A method for splitting multispectral imaging of microscopic pathology

Abstract

The present invention relates to the field of medical microscopic imaging technology, and specifically discloses a microscopic pathology multispectral imaging splitting method, including the steps of image acquisition, model preprocessing, preliminary analysis, comprehensive analysis, and extraction splitting processing; the present invention uses a multispectral imaging device to obtain the original image of the microscopic pathology slice at different wavelengths, so as to capture the detailed information of the tissue more comprehensively. Combined with the pre-trained sample splitting model and the similarity comparison model, the pathological characteristics can be more accurately identified and analyzed, thereby improving the accuracy of diagnosis. The present invention compares the pathological sample image with the corresponding monochrome images before the multispectral splitting, and selects a group of monochrome images with the highest similarity to the pathological sample image. The original image of the corresponding band is obtained according to the wavelength of the monochrome image, and the microscopic pathological structure is split and extracted according to the position contour mark, so as to realize the accurate extraction of specific components or structures in the microscopic pathological sample.

Description

Technical Field

[0001] The invention relates to the technical field of medical microscopic imaging, and in particular to a microscopic pathology multispectral imaging separation method. Background Art

[0002] Traditional pathological analysis mainly relies on microscopic observation and manual interpretation of tissue sections. Although this method is effective, it has some limitations, such as low efficiency, strong subjectivity, and poor repeatability. In recent years, with the development of computer science and image processing technology, digital pathology has gradually emerged. Digital pathology uses digital imaging technology and computer algorithms to digitize and analyze pathological sections, which can improve the accuracy, speed, and consistency of pathological analysis. However, most of the existing digital pathology technologies only focus on image analysis under a single spectrum, ignoring the different characteristics that pathological tissues may exhibit under different spectra.

[0003] Multispectral imaging is a technology that obtains images of objects at different wavelengths, which can capture more information and is of great value in improving the accuracy and comprehensiveness of pathological analysis. However, how to effectively use multispectral imaging technology for pathological analysis, especially how to split multispectral images and extract useful information, remains a challenge.

[0004] Therefore, a microscopic pathology multispectral imaging splitting method is needed, which can fully utilize the advantages of multispectral imaging to accurately analyze and process microscopic pathology samples, so as to improve the accuracy and efficiency of pathological analysis. Summary of the invention

[0005] The purpose of the present invention is to provide a microscopic pathology multispectral imaging separation method to solve the technical problems raised in the background technology.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A microscopic pathology multispectral imaging splitting method comprises the following steps:

[0008] Step 1: Image acquisition

[0009] Using a multispectral imaging device, the same imaging position of a microscopic pathological section is imaged at a plurality of different wavelengths set in advance, and original images of multiple bands are obtained;

[0010] Step 2: Model preprocessing

[0011] Extracting all pathological sample images from a preset pathological sample library and importing them into a pre-trained sample splitting model. The sample splitting model performs multi-spectral splitting on the pathological sample images according to a preset Fourier transform algorithm. Multi-spectral splitting refers to decomposing the pathological sample images into multiple monochrome images.

[0012] Step 3: Preliminary Analysis

[0013] The original images of multiple bands corresponding to the microscopic pathological sections are imported into the pre-trained similarity comparison model, and then the similarity comparison model obtains the pathological sample images one by one and performs comparison analysis processing. The comparison analysis processing refers to obtaining similar pixel points from the original image of the microscopic pathological sections and marking the position contours based on the pixel values ​​on the monochrome image corresponding to the pathological sample images;

[0014] Step 4: Comprehensive Analysis

[0015] The original image is assigned a value through the position contour mark, and at the same time, the corresponding pathological sample image is superimposed with the original image to obtain the superposition rate of each monochrome image in the pathological sample image with the original image, and then the determination coefficient is calculated based on the above data, and it is compared with the preset determination threshold, and the pathology of the microscopic pathological section is determined based on the comparison result;

[0016] Step 5: Extraction and splitting

[0017] The pathological sample image is compared with multiple monochrome images before multispectral splitting, and a group of images that are most similar to the pathological sample image in pathological features are selected from the monochrome images. Then the wavelength of this group of monochrome images is determined, and then the original image of the corresponding band is obtained according to the wavelength. Finally, the position contour marks on the original image are used to split and extract the microscopic pathological structure.

[0018] As a further solution of the present invention: in step 2, each monochromatic image is decomposed according to a plurality of different wavelengths preset by the multi-spectral imaging device.

[0019] As a further solution of the present invention: Fourier transform is to convert the pathological sample image from the spatial domain to the frequency domain, wherein the magnitude of the frequency indicates the severity of the signal change;

[0020] Multi-spectral splitting uses the properties of Fourier transform to separate different wavelengths in pathological sample images. This technology is called spectroscopic means in spectroscopy.

[0021] Subsequently, the pathological sample images separated into different wavelengths are converted back to the spatial domain through inverse Fourier transform, and the obtained image is the corresponding monochrome image.

[0022] As a further solution of the present invention: the specific method of comparison and analysis is as follows:

[0023] SS1. Select a pathological sample image and obtain a monochrome image corresponding to the pathological sample image;

[0024] SS2, select a group of original images and monochrome images with the same wavelength, obtain all corresponding pixel values ​​on the monochrome image, and construct a pixel value interval based on the maximum and minimum values ​​of the pixel values;

[0025] SS3. According to the pixel value interval, pixels within the pixel value interval are obtained in the original image, and the pixels are recorded as similar pixels;

[0026] SS4, according to the connectivity of adjacent similar pixels, obtain continuous adjacent similar pixels and obtain the number of the continuous adjacent similar pixels, then group the continuous adjacent similar pixels whose number exceeds a preset number threshold into a pixel group, and make a position contour mark on the corresponding original image according to the pixel group;

[0027] SS5. According to steps SS2 to SS4, obtain other original images and monochrome images with the same wavelength, obtain corresponding pixel groups, and make corresponding position contour marks on the corresponding original images.

[0028] As a further solution of the present invention: in SS4, the position contour mark refers to obtaining the outermost similar pixel points on all consecutive adjacent similar pixel points corresponding to the pixel group, and then surrounding the outermost similar pixel points with a coil, and the range surrounded by the coil is the position contour mark.

[0029] As a further solution of the present invention: the specific method of step 4 is as follows:

[0030] SD1, assign values ​​to the original images of each band according to the presence or absence of position contour marks on the original image;

[0031] SD2, then subtract the value of the original image of each band from the total number of original images of all bands, and divide the result by the total number of original images of all bands to obtain a corresponding quantity ratio, and record the quantity ratio as BZ;

[0032] SD3, marking corresponding position contours on a plurality of original images obtained from the corresponding pathological sample images, and superimposing them, and obtaining the superposition rate between the corresponding position contour marks of the two images;

[0033] Then, the calculated overlap ratios between the multiple two images are marked as DHi, i=1, 2, ... n, and n represents the number of overlap ratios;

[0034] SD4, pass The corresponding determination coefficient Y is obtained, where β1 and β2 are both preset proportional coefficients;

[0035] SD5. Similarly, the determination coefficients corresponding to all pathological sample images are calculated, and then the determination coefficients are compared with the preset determination thresholds:

[0036] If among all the determination coefficients, there is a determination coefficient greater than a preset determination threshold, then among all the determination coefficients greater than the preset determination threshold, a pathological sample image corresponding to a determination coefficient with the highest value is selected, and the pathology of the microscopic pathological section is the same as the pathology of the pathological sample image.

[0037] As a further solution of the present invention: in SD1, if there is no position contour mark on the original image, the value assigned to the original image is 0;

[0038] If there is a position contour mark on the original image, the value assigned to the original image is 1.

[0039] As a further solution of the present invention: the superposition method is as follows:

[0040] Each original image is used as a reference image one by one, and then the transparency of other original images is reduced, and then the other original images with reduced transparency are superimposed on the reference image one by one, and the superposition rate between the two images is calculated.

[0041] As a further solution of the present invention: the overlap ratio is obtained as follows:

[0042] The number of similar pixels in the pixel groups corresponding to the two images is obtained respectively, and is recorded as S1 and S2, and then the number of similar pixels after the two images are superimposed is obtained, and is recorded as S0;

[0043] Then through The overlap ratio DH between the two images is calculated.

[0044] As a further solution of the present invention: if all the determination coefficients are less than or equal to the preset determination threshold, it means that the microscopic pathological section has no obvious pathology and its corresponding tissue part is normal.

[0045] Beneficial effects of the present invention:

[0046] Improve diagnostic accuracy: The present invention can capture the detailed information of tissues more comprehensively by using multispectral imaging equipment to obtain the original images of microscopic pathological sections at different wavelengths. Combined with the pre-trained sample splitting model and similarity comparison model, it can more accurately identify and analyze pathological features, thereby improving the accuracy of diagnosis.

[0047] Enhanced pathological analysis capabilities: The present invention uses Fourier transform to perform multi-spectral splitting, which can decompose the pathological sample image into multiple monochrome images, and then analyze the pathological characteristics under different wavelengths. This helps to reveal subtle changes in the pathological process and provide richer information for research.

[0048] Achieve accurate extraction of specific components or structures: The present invention compares the pathological sample image with the corresponding monochrome images before multiple multispectral splitting, and selects a group of monochrome images with the highest similarity to the pathological sample image. The original image of the corresponding band is obtained according to the wavelength of the monochrome image, and the microscopic pathological structure is split and extracted according to the position contour mark. This step can achieve accurate extraction of specific components or structures in microscopic pathological samples, providing accurate data support for subsequent research or diagnosis.

[0049] Assisting disease diagnosis and treatment: The present invention can more accurately identify pathological features by performing multispectral imaging splitting on microscopic pathological sections, thereby providing strong support for the diagnosis and treatment of diseases.

[0050] Improve image quality: De-noising can reduce random noise in the image, making the image clearer and easier to analyze and process. Edge enhancement can highlight important features in the image, such as cell boundaries and tissue contours, which helps to better identify and analyze these features.

[0051] Enhance algorithm robustness: By adjusting the brightness and contrast, the visual effect of the image can be made closer to the actual pathological tissue state, which helps doctors or researchers to judge pathological changes more accurately, thereby improving the accuracy of diagnosis. De-noising and image enhancement processing can reduce misleading information in the image, making subsequent image analysis and recognition algorithms more stable and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The present invention will be further described below in conjunction with the accompanying drawings.

[0053] Figure 1 It is a schematic diagram of the process of a microscopic pathology multispectral imaging splitting method of the present invention.

[0054] Figure 2 It is a schematic diagram of the process of comparison and analysis of a microscopic pathology multispectral imaging splitting method of the present invention. DETAILED DESCRIPTION

[0055] 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 only 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] Embodiment 1

[0057] See also Figure 1 and Figure 2 As shown, the present invention is a microscopic pathology multispectral imaging separation method, comprising the following steps:

[0058] Step 1: Image acquisition

[0059] Using multispectral imaging equipment to image microscopic pathological sections at multiple pre-set wavelengths to obtain original images of multiple bands;

[0060] Among them, the original images of multiple bands have the same imaging position for the microscopic pathological sections;

[0061] In this embodiment, the multispectral imaging device includes, but is not limited to, multiple light sources such as visible light, infrared light, ultraviolet light, and X-rays;

[0062] Step 2: Model preprocessing

[0063] Extract all pathological sample images from a pre-set pathological sample library and import them into the pre-trained sample splitting model;

[0064] In this embodiment, the pathological sample image refers to a pathological image material prepared in advance for the purpose of diagnosis, analysis or machine learning in medical research;

[0065] The sample splitting model performs multi-spectral splitting on the pathological sample image according to a preset Fourier transform algorithm. Multi-spectral splitting refers to decomposing the pathological sample image into multiple monochrome images.

[0066] Wherein, each monochromatic image is decomposed according to a plurality of different wavelengths pre-set by the multispectral imaging device;

[0067] In this embodiment, Fourier transform is to convert the pathological sample image from the spatial domain to the frequency domain, wherein the magnitude of the frequency indicates the severity of the signal change;

[0068] Multi-spectral splitting uses the properties of Fourier transform to separate different wavelengths in pathological sample images. This technology is called spectroscopic means in spectroscopy.

[0069] Then, the pathological sample images separated into different wavelengths are converted back to the spatial domain through inverse Fourier transform, and the obtained image is the corresponding monochrome image;

[0070] This embodiment uses Fourier transform to perform multi-spectral splitting, which can decompose the pathological sample image into multiple monochrome images, and then analyze the pathological characteristics under different wavelengths. This helps to reveal subtle changes in the pathological process and provide richer information for research.

[0071] Step 3: Preliminary Analysis

[0072] The original images of multiple bands corresponding to the micro-pathological sections are imported into the pre-trained similarity comparison model, and then the similarity comparison model obtains the pathological sample images one by one and performs comparison analysis;

[0073] The method is as follows:

[0074] SS1. Taking a pathological sample image as an example, obtaining a monochrome image corresponding to the pathological sample image;

[0075] SS2, select a group of original images and monochrome images with the same wavelength, obtain all corresponding pixel values ​​on the monochrome image, and construct a pixel value interval based on the maximum and minimum values ​​of the pixel values;

[0076] SS3. According to the pixel value interval, pixels within the pixel value interval are obtained in the original image, and the pixels are recorded as similar pixels;

[0077] SS4, according to the connectivity of adjacent similar pixels, obtain continuous adjacent similar pixels and obtain the number of the continuous adjacent similar pixels, then group the continuous adjacent similar pixels whose number exceeds a preset number threshold into a pixel group, and make a position contour mark on the corresponding original image according to the pixel group;

[0078] The position contour mark refers to obtaining the outermost similar pixel points on all the similar pixel points that are continuously adjacent to each other in the pixel group, and then forming a coil around the outermost similar pixel points, and the range surrounded by the coil is the position contour mark;

[0079] The quantity threshold is a different quantity value pre-set according to different image resolutions, and is used to prevent the credibility of pixel group division from being affected by image noise;

[0080] SS5. Obtain other original images and monochrome images with the same wavelength in the same manner as steps SS2 to SS4, obtain corresponding pixel groups, and make corresponding position contour marks on the corresponding original images;

[0081] This embodiment uses a multispectral imaging device to obtain the original images of microscopic pathological sections at different wavelengths, which can capture the detailed information of the tissue more comprehensively. Combined with the pre-trained sample splitting model and similarity comparison model, it can more accurately identify and analyze pathological features, thereby improving the accuracy of diagnosis;

[0082] Step 4: Comprehensive Analysis

[0083] SD1, according to the presence or absence of position contour marks on the original image, add values ​​to the original image of each band;

[0084] If there is no position contour mark on the original image, the value assigned to the original image is 0;

[0085] If there is a position contour mark on the original image, the value of the original image is 1;

[0086] SD2, then subtract the value of the original image of each band from the total number of original images of all bands, and divide the result by the total number of original images of all bands to obtain a corresponding quantity ratio, and record the quantity ratio as BZ;

[0087] SD3, marking corresponding position contours on a plurality of original images obtained from the corresponding pathological sample images, and superimposing them, and obtaining the superposition rate between the corresponding position contour marks of the two images;

[0088] Then, the calculated overlap ratios between the multiple two images are marked as DH i, where i = 1, 2, ... n, and n represents the number of overlap ratios;

[0089] The overlapping method is as follows:

[0090] Each original image is used as a reference image one by one, then the transparency of other original images is reduced, then the other original images with reduced transparency are superimposed on the reference image one by one, and the superposition rate between the two images is calculated;

[0091] The overlap ratio is calculated as follows:

[0092] The number of similar pixels in the pixel groups corresponding to the two images is obtained respectively, and is recorded as S1 and S2, and then the number of similar pixels after the two images are superimposed is obtained, and is recorded as S0;

[0093] Then through The overlap ratio DH between the two images is calculated;

[0094] SD4, pass The corresponding determination coefficient Y is obtained, where β1 and β2 are both preset proportional coefficients;

[0095] SD5. Similarly, the determination coefficients corresponding to all pathological sample images are calculated, and then the determination coefficients are compared with the preset determination thresholds:

[0096] If all the determination coefficients are less than or equal to the preset determination threshold, it means that the microscopic pathology section has no obvious pathology and the corresponding tissue part is normal;

[0097] If among all the determination coefficients, there is a determination coefficient greater than a preset determination threshold, then among all the determination coefficients greater than the preset determination threshold, a pathological sample image corresponding to a determination coefficient with the highest value is selected, and the pathology of the microscopic pathological section is the same as that of the pathology of the pathological sample image;

[0098] Step 5: Extraction and splitting

[0099] Acquire multiple monochrome images corresponding to the pathological sample images with the same pathology of the microscopic pathological sections;

[0100] Then, the pathological sample image before multispectral splitting is compared with the corresponding multiple monochrome images before multispectral splitting, and a group of monochrome images whose texture features and shape features of the pathology are most similar to the pathological sample image are selected from the multiple monochrome images;

[0101] Then, the wavelength of the monochromatic image is obtained, and the original image of the corresponding band is obtained according to the wavelength, and then the microscopic pathological structure is split and extracted according to the position contour mark corresponding to the original image of the band.

[0102] In addition, this method can also be used to evaluate treatment effects, monitor disease changes, and provide a basis for clinicians to develop personalized treatment plans.

[0103] This embodiment achieves accurate extraction of specific components or structures in microscopic pathological samples, which can provide a large amount of valuable data for medical research, help reveal the mechanism of disease occurrence and development, and promote scientific research in related fields. At the same time, this method can also be applied to fields such as drug development and biological product testing, providing technical support for medical research and application.

[0104] Embodiment 2

[0105] As the second embodiment of the present invention, when the present application is implemented, compared with the first embodiment, the technical solution of the present embodiment is different from that of the first embodiment only in that:

[0106] In this embodiment, it is also used to remove noise, enhance edges, and adjust brightness and contrast of the original images of each band acquired;

[0107] in,

[0108] The denoising processing method uses any one of a mean filter, a median filter, and a Gaussian filter to denoise the original image, and the mean filter, the median filter, and the Gaussian filter are existing technologies;

[0109] The mean filter is a linear filter that replaces the anchor pixel value by calculating the average value of all pixels in the window area;

[0110] The median filter is a nonlinear filtering technique that replaces the value of each pixel with the median of all pixel values ​​within its neighborhood window;

[0111] The Gaussian filter is also a linear filter that uses normal distribution as weights to perform weighted averaging on the pixels within the window;

[0112] The enhanced edge processing method uses any one of a differential operator and a Laplace operator to highlight the edge in the original image, and the differential operator and the Laplace operator are existing algorithms;

[0113] The brightness adjustment and contrast processing methods both use histogram equalization to achieve brightness adjustment and contrast enhancement of the original image, and histogram equalization is an existing technology.

[0114] This embodiment can reduce random noise in the image through denoising, making the image clearer and facilitating subsequent analysis and processing. Enhanced edge processing can highlight important features in the image, such as cell boundaries, tissue contours, etc., which helps to better identify and analyze these features.

[0115] By adjusting the brightness and contrast, the visual effect of the image can be made closer to the actual pathological tissue state, which helps doctors or researchers to judge pathological changes more accurately, thereby improving the accuracy of diagnosis. De-noising and image enhancement processing can reduce misleading information in the image, making subsequent image analysis and recognition algorithms more stable and reliable.

[0116] Embodiment 3

[0117] As the third embodiment of the present invention, when the present application is specifically implemented, compared with the first and second embodiments, the technical solution of this embodiment is to combine the solutions of the first and second embodiments mentioned above for implementation.

[0118] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0119] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A microscopic pathology multispectral imaging splitting method, characterized in that: The following steps are involved: Step 1: Use a multispectral imaging device to image the same imaging position of a microscopic pathological section at a plurality of pre-set wavelengths, and obtain original images of multiple bands; Step 2: extract all pathological sample images from a pre-set pathological sample library and import them into a pre-trained sample splitting model. The sample splitting model performs multi-spectral splitting on the pathological sample images according to a pre-set Fourier transform algorithm. Multi-spectral splitting refers to decomposing the pathological sample images into multiple monochrome images. Step 3: Import the original images of multiple bands corresponding to the microscopic pathological sections into the pre-trained similarity comparison model, and then the relative comparison model obtains the pathological sample images one by one and performs comparison analysis processing. The comparison analysis processing refers to obtaining similar pixel points from the original image of the microscopic pathological section and marking the position contours based on the pixel values ​​on the monochrome image corresponding to the pathological sample image; Step 4: assign values ​​to the original image through position contour marking, and at the same time, overlap the corresponding pathological sample image with the original image to obtain the overlap rate of each monochrome image in the pathological sample image with the original image, and then calculate the determination coefficient based on the above data, and compare it with the preset determination threshold, and determine the pathology of the microscopic pathological section based on the comparison result; The specific method is as follows: SD1, assign values ​​to the original images of each band according to whether there are position contour marks on the original image; if there are no position contour marks on the original image, the value assigned to the original image is 0; if there are position contour marks on the original image, the value assigned to the original image is 1; SD2, then subtract the value of the original image of each band from the total number of original images of all bands, and divide the result by the total number of original images of all bands to obtain a corresponding quantity ratio, and record the quantity ratio as BZ; SD3, marking corresponding position contours on a plurality of original images obtained from the corresponding pathological sample images, and superimposing them, and obtaining the superposition rate between the corresponding position contour marks of the two images; Then, the calculated overlap ratios between the multiple two images are marked as DH i, where i = 1, 2, ... n, and n represents the number of overlap ratios; SD4, pass The corresponding determination coefficient Y is obtained, where β1 and β2 are both preset proportional coefficients; SD5. Similarly, the determination coefficients corresponding to all pathological sample images are calculated, and then the determination coefficients are compared with the preset determination thresholds: If among all the determination coefficients, there is a determination coefficient greater than a preset determination threshold, then among all the determination coefficients greater than the preset determination threshold, a pathological sample image corresponding to a determination coefficient with the highest value is selected, and the pathology of the microscopic pathological section is the same as that of the pathology of the pathological sample image; Step 5: Compare the pathological sample image with multiple monochrome images before multispectral splitting, and select a group of images that are most similar to the pathological sample image in pathological features from the monochrome images, then determine the wavelength of this group of monochrome images, and then obtain the original image of the corresponding band based on the wavelength, and finally split and extract the microscopic pathological structure through the position contour marks on the original image.

2. A microscopic pathology multispectral imaging separation method according to claim 1, characterized in that: In step 2, each monochromatic image is decomposed according to a plurality of different wavelengths preset by the multispectral imaging device.

3. A microscopic pathology multispectral imaging separation method according to claim 1, characterized in that: Fourier transform is to convert the pathological sample image from the spatial domain to the frequency domain, where the magnitude of the frequency indicates the severity of the signal change; Multi-spectral splitting uses the properties of Fourier transform to separate different wavelengths in pathological sample images. This technology is called spectroscopic means in spectroscopy. Subsequently, the pathological sample images separated into different wavelengths are converted back to the spatial domain through inverse Fourier transform, and the obtained image is the corresponding monochrome image.

4. A microscopic pathology multispectral imaging separation method according to claim 1, characterized in that: The specific method of comparison and analysis is as follows: SS1. Select a pathological sample image and obtain a monochrome image corresponding to the pathological sample image; SS2, select a group of original images and monochrome images with the same wavelength, obtain all corresponding pixel values ​​on the monochrome image, and construct a pixel value interval based on the maximum and minimum values ​​of the pixel values; SS3. According to the pixel value interval, pixels within the pixel value interval are obtained in the original image, and the pixels are recorded as similar pixels; SS4, according to the connectivity of adjacent similar pixels, obtain continuous adjacent similar pixels and obtain the number of the continuous adjacent similar pixels, then group the continuous adjacent similar pixels whose number exceeds a preset number threshold into a pixel group, and make a position contour mark on the corresponding original image according to the pixel group; SS5. According to steps SS2 to SS4, obtain other original images and monochrome images with the same wavelength, obtain corresponding pixel groups, and make corresponding position contour marks on the corresponding original images.

5. A microscopic pathology multispectral imaging separation method according to claim 4, characterized in that: In SS4, the position contour mark refers to obtaining the outermost similar pixel points on all consecutive adjacent similar pixel points corresponding to the pixel group, and then surrounding the outermost similar pixel points with a coil, and the range surrounded by the coil is the position contour mark.

6. A microscopic pathology multispectral imaging separation method according to claim 1, characterized in that: The overlapping method is as follows: Each original image is used as a reference image one by one, and then the transparency of other original images is reduced, and then the other original images with reduced transparency are superimposed on the reference image one by one, and the superposition rate between the two images is calculated.

7. A microscopic pathology multispectral imaging separation method according to claim 6, characterized in that: The overlap ratio is calculated as follows: The number of similar pixels in the pixel groups corresponding to the two images is obtained respectively, and is recorded as S1 and S2, and then the number of similar pixels after the two images are superimposed is obtained, and is recorded as S0; Then through The overlap ratio DH between the two images is calculated.

8. A microscopic pathology multispectral imaging separation method according to claim 1, characterized in that: In SD5, if all the determination coefficients are less than or equal to the preset determination threshold, it means that the microscopic pathology section has no obvious pathology and its corresponding tissue part is normal.

Images