Digital pathological section auxiliary diagnosis interpretation system and method

Abstract

The invention provides a digital pathological section auxiliary diagnosis interpretation system and method, and the system comprises a collection module which is used for obtaining the annotation information of a doctor on a digital pathological section, and extracting a plurality of key features contained in the digital pathological section based on the annotation information, and an interpretation module which is used for analyzing each key feature, the digital pathological section processing module is used for setting up key features corresponding to the digital pathological section, establishing feature information corresponding to each key feature and converting each feature information into a preset text format to obtain text interpretation corresponding to each key feature, and the execution module is used for marking each text interpretation in a section area corresponding to the digital pathological section. The digital pathological section auxiliary diagnosis system comprises a statistical module used for counting text interpretation corresponding to each section area in a digital pathological section and establishing a diagnosis result of the digital pathological section, and a display module used for displaying the digital pathological section and displaying the diagnosis result corresponding to the digital pathological section. And a user is helped to better diagnose the slice.

Description

Technical Field

[0001] The present invention relates to the technical field of pathological slice analysis, and in particular to a digital pathological slice auxiliary diagnosis interpretation system and method. Background Art

[0002] With the development of medical technology, digital pathology slide auxiliary diagnosis system plays an increasingly important role in the field of pathology diagnosis. Traditional pathology diagnosis relies on the observation and judgment of pathologists on tissue slides, but this method has problems such as subjectivity and high labor costs. The digital pathology slide auxiliary diagnosis system is based on a deep learning algorithm model. By training a large amount of annotated pathology slide data, it can efficiently and accurately predict the cancer area in the pathology slide, providing doctors with important auxiliary diagnosis information.

[0003] The digital pathology slice auxiliary diagnosis system has broad prospects and importance in the field of pathology. It can greatly improve the accuracy and efficiency of pathology diagnosis and provide a scientific basis for doctors' decision-making. However, the existing system can only give simple binary classification results after predicting the cancer area, and lacks explanation and description of the results. This brings troubles to doctors, who need to understand the reasons and basis behind the prediction results in order to better understand and apply these results. Therefore, a digital pathology slice auxiliary diagnosis interpretation system that can explain the prediction results has important research and application value.

[0004] The current artificial intelligence pathology-assisted diagnosis system mainly relies on deep learning algorithm models to predict cancer areas, but the predictive results are poorly interpretable. Traditional binary classification results or heat maps cannot intuitively tell doctors why the area is predicted to be a cancer area and the characteristics of the predicted cancer area classification. This lack of explanation limits doctors' understanding and application of the prediction results and reduces the credibility and reliability of the auxiliary diagnosis system. In addition, doctors may have doubts or uncertainties when faced with the prediction results, and cannot deeply understand the basis and reasons for the prediction results, which has a certain impact on the accuracy and reliability of pathological diagnosis.

[0005] Therefore, the present invention proposes a cancer region classification prediction and interpretation function for a digital pathology section auxiliary diagnosis interpretation system, which guides and educates users by explaining and predicting the reasons for corresponding cancer regions and classifications, thereby better assisting users in diagnosing pathology sections.

[0006] Therefore, the present invention provides a digital pathology section auxiliary diagnosis interpretation system and method. Summary of the invention

[0007] The present invention provides a digital pathology slice auxiliary diagnosis interpretation system and method, which helps users to better diagnose slices by providing interpretable evidence of the prediction results of the digital pathology slice auxiliary diagnosis system.

[0008] The present invention provides a digital pathology section auxiliary diagnosis interpretation system, comprising:

[0009] An acquisition module, used to obtain the doctor's annotation information on the digital pathology slice, and extract several key features contained in the digital pathology slice based on the annotation information;

[0010] An interpretation module, used to analyze each of the key features, establish feature information corresponding to each key feature, convert each feature information into a preset text format, and obtain a text interpretation corresponding to each key feature;

[0011] An execution module, used for marking each text explanation in a slice area corresponding to the digital pathology slice, counting the text explanation corresponding to each slice area in the digital pathology slice, and establishing a diagnosis result of the digital pathology slice;

[0012] The display module is used to display the digital pathological slice and the diagnosis result corresponding to the digital pathological slice.

[0013] In one practicable manner,

[0014] The storage module is used to obtain the diagnosis result corresponding to each digital pathology slice and store the digital pathology slice and the corresponding diagnosis result.

[0015] In one practicable manner,

[0016] The acquisition module comprises:

[0017] An information collection unit is used to collect all the annotated information on the digital pathological slice by the doctor, respectively obtain the amount of information contained in each annotated information, classify the annotated information with the same amount of information into one category, and obtain a plurality of information categories;

[0018] An information processing unit, configured to establish a convolution number for a corresponding information class based on the amount of information corresponding to the labeled information in each information class, and to perform convolution of a corresponding number of convolutions on each information class using a preset convolutional neural network to generate information attributes corresponding to each information class;

[0019] A feature generation unit, configured to determine the expression type corresponding to the information class based on the information attribute, and establish texture sub-features, shape sub-features, and edge sub-features of the digital pathological slice;

[0020] The feature establishment unit is used to perform feature enhancement on the texture sub-features, shape sub-features and edge sub-features respectively, and establish key features according to the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information.

[0021] In one practicable manner,

[0022] The interpretation module comprises:

[0023] an analysis and matching unit, configured to perform parameter analysis on all the annotation information on the digital pathology slice to obtain cell parameters corresponding to the digital pathology slice, determine the cancer attribute corresponding to the digital pathology slice based on the cell parameters, and obtain a corresponding semantic vocabulary according to the cancer attribute;

[0024] a marking processing unit, used to mark each key feature in a corresponding digital pathology slice, obtain a cross relationship between different key features, establish an associated label for the corresponding key feature based on the cross relationship, mark each associated label in the digital pathology slice, and obtain a target digital pathology slice;

[0025] A semantic conversion unit, configured to convert key features contained in the target digital pathology slice into corresponding text information based on the semantic word library in combination with preset semantic conversion rules, and to perform text fusion on the text information according to association tags between different key features to obtain feature information corresponding to the key features;

[0026] The format conversion unit is used to convert the feature information into a preset text format to obtain a text explanation corresponding to each key feature.

[0027] In one practicable manner,

[0028] The execution module comprises:

[0029] A first execution unit is used to divide the digital pathology slice into a plurality of slice regions, map the text interpretation into the digital pathology slice, and obtain the slice region into which each text interpretation falls;

[0030] The second execution unit is used to regard a plurality of text interpretations falling into the same slice area as a text set, and perform word segmentation processing on the text interpretations contained in each text set based on medical rules to obtain a plurality of words corresponding to each text set;

[0031] The third execution unit is used to reorganize a number of words corresponding to the same text set to obtain an overall text interpretation of each slice area, count the overall text interpretation corresponding to each slice area, and generate a diagnosis result of the digital pathology slice.

[0032] In one practicable manner,

[0033] The display module comprises:

[0034] An instruction unit, used for receiving a viewing instruction input by a user;

[0035] A receiving unit, configured to receive the diagnosis result generated by the execution module;

[0036] The display unit is used to retrieve and display the corresponding viewing content based on the viewing instruction of the user, and to display the diagnosis result.

[0037] In one practicable manner,

[0038] The feature establishment unit comprises:

[0039] A first establishing subunit is used to downsample the texture sub-feature to obtain a plurality of sampling points, connect the sampling points to establish a texture contour, and use the texture contour to perform enhancement training on the texture sub-feature to obtain a corresponding enhanced texture sub-feature;

[0040] A second establishing subunit is used to rotate and flip the shape sub-feature to obtain feature parameters of the shape sub-feature at different angles, traverse the digital pathology slice using the feature parameters, extract target feature parameters contained in the digital pathology slice, and correct the shape sub-feature based on the target feature parameters to obtain a corresponding enhanced corrected sub-feature;

[0041] The third establishing subunit is used to mark the edge sub-feature in the digital pathology slice, translate the edge sub-feature in the digital pathology slice, obtain the overlapping relationship between the digital pathology slice and the edge sub-feature, obtain the leak-filling edge contained in the digital pathology slice, and correct the edge sub-feature based on the position of the leak-filling edge on the digital pathology slice to obtain the enhanced edge sub-feature.

[0042] The fourth establishing sub-unit is used to obtain the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information to establish key features.

[0043] The present invention provides a digital pathology section auxiliary diagnosis interpretation method, comprising:

[0044] Step 1: Obtain the doctor's annotation information on the digital pathology slice, and extract several key features contained in the digital pathology slice based on the annotation information;

[0045] Step 2: Analyze each of the key features respectively, establish feature information corresponding to each key feature, convert each feature information into a preset text format, and obtain a text explanation corresponding to each key feature;

[0046] Step 3: Mark each text explanation in the slice area corresponding to the digital pathology slice, count the text explanations corresponding to each slice area in the digital pathology slice, and establish the diagnosis result of the digital pathology slice;

[0047] Step 4: Display the digital pathology slice and the diagnosis result corresponding to the digital pathology slice.

[0048] In one practicable manner,

[0049] The step 1 comprises:

[0050] Step 11: Collect all the annotation information on the digital pathology slide by the doctor, obtain the amount of information contained in each annotation information, classify the annotation information with the same amount of information into one category, and obtain several information categories;

[0051] Step 12: Based on the amount of information corresponding to the labeled information in each information class, establish the number of convolutions for the corresponding information class, and use the preset convolutional neural network to perform convolutions for each information class for the corresponding number of convolutions to generate information attributes corresponding to each information class;

[0052] Step 13: determining the expression type corresponding to the information class based on the information attribute, and establishing the texture sub-feature, shape sub-feature and edge sub-feature of the digital pathological slice;

[0053] Step 14: Feature enhancement is performed on the texture sub-features, shape sub-features and edge sub-features respectively, and key features are established according to the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information.

[0054] In one practicable manner,

[0055] The step 14 comprises:

[0056] Step 141: downsampling the texture sub-feature to obtain a plurality of sampling points, connecting the sampling points to establish a texture contour, and using the texture contour to perform enhancement training on the texture sub-feature to obtain a corresponding enhanced texture sub-feature;

[0057] Step 142: rotating and flipping the shape sub-feature to obtain feature parameters of the shape sub-feature at different angles, traversing the digital pathology slice using the feature parameters, extracting target feature parameters contained in the digital pathology slice, and correcting the shape sub-feature based on the target feature parameters to obtain a corresponding enhanced corrected sub-feature;

[0058] Step 143: Mark the edge sub-feature in the digital pathology slice, translate the edge sub-feature in the digital pathology slice, obtain the overlapping relationship between the digital pathology slice and the edge sub-feature, obtain the leak-filling edge contained in the digital pathology slice, and correct the edge sub-feature based on the position of the leak-filling edge on the digital pathology slice to obtain the enhanced edge sub-feature.

[0059] Step 144: Acquire the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information to establish key features.

[0060] The beneficial effects that can be achieved by the present invention are as follows: in order to realize the interpretation of digital pathology slices, the key features of the digital pathology slices are first established according to the annotation information of the doctor on the digital pathology slices, and then the key features are processed to obtain the feature information corresponding to each key feature and create a text explanation corresponding to each key feature, which is then marked in the digital pathology slices to obtain the text explanation corresponding to each slice area, thereby establishing the diagnosis result of the digital pathology slices, and finally the display module displays the diagnosis result, and the user can judge the lesion information occurring in the digital pathology slices according to the diagnosis result, thereby helping the user understand the condition.

[0061] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0062] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The accompanying drawings are used to provide a 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 of the present invention. In the accompanying drawings:

[0064] Figure 1 A schematic diagram of the composition of a digital pathology section auxiliary diagnosis and interpretation system according to an embodiment of the present invention;

[0065] Figure 2 The figure is a schematic diagram of the workflow of a method for auxiliary diagnosis and interpretation of digital pathological sections in an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0067] Example 1

[0068] This embodiment provides a digital pathology slice auxiliary diagnosis interpretation system, such as Figure 1 As shown, including:

[0069] An acquisition module, used to obtain the doctor's annotation information on the digital pathology slice, and extract several key features contained in the digital pathology slice based on the annotation information;

[0070] An interpretation module, used to analyze each of the key features, establish feature information corresponding to each key feature, convert each feature information into a preset text format, and obtain a text interpretation corresponding to each key feature;

[0071] An execution module, used for marking each text explanation in a slice area corresponding to the digital pathology slice, counting the text explanation corresponding to each slice area in the digital pathology slice, and establishing a diagnosis result of the digital pathology slice;

[0072] The display module is used to display the digital pathological slice and the diagnosis result corresponding to the digital pathological slice.

[0073] In this example, the number of annotation information included in a digital pathology slide may be one or more;

[0074] In this example, one key feature corresponds to one feature information, wherein the key feature represents the features displayed in different regions of the digital pathology slice, and the feature information represents the information used to describe the key feature;

[0075] In this example, the default text format can be Chinese, and the user can also adjust different text formats as needed;

[0076] In this example, the textual explanations corresponding to different key features are not necessarily the same;

[0077] In this example, one textual interpretation can explain one or more slice regions, and one slice region can correspond to one or more textual interpretations;

[0078] In this example, the diagnosis result includes: shape characteristics, size characteristics, characteristics of infiltrating surrounding tissues, texture characteristics of the cancer area, reticular structure, etc. corresponding to the cancer area classification;

[0079] In this example, after the digital pathology slice auxiliary diagnosis system completes the predictive diagnosis for the slice, the display module displays the corresponding cancer area classification. After the user moves the mouse to the corresponding cancer area classification result, the area that meets the corresponding classification symptoms is highlighted in the digital pathology slice image. At the same time, the characteristics of the current area diagnosed as the corresponding result are converted into interpretable text above the corresponding area, thereby assisting the user to understand the corresponding diagnostic basis.

[0080] The working principle and beneficial effects of the above technical solution: In order to realize the interpretation of digital pathology sections, the key features of the digital pathology sections are first established according to the doctor's annotation information on the digital pathology sections, and then the key features are processed to obtain the feature information corresponding to each key feature and create a text explanation corresponding to each key feature, which is then marked in the digital pathology sections to obtain the text explanation corresponding to each slice area, thereby establishing the diagnosis result of the digital pathology sections, and finally the display module displays the diagnosis result. The user can judge the lesion information in the digital pathology sections according to the diagnosis result, thereby helping the user understand the condition.

[0081] Example 2

[0082] Based on Example 1, the digital pathology section auxiliary diagnosis and interpretation system further includes:

[0083] The storage module is used to obtain the diagnosis result corresponding to each digital pathology slice and store the digital pathology slice and the corresponding diagnosis result.

[0084] The working principle and beneficial effects of the above technical solution are as follows: each digital pathology slice and its corresponding diagnosis result are stored so that users can view them at any time.

[0085] Example 3

[0086] Based on Example 1, the digital pathology slice auxiliary diagnosis and interpretation system, the acquisition module, includes:

[0087] An information collection unit is used to collect all the annotated information on the digital pathological slice by the doctor, respectively obtain the amount of information contained in each annotated information, classify the annotated information with the same amount of information into one category, and obtain a plurality of information categories;

[0088] An information processing unit, configured to establish a convolution number for a corresponding information class based on the amount of information corresponding to the labeled information in each information class, and to perform convolution of a corresponding number of convolutions on each information class using a preset convolutional neural network to generate information attributes corresponding to each information class;

[0089] A feature generation unit, configured to determine the expression type corresponding to the information class based on the information attribute, and establish texture sub-features, shape sub-features, and edge sub-features of the digital pathological slice;

[0090] The feature establishment unit is used to perform feature enhancement on the texture sub-features, shape sub-features and edge sub-features respectively, and establish key features according to the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information.

[0091] In this example, the amount of information corresponds to the number of convolutions;

[0092] In this instance, the amount of information in all the annotation information contained in the same information class is the same;

[0093] In this example, the information attributes include: texture attributes, shape attributes, and edge attributes;

[0094] In this example, the expression types include: texture expression, shape expression and edge expression;

[0095] In this example, the texture sub-feature corresponds to the enhanced sub-feature, the shape sub-feature corresponds to the enhanced shape sub-feature, and the edge sub-feature corresponds to the enhanced edge sub-feature.

[0096] The working principle and beneficial effects of the above technical solution are as follows: In order to further enhance the sub-features in digital pathology sections, the annotated information with the same amount of information is first classified, and the number of convolutions is determined based on the amount of information in the annotated information. Then, a preset convolutional neural network is used to perform convolutions on the information class a corresponding number of times to generate information attributes corresponding to each information class. Then, the texture sub-features, shape sub-features and edge sub-features of the digital pathology sections are determined based on the information attributes, and then the sub-features are enhanced, thereby establishing the key features of the digital pathology sections.

[0097] Example 4

[0098] Based on Example 1, the digital pathology section auxiliary diagnosis interpretation system, the interpretation module includes:

[0099] an analysis and matching unit, configured to perform parameter analysis on all the annotation information on the digital pathology slice to obtain cell parameters corresponding to the digital pathology slice, determine the cancer attribute corresponding to the digital pathology slice based on the cell parameters, and obtain a corresponding semantic vocabulary according to the cancer attribute;

[0100] a marking processing unit, used to mark each key feature in a corresponding digital pathology slice, obtain a cross relationship between different key features, establish an associated label for the corresponding key feature based on the cross relationship, mark each associated label in the digital pathology slice, and obtain a target digital pathology slice;

[0101] A semantic conversion unit, configured to convert key features contained in the target digital pathology slice into corresponding text information based on the semantic word library in combination with preset semantic conversion rules, and to perform text fusion on the text information according to association tags between different key features to obtain feature information corresponding to the key features;

[0102] The format conversion unit is used to convert the feature information into a preset text format to obtain a text explanation corresponding to each key feature.

[0103] In this example, parameter analysis refers to the process of extracting all parameters from the annotation information;

[0104] In this example, the cancer attributes include: cancerous organ, cancer degree, and cancerous area;

[0105] In this example, each cancer attribute corresponds to a semantic vocabulary;

[0106] In this instance, the cross-relationships represent the overlap between different key features;

[0107] In this example, the associated label means establishing similar labels for key features with cross-relationships;

[0108] In this example, text fusion refers to the process of semantically fusing text semantics of multiple text information.

[0109] The working principle and beneficial effects of the above technical solution are as follows: in order to interpret the key features in text, the digital pathology slices are first subjected to parameter analysis to obtain the cell parameters of the digital pathology slices, and then the cancer attributes of the digital pathology slices are determined, and the corresponding semantic vocabulary is matched according to the cancer attributes. Then, the cross-relationship between different key features is determined according to the positional relationship of the key feature tags in the digital pathology slices, and the associated labels are established according to the cross-relationships. The key features contained in the target digital pathology slices are converted into text information using the semantic vocabulary and preset semantic conversion rules, and the text information is text-fused according to the associated labels between different key features, thereby obtaining the feature information corresponding to the key features. Finally, the feature information is converted into a preset text format according to actual needs, thereby obtaining the text explanation corresponding to each key feature, and the features of the digital pathology slices are converted into understandable text in combination with pathological knowledge, which is convenient for users to understand.

[0110] Example 5

[0111] Based on Example 1, the digital pathology section auxiliary diagnosis and interpretation system, the execution module includes:

[0112] A first execution unit is used to divide the digital pathology slice into a plurality of slice regions, map the text interpretation into the digital pathology slice, and obtain the slice region into which each text interpretation falls;

[0113] The second execution unit is used to regard a plurality of text interpretations falling into the same slice area as a text set, and perform word segmentation processing on the text interpretations contained in each text set based on medical rules to obtain a plurality of words corresponding to each text set;

[0114] The third execution unit is used to reorganize a number of words corresponding to the same text set to obtain an overall text interpretation of each slice area, count the overall text interpretation corresponding to each slice area, and generate a diagnosis result of the digital pathology slice.

[0115] In this example, the process of performing word segmentation processing on the word interpretations contained in each word set based on medical rules includes: preprocessing data according to medical rules, performing word segmentation operations on the word interpretations according to Chinese rules, adding a synonym word list to the word interpretations, repeating the same processing operations on the word interpretations, and removing stop words from the word interpretations;

[0116] In this example, each slice area may correspond to multiple textual explanations.

[0117] The working principle and beneficial effects of the above technical solution: In order to further analyze the diagnostic results of digital pathology sections, the digital pathology sections are divided into several slice areas, and then the text explanation corresponding to each slice area is regarded as a text set, and then each text set is analyzed and processed to obtain several words, and finally the words are reorganized to obtain the overall text explanation corresponding to each slice area, and the diagnostic results of the digital pathology sections are generated for user reference.

[0118] Example 6

[0119] Based on Example 1, the digital pathology section auxiliary diagnosis and interpretation system, the display module, includes:

[0120] An instruction unit, used for receiving a viewing instruction input by a user;

[0121] A receiving unit, configured to receive the diagnosis result generated by the execution module;

[0122] The display unit is used to retrieve and display the corresponding viewing content based on the viewing instruction of the user, and to display the diagnosis result.

[0123] The working principle and beneficial effects of the above technical solution are as follows: in order to facilitate user viewing, the corresponding viewing content is selected and displayed by collecting the user's viewing instructions, and the diagnostic results of the digital pathology sections are displayed in the display unit for user reference.

[0124] Example 7

[0125] On the basis of Example 3, the digital pathology section auxiliary diagnosis and interpretation system, the feature establishment unit includes:

[0126] A first establishing subunit is used to downsample the texture sub-feature to obtain a plurality of sampling points, connect the sampling points to establish a texture contour, and use the texture contour to perform enhancement training on the texture sub-feature to obtain a corresponding enhanced texture sub-feature;

[0127] A second establishing subunit is used to rotate and flip the shape sub-feature to obtain feature parameters of the shape sub-feature at different angles, traverse the digital pathology slice using the feature parameters, extract target feature parameters contained in the digital pathology slice, and correct the shape sub-feature based on the target feature parameters to obtain a corresponding enhanced corrected sub-feature;

[0128] The third establishing subunit is used to mark the edge sub-feature in the digital pathology slice, translate the edge sub-feature in the digital pathology slice, obtain the overlapping relationship between the digital pathology slice and the edge sub-feature, obtain the leak-filling edge contained in the digital pathology slice, and correct the edge sub-feature based on the position of the leak-filling edge on the digital pathology slice to obtain the enhanced edge sub-feature.

[0129] The fourth establishing sub-unit is used to obtain the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information to establish key features.

[0130] In this example, the missed edge represents an edge that exists in the digital pathology slice but is not reflected in the edge sub-feature.

[0131] The working principle and beneficial effects of the above technical solution are as follows: texture sub-features, shape sub-features and edge sub-features are enhanced respectively through different data enhancement technologies to generate multi-angle and multi-level enhanced sub-features, and then key features corresponding to the same annotation information are established based on the enhanced sub-features. The deficiencies of the original sub-features are compensated by filling in the gaps, and accurate key features are obtained.

[0132] Example 8

[0133] This embodiment provides a method for assisting diagnosis and interpretation of digital pathological sections. Figure 2 As shown, including:

[0134] Step 1: Obtain the doctor's annotation information on the digital pathology slice, and extract several key features contained in the digital pathology slice based on the annotation information;

[0135] Step 2: Analyze each of the key features respectively, establish feature information corresponding to each key feature, convert each feature information into a preset text format, and obtain a text explanation corresponding to each key feature;

[0136] Step 3: Mark each text explanation in the slice area corresponding to the digital pathology slice, count the text explanations corresponding to each slice area in the digital pathology slice, and establish the diagnosis result of the digital pathology slice;

[0137] Step 4: Display the digital pathology slice and the diagnosis result corresponding to the digital pathology slice.

[0138] In this example, the number of annotation information included in a digital pathology slide may be one or more;

[0139] In this example, one key feature corresponds to one feature information, wherein the key feature represents the features displayed in different regions of the digital pathology slice, and the feature information represents the information used to describe the key feature;

[0140] In this example, the default text format can be Chinese, and the user can also adjust different text formats as needed;

[0141] In this example, the textual explanations corresponding to different key features are not necessarily the same;

[0142] In this example, one textual interpretation can explain one or more slice regions, and one slice region can correspond to one or more textual interpretations;

[0143] In this example, the diagnosis result includes: shape characteristics, size characteristics, characteristics of infiltrating surrounding tissues, texture characteristics of the cancer area, reticular structure, etc. corresponding to the cancer area classification;

[0144] In this example, after the digital pathology slice auxiliary diagnosis system completes the predictive diagnosis for the slice, the display module displays the corresponding cancer area classification. After the user moves the mouse to the corresponding cancer area classification result, the area that meets the corresponding classification symptoms is highlighted in the digital pathology slice image. At the same time, the characteristics of the current area diagnosed as the corresponding result are converted into interpretable text above the corresponding area, thereby assisting the user to understand the corresponding diagnostic basis.

[0145] The working principle and beneficial effects of the above technical solution: In order to realize the interpretation of digital pathology sections, the key features of the digital pathology sections are first established according to the doctor's annotation information on the digital pathology sections, and then the key features are processed to obtain the feature information corresponding to each key feature and create a text explanation corresponding to each key feature, which is then marked in the digital pathology sections to obtain the text explanation corresponding to each slice area, thereby establishing the diagnosis result of the digital pathology sections, and finally the display module displays the diagnosis result. The user can judge the lesion information in the digital pathology sections according to the diagnosis result, thereby helping the user understand the condition.

[0146] Example 9

[0147] On the basis of Example 8, the digital pathology section auxiliary diagnosis and interpretation method, step 1, comprises:

[0148] Step 11: Collect all the annotation information on the digital pathology slide by the doctor, obtain the amount of information contained in each annotation information, classify the annotation information with the same amount of information into one category, and obtain several information categories;

[0149] Step 12: Based on the amount of information corresponding to the labeled information in each information class, establish the number of convolutions for the corresponding information class, and use the preset convolutional neural network to perform convolutions for each information class for the corresponding number of convolutions to generate information attributes corresponding to each information class;

[0150] Step 13: determining the expression type corresponding to the information class based on the information attribute, and establishing the texture sub-feature, shape sub-feature and edge sub-feature of the digital pathological slice;

[0151] Step 14: Feature enhancement is performed on the texture sub-features, shape sub-features and edge sub-features respectively, and key features are established according to the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information.

[0152] In this example, the amount of information corresponds to the number of convolutions;

[0153] In this instance, the amount of information in all the annotation information contained in the same information class is the same;

[0154] In this example, the information attributes include: texture attributes, shape attributes, and edge attributes;

[0155] In this example, the expression types include: texture expression, shape expression and edge expression;

[0156] In this example, the texture sub-feature corresponds to the enhanced sub-feature, the shape sub-feature corresponds to the enhanced shape sub-feature, and the edge sub-feature corresponds to the enhanced edge sub-feature.

[0157] The working principle and beneficial effects of the above technical solution are as follows: In order to further enhance the sub-features in digital pathology sections, the annotated information with the same amount of information is first classified, and the number of convolutions is determined based on the amount of information in the annotated information. Then, a preset convolutional neural network is used to perform convolutions on the information class a corresponding number of times to generate information attributes corresponding to each information class. Then, the texture sub-features, shape sub-features and edge sub-features of the digital pathology sections are determined based on the information attributes, and then the sub-features are enhanced, thereby establishing the key features of the digital pathology sections.

[0158] Example 10

[0159] On the basis of Example 9, the digital pathology section auxiliary diagnosis and interpretation method, step 14 includes:

[0160] Step 141: downsampling the texture sub-feature to obtain a plurality of sampling points, connecting the sampling points to establish a texture contour, and using the texture contour to perform enhancement training on the texture sub-feature to obtain a corresponding enhanced texture sub-feature;

[0161] Step 142: rotating and flipping the shape sub-feature to obtain feature parameters of the shape sub-feature at different angles, traversing the digital pathology slice using the feature parameters, extracting target feature parameters contained in the digital pathology slice, and correcting the shape sub-feature based on the target feature parameters to obtain a corresponding enhanced corrected sub-feature;

[0162] Step 143: Mark the edge sub-feature in the digital pathology slice, translate the edge sub-feature in the digital pathology slice, obtain the overlapping relationship between the digital pathology slice and the edge sub-feature, obtain the leak-filling edge contained in the digital pathology slice, and correct the edge sub-feature based on the position of the leak-filling edge on the digital pathology slice to obtain the enhanced edge sub-feature.

[0163] Step 144: Acquire the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information to establish key features.

[0164] In this example, the missed edge represents an edge that exists in the digital pathology slice but is not reflected in the edge sub-feature.

[0165] The working principle and beneficial effects of the above technical solution are as follows: texture sub-features, shape sub-features and edge sub-features are enhanced respectively through different data enhancement technologies to generate multi-angle and multi-level enhanced sub-features, and then key features corresponding to the same annotation information are established based on the enhanced sub-features. The deficiencies of the original sub-features are compensated by filling in the gaps, and accurate key features are obtained.

[0166] 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 equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A digital pathology section auxiliary diagnosis interpretation system, characterized in that: include: An acquisition module, used to obtain the doctor's annotation information on the digital pathology slice, and extract several key features contained in the digital pathology slice based on the annotation information; An interpretation module, used to analyze each of the key features, establish feature information corresponding to each key feature, convert each feature information into a preset text format, and obtain a text interpretation corresponding to each key feature; An execution module, used for marking each text explanation in a slice area corresponding to the digital pathology slice, counting the text explanation corresponding to each slice area in the digital pathology slice, and establishing a diagnosis result of the digital pathology slice; The display module is used to display the digital pathological slice and the diagnosis result corresponding to the digital pathological slice.

2. A digital pathology section auxiliary diagnosis and interpretation system as claimed in claim 1, characterized in that: Also includes: The storage module is used to obtain the diagnosis result corresponding to each digital pathology slice and store the digital pathology slice and the corresponding diagnosis result.

3. A digital pathology section auxiliary diagnosis and interpretation system as claimed in claim 1, characterized in that: The acquisition module comprises: An information collection unit is used to collect all the annotated information on the digital pathological slice by the doctor, respectively obtain the amount of information contained in each annotated information, classify the annotated information with the same amount of information into one category, and obtain a plurality of information categories; An information processing unit, configured to establish a convolution number for a corresponding information class based on the amount of information corresponding to the labeled information in each information class, and to perform convolution of a corresponding number of convolutions on each information class using a preset convolutional neural network to generate information attributes corresponding to each information class; A feature generation unit, configured to determine the expression type corresponding to the information class based on the information attribute, and establish texture sub-features, shape sub-features, and edge sub-features of the digital pathological slice; The feature establishment unit is used to perform feature enhancement on the texture sub-features, shape sub-features and edge sub-features respectively, and establish key features according to the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information.

4. A digital pathology section auxiliary diagnosis and interpretation system as claimed in claim 1, characterized in that: The interpretation module comprises: an analysis and matching unit, configured to perform parameter analysis on all the annotation information on the digital pathology slice to obtain cell parameters corresponding to the digital pathology slice, determine the cancer attribute corresponding to the digital pathology slice based on the cell parameters, and obtain a corresponding semantic vocabulary according to the cancer attribute; a marking processing unit, used to mark each key feature in a corresponding digital pathology slice, obtain a cross relationship between different key features, establish an associated label for the corresponding key feature based on the cross relationship, mark each associated label in the digital pathology slice, and obtain a target digital pathology slice; A semantic conversion unit, configured to convert key features contained in the target digital pathology slice into corresponding text information based on the semantic word library in combination with preset semantic conversion rules, and to perform text fusion on the text information according to association tags between different key features to obtain feature information corresponding to the key features; The format conversion unit is used to convert the feature information into a preset text format to obtain a text explanation corresponding to each key feature.

5. A digital pathology section auxiliary diagnosis and interpretation system as claimed in claim 1, characterized in that: The execution module comprises: A first execution unit is used to divide the digital pathology slice into a plurality of slice regions, map the text interpretation into the digital pathology slice, and obtain the slice region into which each text interpretation falls; The second execution unit is used to regard a plurality of text interpretations falling into the same slice area as a text set, and perform word segmentation processing on the text interpretations contained in each text set based on medical rules to obtain a plurality of words corresponding to each text set; The third execution unit is used to reorganize a number of words corresponding to the same text set to obtain an overall text interpretation of each slice area, count the overall text interpretation corresponding to each slice area, and generate a diagnosis result of the digital pathology slice.

6. A digital pathology section auxiliary diagnosis and interpretation system as claimed in claim 1, characterized in that: The display module comprises: An instruction unit, used for receiving a viewing instruction input by a user; A receiving unit, configured to receive the diagnosis result generated by the execution module; The display unit is used to retrieve and display the corresponding viewing content based on the viewing instruction of the user, and to display the diagnosis result.

7. A digital pathology section auxiliary diagnosis and interpretation system as claimed in claim 3, characterized in that: The feature establishment unit comprises: A first establishing subunit is used to downsample the texture sub-feature to obtain a plurality of sampling points, connect the sampling points to establish a texture contour, and use the texture contour to perform enhancement training on the texture sub-feature to obtain a corresponding enhanced texture sub-feature; A second establishing subunit is used to rotate and flip the shape sub-feature to obtain feature parameters of the shape sub-feature at different angles, traverse the digital pathology slice using the feature parameters, extract target feature parameters contained in the digital pathology slice, and correct the shape sub-feature based on the target feature parameters to obtain a corresponding enhanced corrected sub-feature; A third establishing subunit is configured to mark the edge sub-feature in the digital pathology slice, translate the edge sub-feature in the digital pathology slice, obtain an overlap relationship between the digital pathology slice and the edge sub-feature, obtain a leak-padding edge included in the digital pathology slice, and correct the edge sub-feature based on a position of the leak-padding edge on the digital pathology slice to obtain an enhanced edge sub-feature; The fourth establishing sub-unit is used to obtain the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information to establish key features.

8. A method for auxiliary diagnosis and interpretation of digital pathological sections, characterized in that: include: Step 1: Obtain the doctor's annotation information on the digital pathology slice, and extract several key features contained in the digital pathology slice based on the annotation information; Step 2: Analyze each of the key features respectively, establish feature information corresponding to each key feature, convert each feature information into a preset text format, and obtain a text explanation corresponding to each key feature; Step 3: Mark each text explanation in the slice area corresponding to the digital pathology slice, count the text explanations corresponding to each slice area in the digital pathology slice, and establish the diagnosis result of the digital pathology slice; Step 4: Display the digital pathology slice and the diagnosis result corresponding to the digital pathology slice.

9. A digital pathology section auxiliary diagnosis interpretation method as claimed in claim 8, characterized in that: The step 1 comprises: Step 11: Collect all the annotation information on the digital pathology slide by the doctor, obtain the amount of information contained in each annotation information, classify the annotation information with the same amount of information into one category, and obtain several information categories; Step 12: Based on the amount of information corresponding to the labeled information in each information class, establish the number of convolutions for the corresponding information class, and use the preset convolutional neural network to perform convolutions for each information class for the corresponding number of convolutions to generate information attributes corresponding to each information class; Step 13: determining the expression type corresponding to the information class based on the information attribute, and establishing the texture sub-feature, shape sub-feature and edge sub-feature of the digital pathological slice; Step 14: Feature enhancement is performed on the texture sub-features, shape sub-features and edge sub-features respectively, and key features are established according to the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information.

10. A digital pathology section auxiliary diagnosis interpretation method as claimed in claim 9, characterized in that: The step 14 comprises: Step 141: downsampling the texture sub-feature to obtain a plurality of sampling points, connecting the sampling points to establish a texture contour, and using the texture contour to perform enhancement training on the texture sub-feature to obtain a corresponding enhanced texture sub-feature; Step 142: rotating and flipping the shape sub-feature to obtain feature parameters of the shape sub-feature at different angles, traversing the digital pathology slice using the feature parameters, extracting target feature parameters contained in the digital pathology slice, and correcting the shape sub-feature based on the target feature parameters to obtain a corresponding enhanced corrected sub-feature; Step 143: marking the edge sub-feature in the digital pathology slice, translating the edge sub-feature in the digital pathology slice to obtain an overlap relationship between the digital pathology slice and the edge sub-feature, obtaining a leak-padding edge included in the digital pathology slice, and correcting the edge sub-feature based on a position of the leak-padding edge on the digital pathology slice to obtain an enhanced edge sub-feature; Step 144: Acquire the enhanced texture sub-features, enhanced shape sub-features and enhanced edge sub-features corresponding to the same annotation information to establish key features.

Images