present invention can reflect the possibility that the pixel point is located inside the crypt area, and the crypt probability is corrected to avoid the interference of the uncolored matrix on the crypt area identification, so that the finally obtained crypt image is more accurate, and the gray value of the pixel point is mapped according to the second corrected probability, so that the contrast between the pixel points in the crypt area and the pixel points in the non-crypt area is enhanced, so that the obtained crypt image is more accurate, which can help doctors diagnose inflammatory bowel disease and provide important biological evidence for clinicians. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.

[0025] Figure 1 is a flow chart of a method for diagnosing inflammatory bowel disease based on pathological morphological characteristics according to an embodiment of the present invention;

[0026] Figure 2 Schematic diagram of a pathological grayscale image of an intestinal biopsy sample slice in an embodiment of the present invention;

[0027] Figure 3 It is a schematic diagram of adjacent windows in the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0029] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0030] The present invention discloses a method for diagnosing inflammatory bowel disease based on pathological morphological characteristics, referring to Figure 1 , comprising steps S1 to S5:

[0031] S1: Pathological grayscale image of intestinal biopsy sample slice.

[0032] Intestinal biopsy samples were collected and pathological images of intestinal biopsy sample slices were taken using a high-power microscope. In order to facilitate subsequent analysis, the pathological images were converted into pathological grayscale images using a grayscale algorithm. Figure 2 Schematic diagram of a pathological grayscale image of an intestinal biopsy sample slice.

[0033] S2: Build a window with each pixel in the pathological grayscale image as the center, and determine the crypt probability of the pixel according to the grayscale distribution of the pixel in the corresponding window.

[0034] It should be noted that inflammatory bowel disease mainly includes two main types: Crohn's disease and ulcerative colitis. Crohn's disease usually causes destruction of mucosal crypt structure, such as crypt branching, irregular expansion, shortening, elongation, and loss. Ulcerative colitis usually causes disordered crypt structure, causing cryptitis, crypt abscesses, etc., characterized by crypt enlargement, destruction, and fusion. Therefore, to identify inflammatory bowel disease, it is necessary to first segment the crypts in the pathological grayscale image.

[0035] The crypt is a deep depression on the surface of the small or large intestine mucosa. In the pathological grayscale image, the crypt area is brighter, which makes the grayscale value of the pixel inside the crypt area larger. At the same time, the color inside the crypt area is more uniform, that is, the grayscale value of the pixel inside the crypt area is similar. The outside of the crypt contains a variety of intestinal cell structures. Under the action of hematoxylin-eosin staining (HE staining), the cell nucleus containing acidic substances is blue, and the cytoplasm containing alkaline substances is red. The cell structure outside the crypt area has a large difference in the corresponding grayscale value in the pathological grayscale image. Therefore, the present invention combines the grayscale value of the pixel to make a preliminary analysis of the position of the pixel.

[0036] Specifically, taking each pixel in the pathological grayscale image as the center, construct The size of the window is used as the window corresponding to each pixel. Take the window size as an example to explain the window size. The implementer can adjust the window size according to the actual implementation situation.

[0037] According to the grayscale distribution in the window corresponding to each pixel in the pathological grayscale image, the crypt probability of each pixel is determined:

[0038]

[0039] in, Indicates The crypt probability of a pixel point is larger, the larger the value is, the The closer the pixel position is to the interior of the crypt; Indicates The grayscale mean of all pixels in the window corresponding to the pixel; Indicates The gray value of a pixel, Indicates The pixel point corresponds to the window Gray value of pixel; Indicates the window size, that is, the number of pixels contained in the window; represents the hyperbolic tangent function; since the color inside the crypt is lighter, the gray value of the corresponding pixel inside the crypt is larger, so when The larger the The more pixels are likely to be inside the crypt; Indicates The sum of the grayscale differences between the pixel and the corresponding pixel in the window. Since the grayscale values ​​of the pixels inside the crypt are relatively similar, the smaller the sum of the grayscale differences, the more similar the grayscales of the pixels in the corresponding window. The more pixels there are, the more likely they are to be inside the crypt area.

[0040] At this point, the crypt probability of each pixel in the pathological grayscale image is obtained.

[0041] S3: Correcting the crypt probability of the pixel point according to the crypt probability of the pixel point in the adjacent window of the window corresponding to the pixel point to obtain a first corrected probability.

[0042] Because some extracellular matrixes are not colored under the effect of HE staining, the grayscale values ​​of these matrixes in the pathological grayscale image are larger, and the grayscale inside these matrixes is more uniform. When the pixel point is the pixel point in these matrixes, the value of the crypt probability of the pixel point is also larger, resulting in inaccuracy in directly dividing the crypt area according to the crypt probability of the pixel point. Since the area corresponding to these uncolored matrixes is smaller, and the crypt area is usually larger, in order to avoid the interference of these uncolored matrixes, the present invention combines the characteristics of the pixel points in the adjacent windows of the window corresponding to the pixel point to correct the crypt probability of the pixel point.

[0043] Specifically, for a window corresponding to any pixel point in the pathological grayscale image, the windows in the eight neighborhoods of the window are taken as the adjacent windows of the window. Figure 3 , Figure 3 Each dotted box represents a pixel, and each solid box represents the window corresponding to the pixel. Figure 3 The adjacent windows of the window of pixel point A1 are the windows corresponding to pixel points A2, A3, A4, A5, A6, A7, A8, and A9.

[0044] According to the crypt probability of the window corresponding to each pixel point in the pathological grayscale image and the pixel points in the adjacent window, the crypt probability of each pixel point is corrected to obtain the first corrected probability of each pixel point:

[0045]

[0046] in, Indicates The first correction probability of pixels; Indicates Crypt probability of each pixel; Indicates The pixel point corresponds to the window Crypt probability of each pixel; Indicates the window size, that is, the number of pixels contained in the window; Indicates The pixel corresponding to the window The crypt probability of the center pixel of each adjacent window; Indicates The pixel corresponding to the window The variance of the crypt probability of all pixels in adjacent windows; Indicates The number of adjacent windows of the window corresponding to the pixel point; represents the hyperbolic tangent function.

[0047] Because the crypt area is relatively large, When the pixel is located inside the crypt area, The window corresponding to the pixel point and all adjacent windows are more likely to be located inside the crypt area at the same time. The smaller the difference between the crypt probability of the pixel and the crypt probability of the central pixel of all adjacent windows of the window corresponding to the pixel, the smaller the variance of the crypt probability of all pixels in the adjacent windows. The area corresponding to the uncolored matrix is ​​relatively small. When the pixel is located inside the area corresponding to the uncolored matrix, the The window corresponding to the pixel point and all adjacent windows are unlikely to be located inside the area corresponding to the uncolored matrix at the same time. The difference between the crypt probability of a pixel and the crypt probability of the center pixel of the adjacent window of the window corresponding to the pixel may be large. right Correction was performed to exclude interference of non-stained matrix in the crypt area.

[0048] S4: Filter an approximate window of the pixel point according to the difference between the first corrected probability of the pixel point and the first corrected probability of the pixel points in an adjacent window, connect the pixel point with the central pixel point of the approximate window to obtain a first line segment, and determine the second corrected probability of the pixel point according to the first corrected probability of all the pixels on the first line segment.

[0049] It should be noted that since the borders of the intestinal biopsy sections in the pathological grayscale image are stained lighter and the corresponding grayscale values ​​are larger, it may also interfere with the identification of the crypts. Therefore, it is necessary to correct the first correction probability of the pixel points again to obtain a more accurate crypt area.

[0050] Specifically, according to the window corresponding to each pixel point in the pathological grayscale image and the first corrected probability of the pixel points in the adjacent window, the approximation between the window corresponding to each pixel point in the pathological grayscale image and the adjacent window is determined:

[0051]

[0052] in, Indicates The window corresponding to the pixel point and its The proximity of adjacent windows; Indicates The pixel point corresponds to the window The first correction probability of pixels; Indicates The pixel corresponding to the window The first The first correction probability of pixels; Indicates the window size, that is, the number of pixels contained in the window; represents the maximum value function, Indicates The first corrected probability of each pixel in the window corresponding to the pixel is The maximum value of the difference between the first modified probabilities of each pixel in adjacent windows; Indicates The mean of the differences in the first corrected probabilities between all the pixels in the window corresponding to the pixels; Indicates The pixel corresponding to the window The mean of the differences in the first corrected probabilities between all pixels in adjacent windows; Represents an exponential function with a natural constant as its base.

[0053] The first correction probability of the pixel point can reflect the position of the pixel point to a certain extent. When the pixel point is outside the crypt, its correction probability is smaller, and when the pixel point is inside the crypt, its correction probability is larger. Therefore Able to respond The window corresponding to the pixel point and its The maximum difference in position between adjacent windows is The more hours, the The window corresponding to the pixel point and its The adjacent windows may be located inside or outside the crypt region at the same time, and the similarity between the two windows is relatively large. The larger the The window corresponding to the pixel point and its There are differences in the positions of the two adjacent windows. One of the two windows may be located outside the crypt region, and the other window may be located inside the crypt region. Alternatively, one of the two windows may include both the inside and outside of the crypt region. In this case, the approximation of the two windows is small.

[0054] When the window is located inside the crypt region, the first correction probability of the pixels in the window is small, and the mean of the difference in the first correction probability between all the pixels in the window is also small, corresponding to When the window is outside the crypt area, the first correction probability of the pixels in the window is large, and the smaller the mean of the difference in the first correction probability between all the pixels in the window, the smaller the corresponding The smaller is the size, the greater the approximation of the two windows. When the window includes both the inside and outside of the crypt region, the greater the difference in the first correction probability between all pixels in the window, the corresponding The larger the value, the smaller the similarity between the two windows.

[0055] For the window corresponding to each pixel in the pathological grayscale image, the adjacent window with the greatest similarity is used as the approximate window of the window corresponding to each pixel.

[0056] For each pixel point in the pathological grayscale image, the pixel point is connected with the central pixel point of the approximate window of the corresponding window to obtain the first line segment of the pixel point.

[0057] Determine the second revised probability of each pixel point according to the first revised probability of all pixels on the first line segment of each pixel point:

[0058]

[0059] in, Indicates The second correction probability of pixels; Indicates The first correction probability of pixels; Indicates The mean of the first corrected probabilities of all pixels on the first line segment of pixels; Indicates The first line segment of the pixel The first correction probability of pixels; Indicates The first line segment of the pixel The first correction probability of pixels; Indicates The number of pixels on the first line segment of pixels; Represents the normalization function, which is used to calculate the value of each pixel according to its corresponding right Perform linear normalization.

[0060] For the The sum of the differences in the first correction probabilities of all two adjacent pixels on the first line segment of the pixel points. Due to the differences in the characteristics of the pixels at the boundary of the intestinal biopsy slice in the pathological grayscale image, the first correction probabilities of the two adjacent pixels on the corresponding first line segment are different. When the sum of the differences is larger, the The more likely the pixel is to be the pixel at the boundary of the intestinal biopsy slice in the pathological grayscale image, the smaller the sum of the differences is. The more likely the pixel is to be inside the crypt. The larger the The larger the mean of the first correction probabilities of all pixels on the first line segment of the pixels, the larger the The greater the probability that each pixel on the first line segment belongs to a crypt pixel, the The larger the second correction probability of the pixel point, the greater the second correction probability of the pixel point.

[0061] S5: Mapping the grayscale values ​​of the pixel points according to the second corrected probability of the pixel points, and obtaining a crypt image according to the size of the obtained mapped grayscale value, so that the doctor can identify inflammatory bowel disease according to the crypt image.

[0062] For each pixel in the pathological grayscale image, the grayscale value of each pixel is multiplied by its second corrected probability to obtain a mapped grayscale value of each pixel.

[0063] It should be noted that, since the crypt region is brighter, the grayscale value of the pixel in the crypt region is larger, and the second correction probability of the pixel in the crypt region is larger. The second correction probability of the pixel in the non-crypt region is smaller. Therefore, the mapping grayscale value obtained by multiplying the grayscale value of the pixel in the crypt region by its second correction probability is larger, while the mapping grayscale value obtained by multiplying the grayscale value of the pixel in the non-crypt region by its second correction probability is smaller. In this way, the contrast between the pixel in the crypt region and the pixel in the non-crypt region is enhanced.

[0064] According to the mapped gray values ​​of all pixels in the pathological grayscale image, the pathological grayscale image is segmented by Otsu's threshold method to obtain a binary image. In the binary image, the pixels with a grayscale value of 1 are crypt areas, and the pixels with a grayscale value of 0 are non-crypt areas. The binary image is multiplied with the pathological grayscale image to obtain a crypt image. The crypt image is provided to the doctor so that the doctor can identify inflammatory bowel disease based on the crypt image.

[0065] It should be noted that the present invention maps the grayscale values ​​of pixels through the second correction probability of the pixels. The crypt image obtained based on the mapped grayscale values ​​is more accurate, eliminating the interference of uncolored matrix, etc., which can help doctors diagnose inflammatory bowel disease and provide important biological evidence for clinicians.

[0066] An embodiment of the present invention also discloses an inflammatory bowel disease diagnosis device based on pathological morphological characteristics, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an inflammatory bowel disease diagnosis method based on pathological morphological characteristics according to the present invention is implemented.

[0067] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0068] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for diagnosing inflammatory bowel disease based on pathological morphological characteristics, characterized in that: include: A window is constructed with each pixel in the pathological grayscale image of the intestinal biopsy sample slice as the center to determine the crypt probability of the pixel, wherein the crypt probability is positively correlated with the grayscale mean of all pixels in the window corresponding to the pixel, and negatively correlated with the grayscale difference between the pixel and each pixel in its corresponding window; the crypt probability of the pixel is corrected according to the crypt probability of the pixel in the adjacent window of the window corresponding to the pixel to obtain a first corrected probability, wherein the first corrected probability is positively correlated with the crypt probability of the pixel, negatively correlated with the difference between the crypt probability of the pixel and the crypt probability of the central pixel of the adjacent window, and negatively correlated with the variance of the crypt probability of all pixels in the adjacent window; The method comprises the following steps: selecting an approximate window of a pixel point according to the difference between the first revised probability of the pixel point and the first revised probability of the pixel points in the adjacent window, wherein the approximate window is negatively correlated with the maximum value of the difference between the first revised probability of each pixel point in the window corresponding to the pixel point and the first revised probability of each pixel point in the adjacent window, and negatively correlated with the mean value of the difference between the first revised probabilities of all the pixels in the window corresponding to the pixel point; taking the adjacent window with the largest approximation to the window corresponding to the pixel point as the approximate window of the pixel point; and comparing the pixel point with the center of the approximate window. The pixels are connected to obtain a first line segment; the second revised probability of the pixel is determined according to the first revised probability of all the pixels on the first line segment, the second revised probability is positively correlated with the mean of the first revised probabilities of the pixels on the first line segment, and negatively correlated with the difference between the first revised probabilities of adjacent pixels on the first line segment; the grayscale values ​​of the pixel are mapped according to the second revised probability of the pixel, including: for each pixel in the pathological grayscale image, the grayscale value of each pixel is multiplied by its second revised probability to obtain a mapped grayscale value of each pixel, and a crypt image is obtained according to the size of the obtained mapped grayscale value, so that the doctor can identify inflammatory bowel disease based on the crypt image.

2. The method for diagnosing inflammatory bowel disease based on pathological morphological characteristics according to claim 1, characterized in that: The crypt probability of the pixel point satisfies the expression: ; in, Indicates Crypt probability of each pixel; Indicates The grayscale mean of all pixels in the window corresponding to the pixel; Indicates The gray value of a pixel, Indicates The pixel point corresponds to the window Gray value of pixel; Indicates the window size; represents the hyperbolic tangent function.

3. The method for diagnosing inflammatory bowel disease based on pathological morphological characteristics according to claim 1, characterized in that: The first modified probability satisfies the expression: ; in, Indicates The first correction probability of pixels; Indicates Crypt probability of each pixel; Indicates the window size; Indicates The pixel corresponding to the window The crypt probability of the center pixel of each adjacent window; Indicates The pixel corresponding to the window The variance of the crypt probability of all pixels in adjacent windows; Indicates The number of adjacent windows of the window corresponding to the pixel point; represents the hyperbolic tangent function.

4. The method for diagnosing inflammatory bowel disease based on pathological morphological characteristics according to claim 1, characterized in that: The approximation satisfies the expression: ; in, Indicates The window corresponding to the pixel point and its The proximity of adjacent windows; Indicates The pixel point corresponds to the window The first correction probability of pixels; Indicates The pixel corresponding to the window The first The first correction probability of pixels; Indicates the window size; represents the maximum value function; Indicates The mean of the differences in the first corrected probabilities between all the pixels in the window corresponding to the pixels; Indicates The pixel corresponding to the window The mean of the differences in the first corrected probabilities between all pixels in adjacent windows; Represents an exponential function with a natural constant as its base.

5. The method for diagnosing inflammatory bowel disease based on pathological morphological characteristics according to claim 1, characterized in that: The second modified probability satisfies the expression: ; in, Indicates The second correction probability of pixels; Indicates The first correction probability of pixels; Indicates The mean of the first corrected probabilities of all pixels on the first line segment of pixels; Indicates The first line segment of the pixel The first correction probability of pixels; Indicates The first line segment of the pixel The first correction probability of pixels; Indicates The number of pixels on the first line segment of pixels; Represents the normalization function.

6. The method for diagnosing inflammatory bowel disease based on pathological morphological characteristics according to claim 1, characterized in that: The step of obtaining a crypt image according to the obtained mapping grayscale value comprises: The pathological grayscale image is subjected to Otsu's threshold segmentation according to the mapped grayscale values ​​of all pixels in the pathological grayscale image to obtain a binary image; the binary image is multiplied with the pathological grayscale image to obtain a crypt image.

7. A diagnostic device for inflammatory bowel disease based on pathological morphological characteristics, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the method for diagnosing inflammatory bowel disease based on pathological morphological characteristics according to any one of claims 1 to 6 is implemented.