A t-s fuzzy semantic intelligent liver cancer staging method and system

Abstract

The application relates to a T-S fuzzy semantic intelligent liver cancer staging method and system. The method comprises the following steps: S1. liver cancer feature extraction, key liver cancer features are extracted from obtained digital radiology diagnosis reports of patients to form a data set; S2. liver cancer feature fuzzification, a plurality of fuzzy semantic sets are used for fuzzy representation, and relevant membership functions are designed; S3. T-S fuzzy semantic model construction, comprising the following steps: designing a T-S fuzzy semantic rule, solving the membership degree of each rule, identifying the antecedent parameter membership function, identifying the consequent parameter, and outputting the T-S fuzzy semantic model; S4. liver cancer intelligent staging and performance evaluation verification, the liver cancer categories in the Chinese CNLC standard are subjected to digital label processing, a training set and a test set are divided, and the staging performance of the intelligent staging model is verified. The T-S fuzzy semantic intelligent liver cancer staging method can divide the liver cancer into Ia, Ib, IIa and IIb-IV, and is more accurate in grading and staging the liver cancer. The application greatly reduces the estimation error and classification error of the liver cancer state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical data processing, and more particularly to a T-S fuzzy semantic intelligent liver cancer staging method and system. BACKGROUND

[0002] Liver cancer is a malignant tumor with high morbidity and mortality, which seriously threatens people's life and health. About 383,000 people die of liver cancer in China every year, accounting for 51% of global liver cancer death cases, which has brought a heavy burden to China's society and medical treatment. Due to the certain incubation period of liver cancer, the early symptoms are not obvious, which brings great challenges to clinical treatment. In order to improve the treatment effect, the corresponding treatment plan is generally selected according to the malignancy degree (i.e. staging) of liver cancer in clinic. Because the treatment plan, recovery probability and medical expenses of different stages are very different, the accurate staging of liver cancer is regarded as a key link in the diagnosis and treatment of liver cancer in clinic. Liver cancer staging is to describe the severity of malignant tumors according to the degree of primary tumor spread, lymph node status, liver function status, general condition of the body, etc. Accurate liver cancer staging is not only a reliable indicator for accurately predicting the biological behavior and prognosis of malignant tumors, but also can provide accurate patient stratification management and treatment basis for clinicians. Therefore, how to accurately stage liver cancer, especially early accurate staging, is an urgent need in clinic and is highly valued.

[0003] The current liver cancer staging standard in China (such as China Liver Cancer Staging, CNLC) is mainly based on a series of logical and non-logical rules to realize "binary hard decision" on digital radiology diagnosis reports and physician evaluation data. By carefully examining the staging scheme and digital radiology diagnosis report, it can be found that: Figure 1

[0004] The liver cancer staging standard has the following characteristics:

[0005] 1) The staging uses "binary logic hard decision" indicators with very clear single logic rules;

[0006] 2) There are a large number of natural text description semantics, such as "yes", "no", "less than or equal to 5", "greater than 5";

[0007] 3) Tumor size and number play a decisive role in the determination of stages I and II in the staging scheme;

[0008] 4) Different stages have different treatment plans.

[0009] And the digital radiology diagnosis report has the following problems:

[0010] ​1) Tumor size measurement relies on manual labeling by radiologists, and there is a small measurement error;

[0011] 2) The shape of the tumor is irregular, and the boundary is not clear, so it is difficult to accurately enclose the lesion;

[0012] 3) The diagnosis report is the result of the radiologist's "picture speaking" according to the imaging results, and there is a subjective experience difference.

[0013] These data exist in a large number of clinical doctors' subjective judgments and natural language description of the test report. Different experienced doctors have different understanding and cognition of the same natural text, that is, there is semantic ambiguity uncertainty. If the CNLC staging standard is used as the basis for binary logic judgment or a liver cancer patient staging network model is established by a deep learning algorithm to determine the stage of primary liver cancer patients, there is still a lot of room for improvement in the accuracy of the network model. SUMMARY

[0014] To solve the defects of the prior art, the present application provides a T-S fuzzy semantic intelligent liver cancer staging method and system. The T-S fuzzy semantic intelligent liver cancer staging method can greatly improve the correctness of the primary liver cancer staging.

[0015] To achieve the above-mentioned purpose, the technical scheme of the present application is as follows:

[0016] The present application provides a T-S fuzzy semantic intelligent liver cancer staging method, comprising the following steps:

[0017] S1. Liver cancer feature extraction, key liver cancer feature extraction is performed on the obtained patient digital radiology diagnosis report to form a data set;

[0018] S2. Fuzzy liver cancer features, fuzzy representation is performed using multiple fuzzy semantic sets, and a related membership function is designed;

[0019] S3. T-S fuzzy semantic model construction, including designing T-S fuzzy semantic rules, solving the membership degree of each rule, identifying the antecedent parameter membership function, identifying the antecedent parameter membership function, and T-S fuzzy semantic model output;

[0020] S4. Intelligent liver cancer staging and performance evaluation verification, the liver cancer categories in the Chinese CNLC standard are processed by digital labeling, the training set and the test set are divided, and the performance of the intelligent staging model is verified.

[0021] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the step S1 is specifically:

[0022] S11. The patient's digital radiology diagnosis report is processed by word segmentation, and character-level words, sentence-level words, position information and other related text information are extracted;

[0023] S12. Form a 128*7 matrix and map to space to get high-dimensional word vectors, so that words with the same semantics will be close to each other;

[0024] S13. Extract key feature information of liver cancer such as tumor size and tumor number;

[0025] S14. Form an Excel-formatted data set.

[0026] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the specific steps of step S2 are:

[0027] S21. Fuzzification of key features of liver cancer such as tumor size and tumor number, wherein the tumor size is divided into three fuzzy semantic sets:

[0028] Small, i.e. Small, represented by S: 0-30mm;

[0029] Middle, i.e. Middle, represented by M: 30-50mm;

[0030] Large, i.e. Large, represented by L: >50mm;

[0031] The number of tumors is divided into three fuzzy semantic sets:

[0032] Small, i.e. Small, represented by S: 1;

[0033] Middle, i.e. Middle, represented by M: 2-3;

[0034] Large, i.e. Large, represented by L: >3;

[0035] S22. Select a Gaussian function as the membership function of the fuzzy semantic set, and determine the initial mean and variance of the Gaussian function. Assuming that the membership function corresponding to the liver cancer feature information fuzzy semantic set is a Gaussian function:

[0036]

[0037] The mean and standard deviation of the Gaussian function, respectively.

[0038] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the T-S fuzzy semantic rules designed in step S3 are as follows:

[0039] Rule : IF is and is …and is , then

[0040] The T-S fuzzy semantic model is composed of IF-THEN, wherein IF represents the antecedent part of the rule, and THEN represents the consequent part of the rule, represents the liver cancer characteristic information, also referred to as the antecedent parameter of the T-S fuzzy semantic model, is the fuzzy semantic set corresponding to the liver cancer characteristic information after being fuzzified, represents the consequent parameter corresponding to each rule, represents the output corresponding to the ith rule.

[0041] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the membership degree of each rule is solved in step S3: the fuzzy C regression clustering method fusing the liver cancer characteristic information is used to solve the membership degree of each model, and the objective function constructed according to the fuzzy C regression clustering method fusing the liver cancer characteristic information is as follows:

[0042]

[0043] wherein represents the liver cancer characteristic information, represents the center value of the characteristic data, and the membership degree is the degree of the characteristic belonging to the model, is the output value of each rule, is the liver cancer staging result of the doctor's experience judgment, is the Lagrange multiplier, is the fuzzy index, which is an integer, represents the number of rules. According to the gradient descent method, the membership degree is obtained:

[0044]

[0045] wherein .

[0046] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the consequent parameter identification in step S3 is: the ridge regression method is used to realize the consequent parameter identification, wherein the fuzzy C clustering method is used to initialize the consequent parameter, and the objective function of the improved ridge regression model is as follows:

[0047]

[0048] wherein , , , is the weight of the rule, is a regularization parameter, according to optimization theory, the final optimization result is:

[0049] .

[0050] wherein is a unit matrix of dimension is a consequent parameter of the T-S fuzzy semantic model, is normalized liver cancer feature information, is the output of the th rule.

[0051] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the antecedent parameter membership function recognition of step S3 is: to realize the parameter recognition of the antecedent parameter membership function, that is, the recognition of the mean and variance of the Gaussian function corresponding to the fuzzy semantic set of liver cancer feature information:

[0052]

[0053]

[0054]

[0055] is the membership function corresponding to the fuzzy semantic set of liver cancer features, which is designed as a Gaussian function here, respectively, the mean and standard deviation of the Gaussian function.

[0056] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the output of the T-S fuzzy semantic model of step S3 is:

[0057]

[0058] is the weight of the rule, calculated as follows:

[0059] ;

[0060] wherein is the membership function corresponding to the th liver cancer feature information fuzzy semantic set of the th rule, is a continuous multiplication symbol, is a summation symbol.

[0061] Further, in the above T-S fuzzy semantic intelligent liver cancer staging method, the specific steps of step S4 are:

[0062] S41. The data after the accurate intelligent staging processing of liver cancer according to the Chinese CNLC staging standard includes four categories of Ia, Ib, IIa, IIb-IV, and the category labels are represented by 0, 1, 2 and 3 respectively;

[0063] S42. The data of patients are randomly selected from the four categories of data in proportion to form a training data set, and the remaining data is used as a test data set, and the general proportion value can be 2:8, 3:7 or 4:6;

[0064] S43. The training data set and the test data set are input into the constructed T-S fuzzy semantic intelligent staging model for staging processing, and finally the correct rate of staging is counted.

[0065] Another aspect of the present application provides a T-S fuzzy semantic intelligent liver cancer staging system based on the T-S fuzzy semantic intelligent liver cancer staging system, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to realize the following steps:

[0066] S1. Liver cancer feature extraction, key liver cancer feature extraction is performed on the obtained digital radiology diagnosis report of patients to form a data set;

[0067] S2. Fuzzy liver cancer features, fuzzy representation is performed by using a plurality of fuzzy semantic sets, and a related membership function is designed;

[0068] S3. T-S fuzzy semantic model construction, including designing T-S fuzzy semantic rules, solving the membership degree of each rule, identifying the antecedent parameter membership function, identifying the antecedent parameter membership function, and outputting the T-S fuzzy semantic model;

[0069] S4. Intelligent liver cancer staging and performance evaluation verification, the liver cancer categories in the Chinese CNLC standard are processed by digital labeling, the training set and the test set are divided, and the performance of the intelligent staging model is verified.

[0070] Compared with the prior art, the T-S fuzzy semantic intelligent liver cancer staging method has the following advantages and beneficial effects:

[0071] 1. At present, the liver cancer grading result is not very accurate, and it is divided into early, middle and late three stages by deep learning, while the T-S fuzzy semantic intelligent liver cancer staging method can divide Ia, Ib, IIa, IIb-IV, and the liver cancer grading and staging is more fine.

[0072] 2. The fuzzy semantics proposed by the T-S fuzzy semantic intelligent liver cancer staging method can solve the error problem of the prior art, and can divide the liver cancer more correctly and accurately according to the feature data.

[0073] 3、The T-S fuzzy semantics is combined with liver cancer staging, which greatly reduces the estimation error and classification error of liver cancer state. BRIEF DESCRIPTION OF DRAWINGS

[0074] Figure 1 For the Chinese liver cancer staging scheme.

[0075] Figure 2 For the T-S fuzzy semantics intelligent liver cancer staging method flow chart based on the application.

[0076] Figure 3 For the T-S fuzzy semantics intelligent liver cancer staging method flow chart based on the application.

[0077] Figure 4 For the T-S fuzzy semantics intelligent liver cancer staging method flow chart based on the application. DETAILED DESCRIPTION

[0078] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application will be further described in detail below in combination with the drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0079] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, however, the present application can also be implemented in other ways different from those described herein, therefore, the scope of protection of the present application is not limited by the specific embodiments disclosed below.

[0080] In this specification, the illustrative description of certain terms does not necessarily refer to the same embodiment or example. Moreover, the described specific features, steps, methods or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0081] The technical solutions of the present application will be further described below in combination with Figures 2 to 4 and embodiments.

[0082] Embodiment 1

[0083] As Figure 2 , the present embodiment is a T-S fuzzy semantics intelligent liver cancer staging method, which specifically includes the following steps:

[0084] S1. Liver cancer feature extraction, the obtained patient digital radio diagnosis report is extracted for key liver cancer features to form a data set;

[0085] S2. Liver cancer feature fuzzification, fuzzy representation is performed by using multiple fuzzy semantic sets, and a related membership function is designed;

[0086] S3. T-S fuzzy semantic model construction, including designing T-S fuzzy semantic rules, solving the membership degree of each rule, identifying the antecedent parameter membership function, identifying the consequent parameter, and outputting the T-S fuzzy semantic model;

[0087] S4. Liver cancer intelligent staging and performance evaluation verification, the liver cancer categories in the Chinese CNLC standard are digitally labeled, the training set and the test set are divided, and the performance of the intelligent staging model is verified.

[0088] A fuzzy system is an artificial intelligence system based on fuzzy set and fuzzy reasoning theory. Its main feature is that it can convert the fuzzy language in nature into fuzzy rules similar to human reasoning mechanism. At present, fuzzy systems have been widely used in various fields, such as image processing, intelligent decision-making, etc. Compared with other artificial intelligence technologies, fuzzy systems not only have strong learning ability, but also have high interpretability. Fuzzy systems have been widely used in medical diagnosis due to their unique advantages. When an intelligent model is applied in medical diagnosis, people hope that it has strong interpretability and good reliability. Fuzzy systems establish corresponding fuzzy rules for a certain medical disease through fuzzy reasoning mechanism, simulate the process of expert diagnosis of the disease, and help doctors solve complex medical diagnosis problems. Among several existing classical fuzzy systems, Takagi-Sugeno (T-S) fuzzy system is widely discussed and used due to its simple output and good approximation performance. For fuzzy logic systems, the interpretability mainly comes from the fuzzy set in the fuzzy rule, which can correspond to human language description. The present application uses a T-S fuzzy semantic model to realize the staging of liver cancer.

[0089] Embodiment 2

[0090] In combination with Figure 2 , Figure 3 and Figure 4 , the present embodiment provides a T-S fuzzy semantic intelligent liver cancer staging method based on the more detailed steps provided in Embodiment 1:

[0091] S1. Liver cancer feature extraction, extracting key liver cancer features from obtained digital radiology diagnosis report of patients to form an Excel data set;

[0092] S2. Fuzzy liver cancer features, fuzzy representation with multiple fuzzy semantic sets, and design of related membership functions, the specific steps of step S2 are:

[0093] S21. Fuzzification of tumor size and tumor number features, design of different fuzzy semantic sets;

[0094] S22. Selecting a Gaussian function as the membership function of the fuzzy semantic set, determining the initial mean and variance of the Gaussian function. Assuming that the membership function corresponding to the liver cancer characteristic information fuzzy semantic set is a Gaussian function:

[0095]

[0096] respectively the mean and standard deviation of the Gaussian function.

[0097] S3. T-S fuzzy semantic model construction, including designing T-S fuzzy semantic rules, solving the membership degree of each rule, identifying the antecedent parameter, identifying the membership function of the antecedent parameter, T-S fuzzy semantic model output;

[0098] S31: Designing T-S fuzzy rules or models, designing the number of rules according to the liver cancer characteristic information fuzzy semantic set in step S2, and designing T-S fuzzy semantic rules in step S3 as follows:

[0099] Rule : IF is and is …and is , then

[0100] The T-S fuzzy semantic model is composed of IF-THEN, wherein IF represents the antecedent part of the rule, THEN represents the consequent part of the rule, represents the liver cancer characteristic information, also known as the antecedent parameter of the T-S fuzzy semantic model, is the corresponding fuzzy semantic set after the liver cancer characteristic information is fuzzified, represents the consequent parameter corresponding to each rule, represents the output corresponding to the ith rule.

[0101] S32: Solving the membership degree of each model, using the fuzzy C regression clustering method of liver cancer characteristic information to solve the membership degree, and step S3 is described as follows: solving the membership degree of each rule. The membership degree of each model is solved by using the fuzzy C regression clustering method of liver cancer characteristic information, and the objective function constructed according to the fuzzy C regression clustering method of liver cancer characteristic information is as follows:

[0102]

[0103] wherein represents the liver cancer characteristic information, represents the center value of the characteristic data, and the membership degree is the degree of characteristic membership to the model, The output value of each rule, The liver cancer staging result determined by the doctor's experience, The Lagrange multiplier, The fuzzy index, which is an integer, The number of rules. According to the gradient descent method, the membership degree :

[0104]

[0105] wherein .

[0106] S33: Consequent parameter identification, the improved ridge regression method is used to adaptively update the consequent parameters of the T-S fuzzy semantic model. The consequent parameter identification described in step S3 is: the ridge regression method is used to realize the consequent parameter identification, wherein the fuzzy C clustering method is used to initialize the consequent parameters, and the objective function of the improved ridge regression model is as follows:

[0107]

[0108] wherein , , , is the weight of the rule, is a regularization parameter, according to the optimization theory, the final optimization result is:

[0109] .

[0110] wherein is a unit matrix of dimension, is the consequent parameter of the T-S fuzzy semantic model, is the normalized liver cancer feature information, is the output of the rule.

[0111] S34: Antecedent parameter membership function identification, the fuzzy C regression clustering method is used to solve the membership degree to adaptively update the mean and variance of the Gaussian function corresponding to the antecedent parameters of the T-S fuzzy semantic model. The antecedent parameter membership function identification described in step S3 is: the parameter identification of the antecedent parameter membership function is realized, that is, the identification of the mean and variance of the Gaussian function corresponding to the fuzzy semantic set of the liver cancer feature information:

[0112]

[0113]

[0114]

[0115] Let be the membership function corresponding to the fuzzy semantic set of liver cancer features; here, it is designed as a Gaussian function. These are the mean and standard deviation of the Gaussian function, respectively.

[0116] S35: The output of the TS fuzzy semantic model is obtained by multiplying the Gaussian functions corresponding to the antecedent parameters obtained in S34 and normalizing them to obtain the weight of each rule. Then, the output of each rule and its corresponding weight are weighted and summed to obtain the output of the TS fuzzy semantic model. The output of the TS fuzzy semantic model in step S3 is as follows:

[0117]

[0118] The weight of the rule is calculated as follows:

[0119] ;

[0120] in For the first Rule No. Membership functions corresponding to a fuzzy semantic set of liver cancer feature information. The multiplication symbol is used. This is the summation symbol.

[0121] S4. Intelligent staging and performance evaluation of liver cancer: The liver cancer categories in the Chinese CNLC standard are digitally labeled, divided into training and test sets, and the staging performance of the intelligent staging model is verified.

[0122] Example 3

[0123] This embodiment provides a more detailed step-by-step approach to intelligent liver cancer staging based on TS fuzzy semantics, building upon Embodiments 1 and 2.

[0124] S1. Liver Cancer Feature Extraction: Key liver cancer features were extracted from the obtained digital radiological diagnostic reports of patients to form a dataset. The data used in the experiment came from the Department of Interventional Oncology, First Affiliated Hospital of Sun Yat-sen University. With the support of the Department of Interventional Oncology, 335 digital radiological diagnostic reports related to liver cancer were obtained. Figure 3 The technique is used to extract liver cancer feature information. Step S1 specifically involves:

[0125] S11. Perform word segmentation on the patient's digital radiological diagnosis report and extract relevant text information such as character-level words, sentence-level words, and location information;

[0126] S12. Form a 128*7 matrix and map it into space to obtain high-dimensional word vectors, so that words with the same meaning will be close to each other;

[0127] S13. Extract the key feature information of liver cancer, such as tumor size, tumor number, etc.

[0128] S14. Form a data set in Excel format.

[0129] S2. Fuzzy liver cancer features, fuzzy representation with multiple fuzzy semantic set pairs, and design related membership functions. The specific steps of step S2 are:

[0130] S21. Fuzzy liver cancer key features such as tumor size and tumor number, where tumor size is divided into 3 fuzzy semantic sets:

[0131] Small, i.e. Small, represented by S: 0-30mm;

[0132] Middle, i.e. Middle, represented by M: 30-50mm;

[0133] Large, i.e. Large, represented by L: > 50mm;

[0134] Tumor number is divided into 3 fuzzy semantic sets:

[0135] Small, i.e. Small, represented by S: 1;

[0136] Middle, i.e. Middle, represented by M: 2-3;

[0137] Large, i.e. Large, represented by L: > 3;

[0138] S22. Select Gaussian function as fuzzy semantic set membership function, and determine the initial mean and variance of Gaussian function. Assuming that the membership function corresponding to the fuzzy semantic set of liver cancer feature information is Gaussian function:

[0139]

[0140] The mean and standard deviation of the Gaussian function are respectively:

[0141] S3. T-S fuzzy semantic model construction, including design of T-S fuzzy semantic rules, solution of membership degree of each rule, identification of consequent parameters, identification of antecedent parameter membership function, T-S fuzzy semantic model output;

[0142] S31: Design T-S fuzzy rules or model, design the number of rules according to the fuzzy semantic set divided by liver cancer feature information. The design of T-S fuzzy semantic rules is as follows:

[0143]

[0144] The T-S fuzzy semantic model is composed of IF-THEN, wherein IF represents the antecedent part of the rule, THEN represents the consequent part of the rule, R1-R9 represent the rule number, respectively represent the tumor size and the tumor number, also referred to as the antecedent parameters of the T-S fuzzy semantic model, S, M and L are the fuzzy semantic sets corresponding to the liver cancer feature information after being fuzzified, represent the consequent parameters corresponding to each rule, represent the output corresponding to each rule.

[0145] S32: Solve the membership degree of each model, and the membership degree is solved by using the fuzzy C regression clustering method fusing the liver cancer feature information. The membership degree of each rule is solved by using the fuzzy C regression clustering method fusing the liver cancer feature information, and the objective function constructed according to the fuzzy C regression clustering method fusing the liver cancer feature information is as follows:

[0146]

[0147] wherein represents the liver cancer feature information, represents the center value of the feature data, and the membership degree is the degree of the feature belonging to the model, is the output value of each rule, is the liver cancer staging result determined by the doctor's experience, is the Lagrange multiplier, is the fuzzy index, which is an integer, represents the rule number. According to the gradient descent method, the membership degree is obtained.

[0148]

[0149] wherein .

[0150] S33: Consequent parameter identification, the consequent parameters of the T-S fuzzy semantic model are adaptively updated by using the improved ridge regression method. The consequent parameter identification described in step S3 is as follows: the ridge regression method is used to realize the consequent parameter identification, wherein the fuzzy C clustering method is used to initialize the consequent parameters, and the objective function of the improved ridge regression model is as follows:

[0151]

[0152] wherein , , , is the weight of the rule, is a regularization parameter, and according to the optimization theory, the final optimization result is:

[0153] .

[0154] in yes An identity matrix of dimension 1 These are the consequent parameters of the TS fuzzy semantic model. This is the normalized liver cancer characteristic information. For the first Output of the rules.

[0155] S34: Antecedent parameter membership function identification. The membership degrees obtained by fuzzy C-regression clustering are used to adaptively update the mean and variance of the Gaussian function corresponding to the antecedent parameters of the TS fuzzy semantic model. The antecedent parameter membership function identification mentioned in step S3 is: to realize the parameter identification of the antecedent parameter membership function, that is, to identify the mean and variance of the Gaussian function corresponding to the fuzzy semantic set of liver cancer feature information.

[0156]

[0157]

[0158]

[0159] Let be the membership function corresponding to the fuzzy semantic set of liver cancer features; here, it is designed as a Gaussian function. These are the mean and standard deviation of the Gaussian function, respectively.

[0160] S35: The output of the TS fuzzy semantic model is obtained by multiplying the Gaussian functions corresponding to the antecedent parameters obtained in S34 and normalizing them to obtain the weight of each rule. Then, the output of each rule and its corresponding weight are weighted and summed to obtain the output of the TS fuzzy semantic model. The output of the TS fuzzy semantic model in step S3 is as follows:

[0161]

[0162] The weight of the rule is calculated as follows:

[0163] ;

[0164] in For the first Rule No. Membership functions corresponding to a fuzzy semantic set of liver cancer feature information. The multiplication symbol is used. This is the summation symbol.

[0165] S4. Intelligent staging of liver cancer and performance evaluation verification, the liver cancer categories in the Chinese CNLC standard are digitally labeled, the training set and the test set are divided, and the performance of the intelligent staging model is verified, and the specific steps of the step S4 are:

[0166] S41. According to the Chinese CNLC staging standard, the liver cancer staging results obtained by doctors' experience in the data set are digitally labeled, wherein the patients in stage Ia are represented by the number 0, the patients in stage Ib are represented by the number 1, the patients in stage IIa are represented by the number 2, and the patients in stage IIb-IV are represented by the number 3;

[0167] S42. The data of patients are randomly selected from the four types of data in the proportions of 2:8, 3:7 and 4:6 to form the training data set and the test data set;

[0168] S43. The training data set and the test data set are input into the T-S fuzzy semantic intelligent staging model constructed in step S3 for staging processing, and finally the error of state estimation and the accuracy of staging are counted.

[0169] The T-S fuzzy semantic intelligent liver cancer staging method of the present application can divide Ia, Ib, IIa, IIb-IV, and is more accurate in grading and staging liver cancer. The present application greatly reduces the estimation error of liver cancer state and the error of classification.

[0170] In addition, without contradiction, a person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples.

[0171] It should be noted that the above embodiments of the present application are only examples for clearly illustrating the present application, and are not intended to limit the embodiments of the present application. Based on the above description, those skilled in the art can also make other different forms of changes or modifications. Here, it is not necessary and impossible to exhaust all the embodiments. Any modification, equivalent replacement and improvement made within the spirit and principles of the present application shall be included in the protection scope of the claims of the present application.

Claims

1. A liver cancer staging method based on TS fuzzy semantics, characterized in that, Includes the following steps: S1. Liver cancer feature extraction: Key liver cancer features are extracted from the obtained digital radiological diagnostic reports of patients to form a dataset; S2. Fuzzy representation of liver cancer features: multiple fuzzy semantic sets are used for fuzzy representation, and related membership functions are designed. S3.TS fuzzy semantic model construction includes designing TS fuzzy semantic rules, solving the membership degree of each rule, identifying consequent parameters, identifying the membership function of antecedent parameters, and outputting the TS fuzzy semantic model. S4. Intelligent staging and performance evaluation of liver cancer: The liver cancer categories in the Chinese CNLC standard are digitally labeled, divided into training and test sets, and the staging performance of the intelligent staging model is verified. Step S1 specifically involves: S11. Perform word segmentation on the patient's digital radiological diagnosis report and extract relevant text information such as character-level words, sentence-level words, and location information; S12. Form a 128*7 matrix and map it into space to obtain high-dimensional word vectors, so that words with the same meaning will be close to each other; S13. Extract key characteristic information of liver cancer such as tumor size and number; S14. Create a dataset in Excel format; The specific steps of step S2 are as follows: S21. The key features of liver cancer, including tumor size and number, are fuzzified, with tumor size divided into three fuzzy semantic sets: Small, represented by S: 0-30mm; Middle, represented by M: 30-50mm; Large, represented by L: >50mm; The number of tumors is divided into three fuzzy semantic sets: Small, represented by S: 1; The middle, represented by M, consists of 2-3 elements. "Large" is represented by the letter L, which means "more than 3". S22. Select a Gaussian function as the membership function of the fuzzy semantic set, and determine the initial mean and variance of the Gaussian function; The design of the TS fuzzy semantic rules in step S3 is as follows: Rule : IF is and is …and is , then The TS fuzzy semantic model consists of IF-THEN, where IF represents the antecedent of the rule and THEN represents the consequent of the rule. This represents the characteristic information of liver cancer, also known as the antecedent parameters of the TS fuzzy semantic model. This is the fuzzy semantic set corresponding to the fuzzified feature information of liver cancer. This represents the consequent parameter corresponding to each rule. This represents the output corresponding to the i-th rule; Step S3 involves solving the membership degree of each rule by using a fuzzy C-regression clustering method that integrates liver cancer feature information. The objective function constructed based on the fuzzy C-regression clustering method that integrates liver cancer feature information is as follows: in Indicates characteristic information of liver cancer. The central value and membership degree of the feature data. The degree to which a feature belongs to the model. For the output value of each rule, The staging results of liver cancer based on the doctor's experience. For Lagrange multipliers, The fuzzy index takes integer values. Represents the number of rules; The membership degree is obtained using the gradient descent method. : in ; The consequent parameter identification in step S3 involves using an improved ridge regression method, where fuzzy C-clustering is used to initialize the consequent parameters. The objective function of the improved ridge regression model is as follows: in , , , For the weight of the rule, This is the regularization parameter. According to optimization theory, the final optimization result is: in yes An identity matrix of dimension 1 These are the consequent parameters of the TS fuzzy semantic model. This is the normalized liver cancer characteristic information. For the first Output of the rule; Step S3, the identification of the antecedent parameter membership function, is to identify the parameters of the antecedent parameter membership function, that is, to identify the mean and variance of the Gaussian function corresponding to the fuzzy semantic set of liver cancer feature information. Let be the membership function corresponding to the fuzzy semantic set of liver cancer features; here, it is designed as a Gaussian function. These are the mean and standard deviation of the Gaussian function, respectively. The output of the TS fuzzy semantic model in step S3 is: The weight of the rule is calculated as follows: ； in For the first Rule No. Membership functions corresponding to a fuzzy semantic set of liver cancer feature information. The multiplication symbol is used. This is the summation symbol.

2. The intelligent liver cancer staging method based on TS fuzzy semantics according to claim 1, characterized in that: The specific steps of step S4 are as follows: S41. The Chinese CNLC staging standard for liver cancer includes four categories after precise intelligent staging: Ia, Ib, IIa, and IIb-IV, with category labels represented by 0, 1, 2, and 3, respectively. S42. Randomly select patient data from the four types of data in proportion to form a training dataset, and use the remaining data as a test dataset. The proportions are set to 2:8, 3:7, and 4:

6. S43. Input the training dataset and test dataset into the constructed TS fuzzy semantic intelligent periodization model for periodization processing, and finally calculate the periodization accuracy.

3. The staging system for liver cancer based on the TS fuzzy semantic intelligent staging method according to claim 1, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: S1. Liver cancer feature extraction: Key liver cancer features are extracted from the obtained digital radiological diagnostic reports of patients to form a dataset; S2. Fuzzy representation of liver cancer features: multiple fuzzy semantic sets are used for fuzzy representation, and related membership functions are designed. S3.TS fuzzy semantic model construction includes designing TS fuzzy semantic rules, solving the membership degree of each rule, identifying consequent parameters, identifying the membership function of antecedent parameters, and outputting the TS fuzzy semantic model. S4. Intelligent staging and performance evaluation of liver cancer: The liver cancer categories in the Chinese CNLC standard are digitally labeled, divided into training and test sets, and the staging performance of the intelligent staging model is verified.

Images