Prostate automatic pathology report system and method based on artificial intelligence

Abstract

The invention provides an automatic prostate pathology reporting system and method based on artificial intelligence, and the system comprises an image obtaining module which is used for obtaining a section image of a prostate; the image segmentation module is used for splitting and extracting the section image of the prostate; the feature extraction module is used for performing feature extraction on the tissue image; the Gleason model module is used for constructing a Gleason scoring model and inputting the first feature data into the Gleason scoring model for training and tuning; the Gleason scoring module is used for inputting the slice image to be analyzed into the optimal Gleason scoring model for scoring; and the report generation module is used for generating a pathological report based on the Gleason score of the slice image to be analyzed. The prostate pathology report is automatically generated through artificial intelligence, and diagnosis and treatment of patients about prostate diseases are greatly improved. And meanwhile, by combining a medical knowledge database and an artificial intelligence technology, better technical support is provided for doctors, and the doctors can make an optimal treatment scheme conveniently.

Description

Technical Field

[0001] The present invention relates to the technical field of pathological diagnosis, and particularly relates to an artificial intelligence-based automated prostate pathological reporting system and method. Background Art

[0002] Currently, in the medical field, prostate diseases are one of the common health problems among men. Prostate cancer is the most common type of prostate disease, and benign prostatic hyperplasia also widely exists. Accurate pathological diagnosis is crucial for the management and treatment of prostate diseases.

[0003] The traditional prostate pathological diagnosis process relies on experienced pathology experts and requires microscopic observation and analysis of tissue sections. This process is not only time-consuming and laborious, but also easily affected by subjective factors, resulting in differences in the reliability and consistency of diagnosis results. Therefore, there is an urgent need for a system that can automatically analyze the images of prostate tissue sections through machine learning and image recognition algorithms to achieve rapid, accurate, and consistent pathological diagnosis of the prostate.

[0004] The prior art CN 114638292A discloses an artificial intelligence pathological auxiliary diagnosis system based on multi-scale analysis. By utilizing multi-magnification information and fusing different magnification information, the ability to extract context information is enhanced, and concentric multi-magnification image block information is utilized. However, the prior art does not perform Gleason scoring on prostate tissue, nor does it determine the grade of the prostate based on the cancer proportion, and it does not facilitate doctors to view and correct prostate section images intelligently. Summary of the Invention

[0005] An artificial intelligence-based automated prostate pathological reporting system provided by an embodiment of the present invention includes:

[0006] An image acquisition module, configured to acquire a section image of the prostate;

[0007] An image segmentation module, configured to split and extract the section image of the prostate to determine a tissue image;

[0008] A feature extraction module, configured to extract features from the tissue image to obtain first feature data;

[0009] A Gleason model module, configured to construct a Gleason scoring model, and input the first feature data into the Gleason scoring model for training and tuning to obtain an optimal Gleason scoring model;

[0010] A Gleason scoring module, configured to input the section image to be analyzed into the optimal Gleason scoring model to obtain the Gleason score of the section image to be analyzed;

[0011] A report generation module for generating a pathological report based on the Gleason score of the slice image to be analyzed.

[0012] Preferably, the image segmentation module includes:

[0013] An image grayscale unit for converting the slice image into a grayscale image;

[0014] A grayscale statistics unit for counting the number of pixels corresponding to each gray level in the grayscale image based on the grayscale image;

[0015] An inter-class variance unit for traversing the gray levels to determine the inter-class variance corresponding to each gray level;

[0016] A split image unit for selecting the maximum inter-class variance and splitting and extracting the slice image based on the gray level corresponding to the maximum inter-class variance to determine multiple separated images;

[0017] A tissue recognition unit for performing tissue recognition and combination on the multiple separated images to determine a tissue image containing multiple tissues.

[0018] Preferably, the feature extraction module includes:

[0019] A second feature unit for performing Fourier transform on the tissue image to determine tissue second feature data;

[0020] A third feature unit for performing contour extraction on the second feature of the tissue to determine tissue third feature data;

[0021] A fourth feature unit for performing texture extraction on the second feature of the tissue to determine tissue fourth feature data;

[0022] A fifth feature unit for performing color extraction on the second feature of the tissue to determine tissue fifth feature data;

[0023] A feature fusion unit for fusing the tissue second feature, tissue third feature, tissue fourth feature, and tissue fifth feature to obtain first feature data.

[0024] Preferably, the third feature unit includes:

[0025] A sixth feature subunit for performing Gaussian filtering on the second feature data to determine sixth feature data;

[0026] A seventh feature subunit for performing gradient calculation on the sixth feature data to determine seventh feature data;

[0027] An eighth feature subunit for performing non-maximum suppression on the seventh feature data to determine eighth feature data;

[0028] A thresholding processing subunit, configured to preset a high threshold and a low threshold, and perform thresholding processing on the eighth feature data based on the high threshold and the low threshold to determine strong threshold pixels and weak threshold pixels;

[0029] A contour connection subunit, configured to perform contour connection on the strong threshold pixels and the weak threshold pixels respectively to determine the third feature data of the tissue.

[0030] Preferably, the Gleason model module includes:

[0031] A model construction unit, configured to construct a Gleason scoring model, where the Gleason scoring model includes a main pattern scoring sub-model and a secondary pattern scoring sub-model;

[0032] A model training unit, configured to input the first feature data into the Gleason scoring model for training to obtain a model training result;

[0033] A model tuning unit, configured to tune the Gleason scoring model based on the model training result to obtain an optimal Gleason scoring model.

[0034] Preferably, the report generation module includes:

[0035] A tissue proportion confirmation unit, configured to determine the tissue proportion corresponding to each Gleason score of the slice image to be analyzed based on the Gleason score;

[0036] A level scoring unit, configured to obtain a histology level table and determine the histology level based on the Gleason score of the slice image to be analyzed;

[0037] A cancer proportion confirmation unit, configured to determine the cancer proportion based on the tissue proportion corresponding to each Gleason score of the slice image to be analyzed;

[0038] An auxiliary diagnosis unit, configured to generate an auxiliary diagnosis report based on the histology level and the cancer proportion;

[0039] A cancer level unit, configured to construct a cancer proportion - cancer level rule base based on historical diagnosis data and determine the cancer level based on the cancer proportion - cancer level rule;

[0040] A doctor diagnosis unit, configured to determine a doctor diagnosis report based on the cancer level;

[0041] A pathology report unit, configured to determine a pathology report based on the auxiliary diagnosis report and the doctor diagnosis report.

[0042] Preferably, the artificial intelligence-based automated prostate pathology reporting system further includes a report viewing module for obtaining the doctor's viewing and modification instructions to adjust the pathology report.

[0043] Among them, the report viewing module includes:

[0044] A tissue image sorting unit for sorting each tissue in the tissue image based on the tissue proportion corresponding to each histological level.

[0045] A mouse arrow acquisition unit for determining the moving direction and position of the mouse arrow based on the doctor's viewing record.

[0046] A magnification display unit for obtaining the tissue score corresponding to the position of the mouse arrow in the tissue image, determining whether the tissue score corresponding to the tissue where the mouse arrow is located in the tissue image exceeds the set score threshold. If it exceeds the set score threshold, it determines the first magnification display area and makes adjustments based on the moving direction of the mouse arrow and the distance from other tissues, otherwise it does not process.

[0047] A report correction unit for obtaining the doctor's Gleason score modification instruction for the slice image to be analyzed in the pathology report and correcting the pathology report based on the Gleason score modification instruction.

[0048] Preferably, the magnification display unit includes:

[0049] A level judgment subunit for obtaining the position of the mouse arrow and judging whether the tissue score within a set first area centered on the position of the mouse arrow exceeds the set score threshold.

[0050] A magnification display area determination subunit for determining the shape of the magnification display area based on the magnification relationship between the magnification display area and the set first area if the tissue score within the set first area exceeds the set score threshold.

[0051] A magnification display area adjustment subunit for obtaining the moving direction of the mouse arrow and adjusting the position of the magnification display area so that the magnification display area meets the area position adjustment conditions.

[0052] Among them, the area position adjustment conditions include:

[0053] The magnification display area is located on the side opposite to the moving direction of the mouse arrow of the mouse arrow.

[0054] The magnification display area is set on the same side as the set first area and has a blank area between the tissue where the mouse arrow is located in the tissue image and other tissues.

[0055] An organization magnification display subunit, configured to magnify and display the organization within a set first region in a magnification display region according to a magnification ratio relationship;

[0056] A static fixation subunit, configured to obtain a magnification display confirmation instruction from a doctor for the magnification display region, and statically fix the magnified display image of the organization within the first region.

[0057] Preferably, the artificial intelligence-based prostate automated pathology reporting system further includes: a pathology report analysis module, configured to analyze a pathology report and determine a treatment plan based on the analysis result;

[0058] Wherein, the pathology report analysis module includes:

[0059] An analysis library construction unit, configured to construct a pathology report - analysis database based on historical analysis data;

[0060] A first analysis unit, configured to calculate a similarity for the pathology report - analysis database based on the pathology report, and use the analysis data with the greatest similarity to the pathology report in the pathology report - analysis result database as the first analysis result;

[0061] A second analysis unit, configured to obtain a modification instruction from a doctor for the first analysis result and determine a second analysis result;

[0062] A treatment plan library construction unit, configured to construct an analysis result - treatment plan library based on historical treatment data;

[0063] A first treatment plan unit, configured to retrieve a first treatment plan based on the second analysis result;

[0064] A second treatment plan unit, configured to obtain a modification instruction from a doctor for the first treatment plan and determine a second treatment plan;

[0065] A treatment plan sending unit, configured to send the second treatment plan to a patient.

[0066] The present invention further provides an artificial intelligence-based prostate automated pathology reporting method, including:

[0067] Obtaining a slice image of the prostate;

[0068] Splitting and extracting the slice image of the prostate to determine an organization image;

[0069] Performing feature extraction on the organization image to obtain first feature data;

[0070] Constructing a Gleason scoring model, and inputting the first feature data into the Gleason scoring model for training and tuning to obtain an optimal Gleason scoring model;

[0071] Input the slice image to be analyzed into the optimal Gleason scoring model to obtain the Gleason score of the slice image to be analyzed;

[0072] Generate a pathology report based on the Gleason score of the slice image to be analyzed.

[0073] Advantages of the present invention:

[0074] First of all, this system automatically analyzes prostate tissue slice images, extracts key features and locates abnormal areas, greatly reducing the workload of doctors and shortening the diagnosis time.

[0075] Secondly, this system uses artificial intelligence algorithms and has excellent recognition and classification capabilities, can accurately diagnose prostate pathology, reduce the influence of subjective factors, and improve the consistency and reliability of diagnosis results.

[0076] Thirdly, this system will generate a detailed pathology report, including information such as the location of abnormal areas, assessment of canceration degree, and treatment suggestions, providing more comprehensive reference for clinicians and promoting the formulation of treatment decisions.

[0077] Finally, this system can quickly process a large number of prostate tissue slice images, is suitable for the high-throughput pathology diagnosis needs in clinical practice, and improves work efficiency and diagnosis accuracy.

[0078] Other features and advantages of the present invention will be described in the subsequent specification, and partly become obvious from the specification, or be understood by implementing the present invention. The objectives and other advantages of the present invention can be realized and obtained through the structures specifically pointed out in the written specification and the drawings.

[0079] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. Description of the Drawings

[0080] The drawings are used to provide further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings:

[0081] Figure 1 It is a schematic diagram of an artificial intelligence-based prostate automated pathology reporting system in an embodiment of the present invention;

[0082] Figure 2 It is a schematic diagram for determining the cancer proportion in an embodiment of the present invention;

[0083] Figure 3 It is a schematic diagram for generating a pathology report in an embodiment of the present invention;

[0084] Figure 4Schematic diagram of a prostate automated pathology reporting method based on artificial intelligence in an embodiment of the present invention. Detailed implementation manners

[0085] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not intended to limit the present invention.

[0086] An embodiment of the present invention provides a prostate automated pathology reporting system based on artificial intelligence, as Figure 1 shown, including:

[0087] An image acquisition module 1, configured to acquire a sliced image of the prostate;

[0088] An image segmentation module 2, configured to split and extract the sliced image of the prostate to determine a tissue image;

[0089] A feature extraction module 3, configured to extract features from the tissue image to obtain first feature data;

[0090] A Gleason model module 4, configured to construct a Gleason scoring model, and input the first feature data into the Gleason scoring model for training and optimization to obtain an optimal Gleason scoring model;

[0091] A Gleason scoring module 5, configured to input the sliced image to be analyzed into the optimal Gleason scoring model to obtain the Gleason score of the sliced image to be analyzed;

[0092] A report generation module 6, configured to generate a pathology report based on the Gleason score of the sliced image to be analyzed.

[0093] The working principle and beneficial effects of the above technical solution are as follows:

[0094] The image acquisition module 1 acquires a sliced image of the prostate. The image segmentation module 2 splits and extracts the sliced image of the prostate to determine a tissue image; the feature extraction module 3 constructs a feature extraction model using a convolutional neural network, and uses the feature extraction model to extract features from the tissue image to obtain first feature data. The Gleason model module 4 constructs a Gleason scoring model using a deep neural network, and inputs the first feature data into the Gleason scoring model for training and optimization to obtain an optimal Gleason scoring model; the Gleason scoring module 5 inputs the sliced image to be analyzed into the optimal Gleason scoring model to obtain the Gleason score of the sliced image to be analyzed; the report generation module 6 generates a pathology report based on the Gleason score of the sliced image to be analyzed.

[0095] In the embodiments of the present invention, an artificial intelligence is used to automatically generate a prostate pathology report, which greatly improves the diagnosis and treatment of prostate diseases for patients. At the same time, by combining a medical knowledge database and artificial intelligence technology, better technical support is provided for doctors, facilitating doctors to make the optimal treatment plan.

[0096] In one embodiment, the image segmentation module 2 includes:

[0097] An image grayscale unit for converting the slice image into a grayscale image;

[0098] A grayscale statistics unit for counting the number of pixels corresponding to each grayscale level in the grayscale image based on the grayscale image;

[0099] An inter-class variance unit for traversing the grayscale levels to determine the inter-class variance corresponding to each grayscale level;

[0100] A split image unit for selecting the maximum inter-class variance and splitting and extracting the slice image based on the grayscale level corresponding to the maximum inter-class variance to determine a plurality of separated images;

[0101] A tissue recognition unit for performing tissue recognition and combination on the plurality of separated images to determine a tissue image containing a plurality of tissues.

[0102] The working principle and beneficial effects of the above technical solution are as follows:

[0103] The slice image is basically a color image. The image grayscale unit converts the color slice image into a grayscale image. The grayscale statistics unit counts the number of pixels corresponding to each grayscale level in the grayscale image. In this embodiment, the grayscale levels are 0 to 255, and the number of pixels corresponding to each grayscale level in the grayscale image is counted. For example, when the number of pixels corresponding to the k-th grayscale level is p k , the number of pixels corresponding to the k + 1-th grayscale level is p k+1 , ……, and the number of pixels corresponding to the 255-th grayscale level p 255 is calculated all the way. Assuming a certain grayscale level as the segmentation threshold, the pixel proportion and average grayscale of the target image and the background image are obtained. Then, according to the average grayscale of the grayscale image, the inter-class variance corresponding to each grayscale level is calculated. The split image unit selects the maximum inter-class variance and splits and extracts the slice image based on the grayscale level corresponding to the maximum inter-class variance to determine a plurality of separated images. The images of the tissues to be analyzed are included in the separated images. The tissue recognition unit performs tissue recognition and combination on the plurality of separated images to determine a tissue image containing a plurality of tissues. That is, the tissue image contains multiple tissue strips, each tissue strip contains biological cells, and the biological cells contain normal cells, inflammatory cells, and cancer cells.

[0104] In an embodiment of the present invention, the sliced image is converted into a grayscale image, and the sliced image is separated and extracted by the gray level corresponding to the maximum between-class variance, which improves the separation accuracy of the tissue image.

[0105] In one embodiment, the feature extraction module 3 includes:

[0106] A second feature unit, configured to perform a Fourier transform on the tissue image to determine tissue second feature data;

[0107] A third feature unit, configured to perform contour extraction on the second feature of the tissue to determine tissue third feature data;

[0108] A fourth feature unit, configured to perform texture extraction on the second feature of the tissue to determine tissue fourth feature data;

[0109] A fifth feature unit, configured to perform color extraction on the second frequency feature of the tissue to determine tissue fifth feature data;

[0110] A feature fusion unit, configured to fuse the tissue second feature, the tissue third feature, the tissue fourth feature, and the tissue fifth feature to obtain first feature data.

[0111] The working principle and beneficial effects of the above technical solution are as follows:

[0112] A second feature unit, configured to perform a Fourier transform on the tissue image to determine tissue second feature data. A third feature unit, configured to perform contour extraction on the second feature of the tissue to determine tissue third feature data. A fourth feature unit, configured to perform texture extraction on the second frequency feature of the tissue to determine tissue fourth feature data. A fifth feature unit, configured to perform color extraction on the second frequency feature of the tissue to determine tissue fifth feature data. A feature fusion unit, configured to fuse the tissue second feature, the tissue third feature, the tissue fourth feature, and the tissue fifth feature to obtain first feature data.

[0113] In an embodiment of the present invention, by performing a Fourier transform on the tissue image and extracting features, a premise is provided for tissue image diagnosis.

[0114] In one embodiment, the third feature unit includes:

[0115] A sixth feature subunit, configured to perform Gaussian filtering on the second feature data to determine sixth feature data;

[0116] A seventh feature subunit, configured to perform gradient calculation on the sixth feature data to determine seventh feature data;

[0117] An eighth feature subunit, configured to perform non-maximum suppression on the seventh feature data to determine eighth feature data;

[0118] A thresholding sub-unit for presetting a high threshold and a low threshold, and performing thresholding on the eighth feature data based on the high threshold and the low threshold to determine strong threshold pixels and weak threshold pixels;

[0119] A contour connection sub-unit for respectively performing contour connection on the strong threshold pixels and the weak threshold pixels to determine the third feature data of the tissue.

[0120] The working principle and beneficial effects of the above technical solution are as follows:

[0121] The sixth feature sub-unit performs Gaussian filtering on the second feature data using a set Gaussian kernel to achieve noise reduction processing of the image. The seventh feature sub-unit performs gradient calculation on the sixth feature data. The eighth feature sub-unit performs non-maximum suppression on the seventh feature data to highlight the maximum values in the image. The thresholding sub-unit presets a high threshold and a low threshold, and performs thresholding on the eighth feature data based on the high threshold and the low threshold to determine strong threshold pixels and weak threshold pixels, thereby extracting the edges of the image. The contour connection sub-unit is used to respectively perform contour connection on the strong threshold pixels and the weak threshold pixels to determine the third feature data of the tissue, that is, the contour shape of the tissue.

[0122] In the embodiment of the present invention, the contour in the image is extracted through Gaussian filtering and non-maximum suppression processing to obtain the third feature data of the tissue, improving the accuracy of contour extraction.

[0123] In one embodiment, the Gleason model module 4 includes:

[0124] A model construction unit for constructing a Gleason scoring model, where the Gleason scoring model includes a main pattern scoring sub-model and a secondary pattern scoring sub-model;

[0125] A model training unit for inputting the first feature data into the Gleason scoring model for training to obtain a model training result;

[0126] A model tuning unit for tuning the Gleason scoring model based on the model training result to obtain an optimal Gleason scoring model.

[0127] The working principle and beneficial effects of the above technical solution are as follows:

[0128] The model construction unit constructs a Gleason scoring model through a deep neural network. Among them, the Gleason scoring model includes a main pattern scoring sub-model and a secondary pattern scoring sub-model. In this embodiment, the Gleason grading method is adopted. The Gleason grading method is a method for histological grading of prostate adenocarcinoma that is widely used at present. The Gleason score includes the main pattern score and the secondary pattern score of the prostate. The model training unit inputs the first feature data into the Gleason scoring model for training to obtain a model training result. The model tuning unit tunes the Gleason scoring model based on the model training result to obtain an optimal Gleason scoring model.

[0129] In the embodiment of the present invention, an artificial intelligence is used to construct a Gleason scoring model to score the prostate, which overcomes the subjectivity of humans and improves the accuracy of the Gleason score.

[0130] In one embodiment, the report generation module 6 includes:

[0131] A tissue proportion confirmation unit for determining the tissue proportion corresponding to each Gleason score of the slice image to be analyzed based on the Gleason score;

[0132] A level scoring unit for obtaining a histological level table and determining the histological level based on the Gleason score of the slice image to be analyzed;

[0133] A cancer proportion confirmation unit for determining the cancer proportion based on the tissue proportion corresponding to each Gleason score of the slice image to be analyzed;

[0134] An auxiliary diagnosis unit for generating an auxiliary diagnosis report based on the histological level and the cancer proportion;

[0135] A cancer level unit for constructing a cancer proportion - cancer level rule base based on historical diagnosis data and determining the cancer level based on the cancer proportion - cancer level rule;

[0136] A doctor diagnosis unit for determining a doctor diagnosis report based on the cancer level;

[0137] A pathology report unit for determining a pathology report based on the auxiliary diagnosis report and the doctor diagnosis report.

[0138] The working principle and beneficial effects of the above technical solutions are as follows:

[0139] The tissue proportion confirmation unit determines the tissue proportion corresponding to each Gleason score of the slice image to be analyzed according to the Gleason score. For example, in this embodiment, the Gleason scores of the tissue are 3 and 4. It is determined that the tissue proportion of 3 points is 5.2%, and the tissue proportion of 4 points is 24%. The tissue proportion of 5 points is 0%. The tissues with a Gleason score above 3 belong to cancerous cells. Among them, the tissue with a Gleason score of 3 accounts for 17.8% of the cancerous tissue, the tissue with a Gleason score of 4 accounts for 82.2% of the cancerous tissue, and the tissue with a Gleason score of 5 accounts for 0% of the cancerous tissue.

[0140] The level scoring unit is used to preset the histological grading rules and determine the histological grade based on the histological grading rules for the Gleason score of the slice image to be analyzed, as shown in Table 1. In this embodiment, the level scoring unit determines that the histological grade is moderately differentiated according to 4 + 3 = 7 points.

[0141] Table 1 Table for determining histological grade by Gleason score

[0142]

[0143]

[0144] The cancer proportion confirmation unit determines the cancer proportion based on the tissue proportion corresponding to each Gleason score of the slice image to be analyzed. As Figure 3 shown, in this embodiment, the cancer proportion can be determined according to 3 points and 4 points, that is, equal to 5.2% + 24% = 29.2%.

[0145] The auxiliary diagnosis unit is used to generate an auxiliary diagnosis report based on the histological grade and the cancer proportion.

[0146] The cancer level unit is used to construct a cancer proportion - cancer level rule base based on historical diagnosis data and determine the cancer level based on the cancer proportion - cancer level rule. In this embodiment, according to the cancer proportion of 29.2%, it is determined as moderately graded prostate cancer. The doctor diagnosis unit is used to determine the doctor diagnosis report based on the cancer level. As Figure 3 shown, the pathology report unit determines the pathology report based on the auxiliary diagnosis report and the doctor diagnosis report.

[0147] In the embodiment of the present invention, a pathological report is generated through the Gleason score, which overcomes the high error rate of manual diagnosis, improves the accuracy of system diagnosis, and at the same time, for the tissue proportion corresponding to each Gleason score, the evaluation is more comprehensive.

[0148] In one embodiment, the prostate automated pathology reporting system based on artificial intelligence further includes a report viewing module for obtaining the doctor's viewing instruction and modification instruction to adjust the pathology report;

[0149] Among them, the report viewing module includes:

[0150] An organizational image sorting unit, which is used to sort each tissue in the tissue image based on the tissue proportion corresponding to each histological level;

[0151] A mouse arrow acquisition unit, which is used to determine the moving direction and position of the mouse arrow based on the doctor's viewing record;

[0152] An enlarged display unit, which is used to obtain the tissue score corresponding to the position of the mouse arrow in the tissue image, determine whether the tissue score corresponding to the tissue where the mouse arrow is located in the tissue image exceeds the set score threshold. If it exceeds the set score threshold, it determines the first enlarged display area and makes adjustments based on the moving direction of the mouse arrow and the distance from other tissues, otherwise it does not process;

[0153] A report correction unit, which is used to obtain the Gleason score modification instruction for the slice image to be analyzed in the pathological report by the doctor, and correct the pathological report based on the Gleason score modification instruction.

[0154] The working principle and beneficial effects of the above technical solutions are as follows:

[0155] The report viewing module facilitates the adjustment of the pathological report according to the doctor's viewing instruction and modification instruction. The organizational image sorting unit sorts each tissue in the tissue image based on the tissue proportion corresponding to each histological level. The tissue proportion corresponding to each histological level of each tissue is different. The larger the score, the higher the degree of canceration. Therefore, for the tissue with a large proportion of tissues with a large score in the tissue, it is placed in the front in the tissue image for the doctor to observe in time.

[0156] The mouse arrow acquisition unit determines the moving direction and position of the mouse arrow based on the doctor's viewing record. For example, when the doctor's viewing record slides from right to left, it can be determined that the moving direction of the mouse arrow slides to the left and the position where the mouse arrow is located.

[0157] The enlarged display unit obtains the tissue score corresponding to the position of the mouse arrow in the tissue image, determines whether the tissue score corresponding to the tissue where the mouse arrow is located in the tissue image exceeds the set score threshold. If it exceeds the set score threshold, it determines the first enlarged display area and makes adjustments based on the moving direction of the mouse arrow and the distance from other tissues, otherwise it does not process. For example, if the mouse arrow moves on a tissue with a score above 3, the enlarged display unit enlarges and displays within a set area centered on the mouse arrow. It is convenient for the doctor to observe cancer cells above 3 points.

[0158] The report correction unit obtains the Gleason score modification instruction of the doctor for the slice image to be analyzed in the pathological report, and corrects the pathological report based on the Gleason score modification instruction. For example, when the doctor finds that the tissue score is incorrect, the Gleason score of the slice image to be analyzed can be modified. The report correction unit obtains the Gleason score modification instruction of the doctor and simultaneously corrects the pathological report synchronously.

[0159] In the embodiment of the present invention, when the mouse arrow moves to the cancerous tissue area, the report viewing module enlarges and displays the cancerous tissue area, which is convenient for the doctor to observe.

[0160] In one embodiment, it is characterized in that the enlarged display unit includes:

[0161] The level judgment subunit is used to obtain the position of the mouse arrow and judge whether the tissue score within the set first area centered on the position of the mouse arrow exceeds the set score threshold;

[0162] The enlarged display area determination subunit is used to determine the shape of the enlarged display area based on the magnification relationship between the enlarged display area and the set first area if the tissue score within the set first area exceeds the set score threshold;

[0163] The enlarged display area adjustment subunit is used to obtain the moving direction of the mouse arrow and adjust the position of the enlarged display area so that the enlarged display area meets the area position adjustment condition;

[0164] Among them, the area position adjustment condition includes:

[0165] The enlarged display area is located on the side opposite to the moving direction of the mouse arrow of the mouse arrow;

[0166] The enlarged display area is set on the same side as the set first area and has a blank area between the tissue at the position of the mouse arrow in the tissue image and other tissues;

[0167] The tissue enlarged display subunit is used to enlarge and display the tissue within the set first area in the enlarged display area according to the magnification relationship;

[0168] The static fixing subunit is used to obtain the enlarged display confirmation instruction of the doctor for the enlarged display area and statically fix the enlarged display image of the tissue within the first area.

[0169] The working principle and beneficial effects of the above technical solutions are:

[0170] The level judgment subunit obtains the position of the mouse arrow and determines whether the tissue score within a set first region centered on the position where the mouse arrow is located exceeds a set score threshold. In this embodiment, the set score threshold is 3 points. Tissues with a score above 3 points belong to cancerous tissues, and the cancerous tissues need to be magnified and displayed. If the tissue score within the set first region exceeds the set score threshold, the magnified display area determination subunit determines the shape of the magnified display area based on the magnification relationship between the magnified display area and the set first region. In this embodiment, the set magnification is 5 times, and the shape is a rectangle with an aspect ratio of 2:1. The magnified display area adjustment subunit obtains the moving direction of the mouse arrow and adjusts the position of the magnified display area so that the magnified display area meets the regional position adjustment conditions. Among them, the regional position adjustment conditions include:

[0171] The magnified display area is located on the side opposite to the moving direction of the mouse arrow of the mouse arrow;

[0172] The magnified display area is set on the same side as the set first region and is the blank area between the tissue where the mouse arrow is located in the tissue image and other tissues.

[0173] In this embodiment, if the moving direction of the arrow is to move left, the magnified display area is on the right side of the mouse arrow. When the set first region is on the upper side of the tissue, and the magnified display area is also on the upper side of the tissue and is the blank area between the tissue where the mouse arrow is located and the tissue adjacent above the tissue where the mouse arrow is located. And the magnified display area does not overlap with other tissues, which is convenient for doctors to observe the cancerous tissue and other tissues.

[0174] The tissue magnified display subunit magnifies and displays the tissue within the set first region in the magnified display area according to the magnification relationship.

[0175] The static fixation subunit obtains the magnified display confirmation instruction of the doctor in the magnified display area and statically fixes the magnified display image of the tissue within the first region. That is, the area magnified and displayed by the doctor is fixed. At the same time, the doctor can also perform the same magnified display on other tissues by moving the mouse arrow.

[0176] In the embodiment of the present invention, by setting a magnified display area for cancerous cells in the tissue and not overlapping with other tissues, it is convenient for doctors to observe.

[0177] In one embodiment, the prostate automated pathology reporting system based on artificial intelligence further includes: a pathology report analysis module for analyzing the pathology report and determining a treatment plan based on the analysis result;

[0178] Among them, the pathology report analysis module includes:

[0179] An analysis library construction unit for constructing a pathological report - analysis database based on historical analysis data;

[0180] A first analysis unit for calculating the similarity of the pathological report - analysis database based on the pathological report, and taking the analysis data with the greatest similarity to the pathological report in the pathological report - analysis result database as the first analysis result;

[0181] A second analysis unit for obtaining the doctor's modification instruction for the first analysis result and determining the second analysis result;

[0182] A treatment plan library construction unit for constructing an analysis result - treatment plan library based on historical treatment data;

[0183] A first treatment plan unit for retrieving the first treatment plan based on the second analysis result;

[0184] A second treatment plan unit for obtaining the doctor's modification instruction for the first treatment plan and determining the second treatment plan;

[0185] A treatment plan sending unit for sending the second treatment plan to the patient.

[0186] The working principle and beneficial effects of the above technical solution are as follows:

[0187] The pathological report analysis module analyzes the pathological report and determines the treatment plan based on the analysis result.

[0188] The analysis library construction unit constructs a pathological report - analysis database based on historical analysis data. In this embodiment, according to the recorded pathological reports of previous patients, the pathological report - analysis database is constructed. The first analysis unit calculates the similarity of the pathological report - analysis database based on the pathological report, and takes the analysis data with the greatest similarity to the pathological report in the pathological report - analysis result database as the first analysis result. For example, input a pathological report with a Gleason score of 3 + 4 into the pathological report - analysis database. According to the similarity calculation, determine the analysis data with the greatest similarity, that is, analyze that the result is due to prostate hyperplasia leading to the middle stage of prostate cancer. The second analysis unit is used to obtain the doctor's modification instruction for the first analysis result and determine the second analysis result. The treatment plan library construction unit constructs an analysis result - treatment plan library based on historical treatment data. The first treatment plan unit retrieves the first treatment plan based on the second analysis result. In this embodiment, according to the middle stage of prostate cancer, it is determined that the patient needs to undergo clinical surgical treatment. The second treatment plan unit obtains the doctor's modification instruction for the first treatment plan and determines the second treatment plan. The treatment plan sending unit is used to send the second treatment plan to the patient.

[0189] The present invention analyzes the pathological report and determines the treatment plan, which is convenient for doctors to quickly formulate treatment plans and improves the treatment pertinence.

[0190] As Figure 4 shown, an embodiment of the present invention also provides an artificial intelligence-based automated prostate pathology reporting method, including:

[0191] Step 1: Obtain a sliced image of the prostate;

[0192] Step 2: Split and extract the sliced image of the prostate to determine a tissue image;

[0193] Step 3: Extract features from the tissue image to obtain first feature data;

[0194] Step 4: Construct a Gleason scoring model, and input the first feature data into the Gleason scoring model for training and tuning to obtain an optimal Gleason scoring model;

[0195] Step 5: Input the sliced image to be analyzed into the optimal Gleason scoring model to obtain the Gleason score of the sliced image to be analyzed;

[0196] Step 6: Generate a pathology report based on the Gleason score of the sliced image to be analyzed.

[0197] The working principle and beneficial effects of the above technical solution are as follows:

[0198] In Step 1, a sliced image of the prostate is obtained. In Step 2, the sliced image of the prostate is split and extracted to determine a tissue image. In Step 3, features are extracted from the tissue image to obtain first feature data. In Step 4, a Gleason scoring model is constructed, and the first feature data is input into the Gleason scoring model for training and tuning to obtain an optimal Gleason scoring model. In Step 5, the sliced image to be analyzed is input into the optimal Gleason scoring model to obtain the Gleason score of the sliced image to be analyzed. In Step 6, a pathology report is generated based on the Gleason score of the sliced image to be analyzed.

[0199] In an embodiment of the present invention, an artificial intelligence is used to automatically generate a prostate pathology report, which greatly improves the diagnosis and treatment of prostate diseases for patients. At the same time, by combining a medical knowledge database and artificial intelligence technology, better technical support is provided for doctors, facilitating doctors to make an optimal treatment plan.

[0200] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. An artificial intelligence-based automated prostate pathology reporting system, characterized in that, Including: An image acquisition module for acquiring slice images of the prostate; An image segmentation module for splitting and extracting the slice images of the prostate to determine tissue images; A feature extraction module for extracting features from the tissue images to obtain first feature data; A Gleason model module for constructing a Gleason scoring model and inputting the first feature data into the Gleason scoring model for training and tuning to obtain an optimal Gleason scoring model; A Gleason scoring module for inputting the slice image to be analyzed into the optimal Gleason scoring model to obtain the Gleason score of the slice image to be analyzed; A report generation module for generating a pathology report based on the Gleason score of the slice image to be analyzed.

2. The automated prostate pathological report system based on artificial intelligence according to claim 1, wherein The image segmentation module includes: An image grayscale unit for converting the slice image into a grayscale image; A grayscale statistics unit for statistically calculating the number of pixels corresponding to each gray level in the grayscale image based on the grayscale image; An inter-class variance unit for traversing the gray levels to determine the inter-class variance corresponding to each gray level; A split image unit for selecting the maximum inter-class variance and splitting and extracting the slice image based on the gray level corresponding to the maximum inter-class variance to determine multiple separated images; A tissue recognition unit for performing tissue recognition and combination on the multiple separated images to determine a tissue image containing multiple tissues.

3. The prostate automated pathology reporting system based on artificial intelligence according to claim 1, characterized in that, The feature extraction module includes: A second feature unit for performing Fourier transform on the tissue image to determine tissue second feature data; A third feature unit for performing contour extraction on the second feature of the tissue to determine tissue third feature data; A fourth feature unit for performing texture extraction on the second feature of the tissue to determine tissue fourth feature data; A fifth feature unit for performing color extraction on the second feature of the tissue to determine tissue fifth feature data; A feature fusion unit for fusing the tissue second feature, tissue third feature, tissue fourth feature, and tissue fifth feature to obtain first feature data.

4. The prostate automated pathology reporting system based on artificial intelligence according to claim 3, wherein The third feature unit includes: A sixth feature sub-unit for performing Gaussian filtering on the second feature data to determine sixth feature data; A seventh feature sub-unit for performing gradient calculation on the sixth feature data to determine seventh feature data; An eighth feature sub-unit for performing non-maximum suppression on the seventh feature data to determine eighth feature data; A thresholding processing sub-unit for presetting high and low thresholds and performing thresholding processing on the eighth feature data based on the high and low thresholds to determine strong threshold pixels and weak threshold pixels; A contour connection sub-unit for performing contour connection on the strong threshold pixels and weak threshold pixels respectively to determine tissue third feature data.

5. The automated prostate pathology reporting system based on artificial intelligence according to claim 1, wherein The Gleason model module includes: A model construction unit for constructing a Gleason scoring model, where the Gleason scoring model includes a main pattern scoring sub-model and a secondary pattern scoring sub-model; A model training unit for inputting the first feature data into the Gleason scoring model for training to obtain a model training result; A model optimization unit, configured to optimize the Gleason scoring model based on the model training results to obtain an optimal Gleason scoring model.

6. The automated prostate pathology reporting system based on artificial intelligence according to claim 1, wherein The report generation module includes: A tissue proportion confirmation unit, configured to determine the tissue proportion corresponding to each Gleason score of the slice image to be analyzed based on the Gleason score; A level scoring unit, configured to obtain a histological level table and determine the histological level based on the Gleason score of the slice image to be analyzed; A cancer proportion confirmation unit, configured to determine the cancer proportion based on the tissue proportion corresponding to each Gleason score of the slice image to be analyzed; An auxiliary diagnosis unit, configured to generate an auxiliary diagnosis report based on the histological level and the cancer proportion; A cancer level unit, configured to construct a cancer proportion - cancer level rule base based on historical diagnosis data and determine the cancer level based on the cancer proportion - cancer level rule; A doctor diagnosis unit, configured to determine a doctor diagnosis report based on the cancer level; A pathology report unit, configured to determine a pathology report based on the auxiliary diagnosis report and the doctor diagnosis report; 7. The prostate automated pathology reporting system based on artificial intelligence according to claim 1, characterized in that, It further includes a report viewing module, configured to obtain the doctor's viewing instruction and modification instruction for adjusting the pathology report; Among them, the report viewing module includes: A tissue image sorting unit, configured to sort each tissue in the tissue image based on the tissue proportion corresponding to each histological level; A mouse arrow acquisition unit, configured to determine the moving direction and position of the mouse arrow based on the doctor's viewing record; A magnification display unit, configured to obtain the tissue score corresponding to the position of the mouse arrow in the tissue image, determine whether the tissue score corresponding to the tissue where the mouse arrow is located in the tissue image exceeds the set score threshold. If it exceeds the set score threshold, determine the first magnification display area and make adjustments based on the moving direction of the mouse arrow and the distance from other tissues, otherwise do not process; A report correction unit, configured to obtain the Gleason score modification instruction for the slice image to be analyzed in the pathology report by the doctor, and correct the pathology report based on the Gleason score modification instruction; 8. The automated prostate pathology reporting system based on artificial intelligence according to claim 7, wherein The magnification display unit includes: A level judgment subunit, configured to obtain the position of the mouse arrow and judge whether the tissue score within a set first area centered on the position of the mouse arrow exceeds the set score threshold; A magnification display area determination subunit, configured to, if the tissue score within the set first area exceeds the set score threshold, determine the shape of the magnification display area based on the magnification relationship between the magnification display area and the set first area; A magnification display area adjustment subunit, configured to obtain the moving direction of the mouse arrow and adjust the position of the magnification display area so that the magnification display area meets the area position adjustment condition; Among them, the area position adjustment condition includes: The magnification display area is located on the side opposite to the moving direction of the mouse arrow of the mouse arrow; The magnification display area is set on the same side of the set first area and has a blank area between the tissue where the mouse arrow is located in the tissue image and other tissues; An organization magnification display subunit for magnifying and displaying the organization within a set first region in a magnification display region according to a magnification ratio relationship; A static fixation subunit for obtaining a magnification display confirmation instruction from a doctor for the magnified display image of the organization within the first region and statically fixing the magnified display image of the organization within the first region.

9. The prostate automated pathology report system based on artificial intelligence according to claim 1, characterized in that, It further includes: A pathological report analysis module for analyzing a pathological report and determining a treatment plan based on the analysis result; Among them, the pathological report analysis module includes: An analysis library construction unit for constructing a pathological report - analysis database based on historical analysis data; A first analysis unit for calculating the similarity of the pathological report - analysis database based on the pathological report and taking the analysis data with the greatest similarity to the pathological report in the pathological report - analysis result database as the first analysis result; A second analysis unit for obtaining a modification instruction from a doctor for the first analysis result and determining a second analysis result; A treatment plan library construction unit for constructing an analysis result - treatment plan library based on historical treatment data; A first treatment plan unit for retrieving a first treatment plan based on the second analysis result; A second treatment plan unit for obtaining a modification instruction from a doctor for the first treatment plan and determining a second treatment plan; A treatment plan sending unit for sending the second treatment plan to the patient.

10. An artificial intelligence-based automated prostate pathology reporting method, characterized in that, It includes: Obtaining a sliced image of the prostate; Splitting and extracting the sliced image of the prostate to determine a tissue image; Performing feature extraction on the tissue image to obtain first feature data; Constructing a Gleason scoring model and inputting the first feature data into the Gleason scoring model for training and tuning to obtain an optimal Gleason scoring model; Inputting the sliced image to be analyzed into the optimal Gleason scoring model to obtain the Gleason score of the sliced image to be analyzed; Generating a pathological report based on the Gleason score of the sliced image to be analyzed.

Images