Big data-based prediction and analysis platform and method for the treatment effect of femoral head necrosis

Abstract

The present invention relates to the field of medical information processing technology, and in particular to a big data-based femoral head necrosis treatment effect prediction and analysis platform and method. The platform includes a femoral head necrosis archive generation module, a femoral head image feature analysis module, a femoral head treatment feature analysis module, and a treatment effect prediction and analysis module. It can obtain multi-source medical data for femoral head necrosis treatment and perform data standard cleaning and medical archive integration, while performing femoral head feature analysis to obtain femoral head necrosis tissue image features; based on the femoral head necrosis treatment data archive, femoral head treatment staging and femoral head necrosis treatment feature analysis are performed to obtain basic femoral head necrosis treatment features; femoral head necrosis tissue image features and femoral head necrosis treatment basic features are screened for treatment effect influencing features and treatment effect prediction analysis are performed to predict and output the corresponding femoral head necrosis treatment effect. The present invention can achieve accurate prediction of the femoral head necrosis treatment effect.

Description

Technical Field

[0001] The present invention relates to the field of medical information processing technology, and in particular to a big data-based platform and method for predicting and analyzing the treatment effect of femoral head necrosis. Background Art

[0002] Avascular Necrosis of the Femoral Head (ANFH) is a common orthopedic disease that refers to the gradual necrosis of bone tissue in the femoral head due to interrupted blood supply, which in turn affects joint function. In recent years, with the rapid development of big data technology and artificial intelligence algorithms, by collecting and analyzing a large amount of clinical data, imaging data, genetic data and other information about patients with ANFH, machine learning and data mining technologies can be used to build more accurate prediction models to provide data support for treatment decisions. However, traditional treatment effect prediction methods mainly rely on the doctor's clinical experience, imaging examinations and some biomarker tests. Although these methods can reflect the effect of treatment to a certain extent, they can usually only reflect the current treatment status, and there are large errors in the estimation of treatment effects for different patients. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide a big data-based femoral head necrosis treatment effect prediction and analysis platform and method to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a big data-based femoral head necrosis treatment effect prediction and analysis platform is proposed, which includes the following modules:

[0005] The femoral head necrosis file generation module is used to obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and treatment plans for femoral head necrosis. It also performs data standardization cleaning and medical file integration on the multi-source medical data for the treatment of femoral head necrosis to generate a femoral head necrosis treatment data file;

[0006] The femoral head image feature analysis module is used to perform femoral head feature analysis on the corresponding femoral head medical images in the femoral head necrosis treatment data file to obtain the femoral head necrosis tissue image features, including the femoral head necrosis tissue structure, femoral head necrosis tissue morphology and femoral head necrosis tissue density;

[0007] The femoral head treatment feature analysis module is used to perform femoral head treatment stage on the corresponding femoral head necrosis treatment plan based on the corresponding femoral head clinical symptoms in the femoral head necrosis treatment data file, so as to obtain the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage; based on the patient's basic information and femoral head necrosis tissue density, the femoral head necrosis treatment feature analysis is performed on the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage to obtain the basic characteristics of femoral head necrosis treatment;

[0008] The treatment effect prediction and analysis module is used to screen the characteristics that affect the treatment effect of femoral head necrosis tissue imaging features and the basic characteristics of femoral head necrosis treatment to obtain the significant characteristics that affect the treatment effect of femoral head necrosis; construct a femoral head necrosis treatment effect prediction model, and input the significant characteristics that affect the treatment effect of femoral head necrosis into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans.

[0009] Furthermore, the femoral head necrosis file generation module includes the following functions:

[0010] Obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and a treatment plan for femoral head necrosis, wherein the basic patient information includes the patient's age, gender, previous medical history of the femoral head, and medication history; the clinical symptoms of the femoral head include the degree of femoral head pain and the range of limited femoral head movement; the medical images of the femoral head include X-ray images, CT images, and MRI images corresponding to the femoral head; and the treatment plan for femoral head necrosis includes treatment methods, medication dosage, and surgical records;

[0011] The multi-source medical data on the treatment of femoral head necrosis was cleaned to remove noise, duplicate data, and abnormal data. For image data, labeling errors were corrected and image coding was standardized using the unified medical image transmission standard DICOM. For text data, natural language processing was used to perform normalization and delete duplicate medical examination records to obtain multi-source standardized data on the treatment of femoral head necrosis.

[0012] The multi-source data on the treatment of femoral head necrosis corresponding to the same patient in the multi-source standard data on the treatment of femoral head necrosis are correlated and matched and integrated to generate a data file for the treatment of femoral head necrosis.

[0013] Furthermore, the femoral head image feature analysis module includes the following functions:

[0014] An in-depth analysis of the physical characteristics of the femoral head medical images corresponding to the femoral head necrosis treatment data files was performed. For X-ray images, the attenuation law of X-rays when passing through the femoral head tissue was studied. The relationship between the grayscale value and tissue density in the image was analyzed based on the differences in the degree of X-ray absorption by different tissues. For CT images, the femoral head tissue characteristics reflected by the voxel values ​​at different levels were deeply analyzed. For MRI images, the hydrogen proton density and relaxation time corresponding to the magnetic resonance signal and the femoral head tissue were studied to obtain a femoral head medical image physical characteristics dataset.

[0015] Perform femoral head necrotic tissue area segmentation on femoral head medical images to generate femoral head necrotic tissue area segmentation images;

[0016] The multi-scale texture feature analysis of the femoral head necrotic tissue area segmentation images at different scales was performed based on wavelet transform to obtain the texture feature set of the femoral head necrotic tissue area.

[0017] Based on the physical property dataset of femoral head medical images, the texture feature set of the femoral head necrotic tissue region was analyzed to invert the tissue structure corresponding to the femoral head necrotic tissue according to the corresponding grayscale and signal intensity in the image. The density distribution corresponding to the femoral head necrotic tissue was calculated by combining the known relationship between the X-ray attenuation coefficient and tissue density with the image voxel value. At the same time, the morphological characteristics corresponding to the femoral head necrotic tissue were inverted according to the relationship between T1 and T2 relaxation times and tissue morphology to obtain the image characteristics of the femoral head necrotic tissue, including the femoral head necrotic tissue structure, femoral head necrotic tissue morphology and femoral head necrotic tissue density.

[0018] Furthermore, the multi-scale texture feature analysis of the segmented images of the femoral head necrotic tissue area at different scales based on wavelet transform is performed. Specifically, at a small scale, the fine texture features corresponding to the femoral head tissue, including the microstructural texture corresponding to the trabeculae, are extracted through the high-frequency sub-band coefficients corresponding to the wavelet transform, while at a large scale, the fractal dimension is used to calculate and analyze the texture roughness and self-similarity corresponding to the entire femoral head.

[0019] Furthermore, the femoral head treatment feature analysis module includes the following functions:

[0020] Based on the corresponding femoral head pain level and femoral head motion limitation range in the femoral head necrosis treatment data file, the corresponding femoral head necrosis treatment plan is divided into femoral head treatment stages to obtain the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage;

[0021] Based on the corresponding femoral head medical history and drug use history in the patient's basic information, the femoral head necrosis treatment and allergy probability analysis is performed on the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage to obtain the femoral head necrosis treatment probability and femoral head necrosis treatment allergy probability;

[0022] Based on the density of femoral head necrotic tissue, the necrotic tissue attenuation evaluation and analysis were performed on the femoral head necrotic tissue treatment cases corresponding to each femoral head symptom stage to obtain the femoral head necrotic tissue treatment attenuation efficiency;

[0023] The probability of femoral head necrosis treatment, the probability of femoral head necrosis treatment allergy and the attenuation efficiency of femoral head necrosis tissue treatment are taken as basic features to obtain the basic features of femoral head necrosis treatment.

[0024] Furthermore, the femoral head symptom staging is specifically based on the femoral head pain level between 1-3 points and the femoral head movement limitation range between 10%-20% as the early stage, including the hip flexion angle from the normal 120°-140° to 100°-120°, the abduction angle from the normal 30°-45° to 25°-35°, and the internal rotation angle from the normal 30°-40° to 20°-30°; the femoral head pain level between 4-6 points and The mid-stage is when the restricted range of femoral head motion expands to between 20%-40%, including the hip flexion angle reduced to 80°-100°, the abduction angle reduced to 20°-25°, and the internal rotation angle reduced to 10°-20°; the late stage is when the femoral head pain level is between 7-10 points and the restricted range of femoral head motion expands by more than 40%, including the hip flexion angle less than 80°, the abduction angle less than 20°, and the internal rotation angle less than 10°.

[0025] Furthermore, the necrotic tissue attenuation evaluation and analysis of the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage based on the femoral head necrosis tissue density includes:

[0026] Based on the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage, the tissue density treatment simulation of the corresponding femoral head necrosis tissue density is performed to generate the density attenuation process of the femoral head necrosis tissue under each treatment effect;

[0027] The actual attenuation amount of the density attenuation process of the femoral head necrosis tissue under each treatment effect is statistically analyzed to obtain the actual attenuation amount of the tissue density under each treatment effect;

[0028] The theoretical maximum attenuation of femoral head necrotic tissue is obtained, and based on the theoretical maximum attenuation of femoral head necrotic tissue, the actual attenuation of tissue density corresponding to femoral head necrotic tissue under various treatment effects is quantitatively calculated to obtain the treatment attenuation efficiency of femoral head necrotic tissue.

[0029] Furthermore, the treatment effect prediction and analysis module includes the following functions:

[0030] Conduct treatment effect correlation mining analysis on the imaging characteristics of femoral head necrosis tissue and the various femoral head necrosis characteristic factors within the basic characteristics of femoral head necrosis treatment to obtain the correlation between each femoral head necrosis characteristic factor and the treatment effect;

[0031] Based on the correlation between each characteristic factor of femoral head necrosis and the treatment effect, the correlation between the imaging characteristics of femoral head necrosis tissue and each characteristic factor of femoral head necrosis within the basic characteristics of femoral head necrosis treatment was evaluated to obtain the characteristic correlation factor between each characteristic factor of femoral head necrosis and the treatment effect;

[0032] Causal inference of the therapeutic effect was performed on the imaging characteristics of femoral head necrosis tissue and each characteristic factor of femoral head necrosis within the basic characteristics of femoral head necrosis treatment, and the number of potential causal relationships between each characteristic factor of femoral head necrosis and the therapeutic effect was obtained;

[0033] Based on the characteristic correlation factors and potential causal relationship numbers between each characteristic factor of femoral head necrosis and the treatment effect, the imaging characteristics of femoral head necrosis tissue and the basic characteristics of femoral head necrosis treatment were screened for the characteristics affecting the treatment effect, so as to obtain the significant characteristics affecting the treatment effect of femoral head necrosis;

[0034] A convolutional neural network was used to construct a femoral head necrosis treatment effect prediction model, and the significant features affecting the femoral head necrosis treatment effect prediction model was input into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis, so as to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans.

[0035] Furthermore, the femoral head necrosis treatment effect prediction model is specifically a pyramid network architecture that extracts treatment effect features corresponding to multiple scales. In the shallow network, the architecture uses a 3x3 corresponding convolution kernel and a 1x1 pooling window to extract the detailed features corresponding to the femoral head, including tiny trabecular changes and the boundaries of early necrotic areas, while in the deep network, the architecture uses a 5x5 corresponding convolution kernel and a 1x1 pooling window to extract the overall features corresponding to the femoral head, including joint morphology and the macroscopic distribution of necrotic areas. By introducing energy constraints, the activation energy and information transfer energy of each neuron in the network are included in the optimization target to focus on the prediction error corresponding to the model. At the same time, cross-validation is used to improve the corresponding prediction performance of the model, so as to predict the corresponding femoral head necrosis treatment effect at different network layers.

[0036] Furthermore, the present invention also provides a method for predicting and analyzing the therapeutic effect of femoral head necrosis based on big data, which is implemented based on the above-mentioned platform for predicting and analyzing the therapeutic effect of femoral head necrosis based on big data. The method for predicting and analyzing the therapeutic effect of femoral head necrosis based on big data includes:

[0037] Obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and treatment plans for femoral head necrosis. Perform data standardization cleaning and medical file integration on the multi-source medical data on the treatment of femoral head necrosis to generate a femoral head necrosis treatment data file.

[0038] Perform femoral head feature analysis on the corresponding femoral head medical images in the femoral head necrosis treatment data file to obtain the femoral head necrosis tissue image features, including femoral head necrosis tissue structure, femoral head necrosis tissue morphology, and femoral head necrosis tissue density;

[0039] Based on the corresponding femoral head clinical symptoms in the femoral head necrosis treatment data file, the corresponding femoral head necrosis treatment plan is divided into femoral head treatment stages to obtain the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage; based on the patient's basic information and femoral head necrosis tissue density, the femoral head necrosis treatment characteristics of the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage are analyzed to obtain the basic characteristics of femoral head necrosis treatment;

[0040] The imaging features of femoral head necrosis tissue and the basic characteristics of femoral head necrosis treatment are screened for features that affect the treatment effect, so as to obtain the significant features that affect the treatment effect of femoral head necrosis; a femoral head necrosis treatment effect prediction model is constructed, and the significant features that affect the treatment effect of femoral head necrosis are input into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis, so as to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans.

[0041] Beneficial effects of the present invention:

[0042] The big data-based femoral head necrosis treatment effect prediction and analysis platform proposed in the present invention is composed of a femoral head necrosis file generation module, a femoral head image feature analysis module, a femoral head treatment feature analysis module and a treatment effect prediction and analysis module. Compared with the existing technology, the beneficial effect of this application lies in collecting and organizing multi-source medical data on femoral head necrosis treatment, covering patient basic information, clinical symptoms, medical images and treatment plans, and standardizing and cleaning these data. The diversity of the data includes basic information such as the patient's age, gender, medical history, living habits, and clinical symptoms of femoral head necrosis, such as pain, limited activity, etc. , as well as femoral head injury information obtained through imaging examinations (such as X-rays, MRI, CT, etc.), through standardized cleaning, redundant data can be removed, missing data can be repaired, and data formats from different sources can be standardized to make them consistent and comparable. The integration of medical records can also ensure a complete record of each patient's treatment process and related information. This process lays a solid foundation for subsequent data analysis and model establishment. The integrated and cleaned data can provide more accurate and comprehensive patient information and provide a real and effective basis for subsequent analysis of the treatment effect of femoral head necrosis, thereby ensuring the accuracy and reliability of the final treatment prediction results. Secondly, a detailed feature analysis of the femoral head medical images in the femoral head necrosis treatment data archive is performed to extract the imaging features of femoral head necrosis. These imaging features include the structure, morphology, and density of femoral head necrotic tissue. Through imaging technology, a quantitative description of femoral head lesions can be achieved, such as morphological changes in the femoral head necrotic area, tissue damage, and density changes in the necrotic area. This analysis can reveal the imaging manifestations of femoral head necrosis at different stages and provide quantitative support for subsequent prediction of treatment effects. Then, through the analysis of the patient's clinical symptoms, the treatment plan for femoral head necrosis can be divided into different stages or sub-plans. For example, early femoral head necrosis only requires conservative treatment, while late stage requires surgical intervention. Through further analysis of the patient's basic information (such as age, gender, etc.) and imaging characteristics (such as tissue density), the most appropriate treatment plan can be selected for each treatment stage, and the treatment strategy can be adjusted individually. Feature analysis not only helps to formulate staged treatment plans, but also reflects the treatment effects at different symptom stages. This analysis process can help doctors accurately evaluate the current corresponding treatment status and provide individualized treatment plans for patients at different stages, thereby reducing the error in the estimated treatment effects of different patients in the subsequent process.Finally, by analyzing the factors affecting the treatment of femoral head necrosis, we screened out the features that have a significant impact on the treatment effect, and established a prediction model for the treatment effect of femoral head necrosis. The treatment effect of femoral head necrosis is affected by many factors, such as the patient's age, gender, treatment plan, disease progression stage and imaging characteristics. By screening out the features that significantly affect the treatment effect, the accuracy of the treatment effect prediction can be improved. The prediction model can be trained based on historical data and use machine learning or statistical methods to establish an efficient prediction system. Through this model, the effects of different treatment plans in specific patient groups can be predicted in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0044] Figure 1 Schematic diagram of the module of the femoral head necrosis treatment effect prediction and analysis platform based on big data of the present invention;

[0045] Figure 2 for Figure 1 Schematic diagram of the functional flow of the femoral head necrosis file generation module;

[0046] Figure 3 for Figure 1 Schematic diagram of the functional flow of the femoral head image feature analysis module. DETAILED DESCRIPTION

[0047] The following is a clear and complete description of the technical platform of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0048] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or across different networks and / or processor platforms and / or microcontroller platforms.

[0049] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0050] To achieve this, please refer to Figures 1 to 3 The present invention provides a big data-based prediction and analysis platform for the treatment effect of femoral head necrosis, which includes the following modules:

[0051] The femoral head necrosis file generation module is used to obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and treatment plans for femoral head necrosis. It also performs data standardization cleaning and medical file integration on the multi-source medical data for the treatment of femoral head necrosis to generate a femoral head necrosis treatment data file;

[0052] The femoral head image feature analysis module is used to perform femoral head feature analysis on the corresponding femoral head medical images in the femoral head necrosis treatment data file to obtain the femoral head necrosis tissue image features, including the femoral head necrosis tissue structure, femoral head necrosis tissue morphology and femoral head necrosis tissue density;

[0053] The femoral head treatment feature analysis module is used to perform femoral head treatment stage on the corresponding femoral head necrosis treatment plan based on the corresponding femoral head clinical symptoms in the femoral head necrosis treatment data file, so as to obtain the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage; based on the patient's basic information and femoral head necrosis tissue density, the femoral head necrosis treatment feature analysis is performed on the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage to obtain the basic characteristics of femoral head necrosis treatment;

[0054] The treatment effect prediction and analysis module is used to screen the characteristics that affect the treatment effect of femoral head necrosis tissue imaging features and the basic characteristics of femoral head necrosis treatment to obtain the significant characteristics that affect the treatment effect of femoral head necrosis; construct a femoral head necrosis treatment effect prediction model, and input the significant characteristics that affect the treatment effect of femoral head necrosis into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans.

[0055] In the embodiment of the present invention, please refer to Figure 1FIG. 1 is a schematic diagram of a module of a big data-based femoral head necrosis treatment effect prediction and analysis platform of the present invention. In this example, the big data-based femoral head necrosis treatment effect prediction and analysis platform includes the following modules:

[0056] S1: A femoral head necrosis archive generation module is used to obtain multi-source medical data on femoral head necrosis treatment, including basic patient information, femoral head clinical symptoms, femoral head medical images, and femoral head necrosis treatment plans. It also performs data standardization cleaning and medical archive integration on the multi-source medical data on femoral head necrosis treatment to generate a femoral head necrosis treatment data archive;

[0057] In an embodiment of the present invention, multi-source medical data for the treatment of femoral head necrosis are collected from multiple channels. The basic information of the patient is obtained from the hospital registration system and medical record system, covering age, gender, past medical history and history of drug use. The clinical symptoms of the femoral head are recorded by the doctor through interviews and physical examinations, including the degree of pain and the range of motion restriction. Medical images of the femoral head are obtained from the archiving system of the imaging department, including X-ray, CT and MRI images. The treatment plan for femoral head necrosis is obtained from the treatment record database, including treatment methods, drug dosages and surgical records. For the collected data, statistical analysis methods are used to identify and remove noise, duplication and abnormal data. The image data is encoded according to the DICOM standard and annotation errors are corrected. The text data is normalized using natural language processing technology and duplicate examination records are deleted. Finally, the patient's unique identifier is used as an index to associate and integrate various types of data for the same patient and store them in a special database, ultimately forming a femoral head necrosis treatment data archive.

[0058] S2: Femoral head image feature analysis module, used to perform femoral head feature analysis on the corresponding femoral head medical images in the femoral head necrosis treatment data file to obtain the femoral head necrosis tissue image features, including the femoral head necrosis tissue structure, femoral head necrosis tissue morphology and femoral head necrosis tissue density;

[0059] In an embodiment of the present invention, X-ray, CT and MRI images in the femoral head necrosis treatment data archive are analyzed. For X-ray images, the relationship between grayscale value and tissue density is established based on the attenuation law of X-rays passing through different tissues, and the tissue structure is analyzed. For CT images, the voxel values ​​of different layers are analyzed, and the tissue characteristics are determined in combination with medical knowledge. For MRI images, the relationship between magnetic resonance signals and hydrogen proton density and relaxation time is studied, and the necrotic tissue area of ​​the image is segmented by threshold segmentation and morphological operations. At a small scale, the trabecular microstructure texture is extracted by wavelet transform high-frequency sub-band coefficients; at a large scale, the fractal dimension is used to analyze the overall texture roughness and self-similarity. According to the physical properties and texture characteristics of the image, the tissue structure is inverted, the density distribution is calculated, and the morphological characteristics are inferred, and finally the image characteristics of femoral head necrosis tissue are obtained.

[0060] S3: Femoral head treatment feature analysis module, used to perform femoral head treatment stage for corresponding femoral head necrosis treatment plans based on the corresponding femoral head clinical symptoms in the femoral head necrosis treatment data file, so as to obtain femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage; based on the patient's basic information and femoral head necrosis tissue density, perform femoral head necrosis treatment feature analysis on the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage, and obtain the basic characteristics of femoral head necrosis treatment;

[0061] In an embodiment of the present invention, the treatment plan is divided into early, middle and late stages based on the clinical symptoms of the femoral head in the femoral head necrosis treatment data file, such as pain level and limited range of activity. The early stage is a pain level of 1-3 points and a limited range of activity of 10%-20%, the middle stage is a pain level of 4-6 points and a limited range of activity of 20%-40%, and the late stage is a pain level of 7-10 points and a limited range of activity of more than 40%. Treatment sub-cases are formulated for each stage. Combined with the patient's basic information, such as age, medical history and drug use history, as well as the density of femoral head necrosis tissue, big data analysis and probability statistical models are used to analyze the treatment probability and allergy probability of the treatment sub-cases, obtain the initial and final density of the tissue through medical imaging, calculate the attenuation efficiency of the necrotic tissue, and use the treatment probability, allergy probability and attenuation efficiency as basic characteristics to finally obtain the basic characteristics of femoral head necrosis treatment.

[0062] S4: Treatment effect prediction and analysis module, which is used to screen the characteristics affecting the treatment effect of femoral head necrosis tissue imaging features and basic characteristics of femoral head necrosis treatment to obtain the significant characteristics affecting the treatment effect of femoral head necrosis; construct a femoral head necrosis treatment effect prediction model, and input the significant characteristics affecting the treatment effect of femoral head necrosis into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis, so as to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans.

[0063] In an embodiment of the present invention, screening criteria are set by analyzing the imaging features of femoral head necrosis tissue (tissue structure, morphology, density) and the basic characteristics of femoral head necrosis treatment (treatment probability, allergy probability, attenuation efficiency). For example, the absolute value of the feature correlation factor is greater than 0.5 and the number of potential causal relationships is greater than 2, and significant influencing features are screened out. Python and a deep learning framework are used to construct a convolutional neural network with a pyramid network architecture as a prediction model. A 3x3 convolution kernel and a 1x1 pooling window are used in the shallow layer to extract detail features, and a 5x5 convolution kernel and a 1x1 pooling window are used in the deep layer to extract overall features. By introducing energy constraints, neuron activation and information transfer energy are included in the optimization target. At the same time, cross-validation is used to divide the data set for training and verification of the model, and the screened significant features are input into the model to predict the treatment effects of femoral head necrosis under different treatment plans, such as the degree of pain relief and recovery of joint function.

[0064] Furthermore, the femoral head necrosis file generation module includes the following functions:

[0065] Obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and a treatment plan for femoral head necrosis, wherein the basic patient information includes the patient's age, gender, previous medical history of the femoral head, and medication history; the clinical symptoms of the femoral head include the degree of femoral head pain and the range of limited femoral head movement; the medical images of the femoral head include X-ray images, CT images, and MRI images corresponding to the femoral head; and the treatment plan for femoral head necrosis includes treatment methods, medication dosage, and surgical records;

[0066] The multi-source medical data on the treatment of femoral head necrosis was cleaned to remove noise, duplicate data, and abnormal data. For image data, labeling errors were corrected and image coding was standardized using the unified medical image transmission standard DICOM. For text data, natural language processing was used to perform normalization and delete duplicate medical examination records to obtain multi-source standardized data on the treatment of femoral head necrosis.

[0067] The multi-source data on the treatment of femoral head necrosis corresponding to the same patient in the multi-source standard data on the treatment of femoral head necrosis are correlated and matched and integrated to generate a data file for the treatment of femoral head necrosis.

[0068] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 The functional flow chart of the femoral head necrosis file generation module in this embodiment includes the following functions:

[0069] S11: Acquire multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and a treatment plan for femoral head necrosis, wherein the basic patient information includes the patient's age, gender, past medical history of the femoral head, and history of medication use; the clinical symptoms of the femoral head include the degree of femoral head pain and the range of limited femoral head movement; the medical images of the femoral head include X-ray images, CT images, and MRI images of the femoral head; and the treatment plan for femoral head necrosis includes treatment methods, medication dosage, and surgical records;

[0070] In an embodiment of the present invention, data related to the treatment of femoral head necrosis are collected from multiple medical data sources, and the patient's basic information is obtained from the hospital's electronic medical record system. The system will record the patient's age, gender, previous medical history of the femoral head and history of drug use in detail. For the clinical symptoms of the femoral head, the doctor's interview and physical examination of the patient are recorded, and the degree of femoral head pain is determined using a pain scoring scale. The range of restricted femoral head movement is determined by measuring the angle of motion of the hip joint. The medical imaging data of the femoral head is obtained from the hospital's picture archiving and communication system (PACS), which stores the patient's X-ray images, CT images and MRI images. The data of the femoral head necrosis treatment plan is obtained from the hospital's treatment record database, which includes information such as treatment methods (such as conservative treatment, surgical treatment, etc.), drug dosage and surgical records. These data obtained from different data sources are integrated to finally form multi-source medical data for the treatment of femoral head necrosis.

[0071] S12: Perform data standardization and cleaning on multi-source medical data on the treatment of femoral head necrosis to remove corresponding noise data, duplicate data, and abnormal data. For image data, correct the corresponding image labeling errors and use the unified medical image transmission standard DICOM to perform image coding standardization. For text data, use natural language processing to perform standardization and delete duplicate medical examination records to obtain multi-source standard data on the treatment of femoral head necrosis;

[0072] In an embodiment of the present invention, a cleaning operation is performed on multi-source medical data for the treatment of femoral head necrosis. For numerical data, such as patient age, drug dosage, etc., a reasonable value range is set to identify and remove abnormal data. At the same time, data records are compared and duplicate data are deleted. For noise data, a filtering algorithm is used for processing, such as median filtering, to remove random interference in the data. For image data, professional medical imaging personnel are arranged to review the image annotations and correct annotation errors. Then all image data are encoded according to the medical image transmission standard DICOM to ensure that the format of the image data is unified. For text data, natural language processing techniques, such as word segmentation, part-of-speech tagging, named entity recognition, etc., are used to normalize the text. At the same time, duplicate medical examination records are deleted by comparing the text content. After these processing steps, standardized and accurate multi-source standard data for the treatment of femoral head necrosis are finally obtained.

[0073] S13: Perform correlation matching and integrated file integration on the multi-source data on the treatment of femoral head necrosis corresponding to the same patient in the multi-source standard data on the treatment of femoral head necrosis to generate a data file on the treatment of femoral head necrosis.

[0074] In an embodiment of the present invention, the patient's basic information, femoral head clinical symptoms, femoral head medical images and femoral head necrosis treatment plan data of the same patient are integrated by using the patient's unique identifier (such as ID number, medical record number, etc.) in the multi-source standard data for the treatment of femoral head necrosis. First, all relevant records of the patient are searched in different types of data tables using the patient identifier as an index. Then, these records are organized according to a certain structure, for example, a comprehensive data structure is created that includes a patient basic information table, a clinical symptom table, an imaging record table and a treatment plan table. For imaging data, its storage path is associated with the corresponding patient record. Finally, the integrated data is stored in a special database to form a complete femoral head necrosis treatment data file, which is convenient for subsequent analysis and utilization of the data.

[0075] Furthermore, the femoral head image feature analysis module includes the following functions:

[0076] An in-depth analysis of the physical characteristics of the femoral head medical images corresponding to the femoral head necrosis treatment data files was performed. For X-ray images, the attenuation law of X-rays when passing through the femoral head tissue was studied. The relationship between the grayscale value and tissue density in the image was analyzed based on the differences in the degree of X-ray absorption by different tissues. For CT images, the femoral head tissue characteristics reflected by the voxel values ​​at different levels were deeply analyzed. For MRI images, the hydrogen proton density and relaxation time corresponding to the magnetic resonance signal and the femoral head tissue were studied to obtain a femoral head medical image physical characteristics dataset.

[0077] Perform femoral head necrotic tissue area segmentation on femoral head medical images to generate femoral head necrotic tissue area segmentation images;

[0078] The multi-scale texture feature analysis of the femoral head necrotic tissue area segmentation images at different scales was performed based on wavelet transform to obtain the texture feature set of the femoral head necrotic tissue area.

[0079] Based on the physical property dataset of femoral head medical images, the texture feature set of the femoral head necrotic tissue region was analyzed to invert the tissue structure corresponding to the femoral head necrotic tissue according to the corresponding grayscale and signal intensity in the image. The density distribution corresponding to the femoral head necrotic tissue was calculated by combining the known relationship between the X-ray attenuation coefficient and tissue density with the image voxel value. At the same time, the morphological characteristics corresponding to the femoral head necrotic tissue were inverted according to the relationship between T1 and T2 relaxation times and tissue morphology to obtain the image characteristics of the femoral head necrotic tissue, including the femoral head necrotic tissue structure, femoral head necrotic tissue morphology and femoral head necrotic tissue density.

[0080] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 1The functional flow diagram of the femoral head image feature analysis module is as follows. In this embodiment, the femoral head image feature analysis module includes the following functions:

[0081] S21: Conduct an in-depth analysis of the physical characteristics of the femoral head medical images corresponding to the femoral head necrosis treatment data files. For X-ray images, study the attenuation law of X-rays when passing through the femoral head tissue, and analyze the relationship between the grayscale value and tissue density in the image based on the differences in the degree of X-ray absorption of different tissues. For CT images, deeply analyze the femoral head tissue characteristics reflected by the voxel values ​​at different levels. For MRI images, study the magnetic resonance signal and the hydrogen proton density and relaxation time corresponding to the femoral head tissue, and obtain a femoral head medical image physical characteristics dataset;

[0082] In an embodiment of the present invention, three types of femoral head medical images, namely X-ray, CT and MRI, are extracted from the femoral head necrosis treatment data archive. For X-ray images, the X-ray physics principle is used to measure the intensity change of X-rays before and after passing through the femoral head tissue, and its attenuation law is analyzed. By experimentally comparing the X-ray absorption of tissues with different densities (such as bones, muscles, and fat), a mathematical model of grayscale values ​​and tissue density in the image is established. For CT images, the voxel values ​​of each layer are analyzed in detail. According to the CT imaging principle, the voxel value is related to the electron density of the tissue. By consulting medical information and databases, the femoral head tissue characteristics corresponding to different voxel value ranges, such as normal bone tissue, necrotic bone tissue, etc., are determined. For MRI images, professional magnetic resonance imaging analysis software is used to measure the magnetic resonance signal intensity, and combined with the known hydrogen proton density and relaxation time (T1, T2) theory, the relationship between the signal and them is studied. All data obtained from the above analysis are sorted and summarized to finally form a femoral head medical image physical property data set.

[0083] S22: performing femoral head necrotic tissue region segmentation on the femoral head medical image to generate a femoral head necrotic tissue region segmentation image;

[0084] In an embodiment of the present invention, medical images of the femoral head are processed by adopting a method based on a combination of threshold segmentation and morphological operations. For X-ray, CT and MRI images, the approximate grayscale value or signal intensity range of the femoral head necrotic tissue in the image is first determined based on the image physical property data set, and this is used as a threshold for preliminary segmentation. For example, in CT images, the voxel value of necrotic tissue is usually in a specific interval. The interval is used as a threshold to binarize the image to obtain a preliminary necrotic tissue area. Then, morphological operations such as dilation and erosion are used to remove small noise and holes in the segmentation results, fill the blank parts in the necrotic tissue area, and make the segmentation results more accurate and complete. Finally, the boundary of the segmented image is corrected to ensure that the boundary of the necrotic tissue area is clear and accurate, thereby generating a segmented image of the femoral head necrotic tissue area.

[0085] S23: Using wavelet transform to perform multi-scale texture feature analysis on the femoral head necrotic tissue area segmentation images at different scales, the texture feature set of the femoral head necrotic tissue area is obtained;

[0086] In an embodiment of the present invention, the image of the femoral head necrotic tissue area is segmented and wavelet transformed to decompose the image into sub-bands of different scales. At a small scale, the high-frequency sub-band coefficients after wavelet transformation are focused on. These coefficients reflect the detail information in the image. The microstructural texture features corresponding to the trabeculae in the femoral head tissue are obtained by extracting features from the high-frequency sub-band coefficients, such as calculating texture energy and contrast. For example, the arrangement and density of the trabeculae will show specific patterns in the high-frequency sub-band coefficients. By analyzing these patterns, subtle texture features can be extracted. At a large scale, the fractal dimension calculation method is used to analyze the entire femoral head necrotic tissue area. The fractal dimension can measure the roughness and self-similarity of the texture. By calculating the fractal dimension, the quantitative features of the overall texture of the femoral head can be obtained. The texture features extracted at small and large scales are summarized to finally obtain a texture feature set of the femoral head necrotic tissue area.

[0087] S24: Based on the femoral head medical imaging physical property dataset, the texture feature set of the femoral head necrotic tissue area is analyzed to invert the tissue structure corresponding to the femoral head necrotic tissue according to the corresponding grayscale and signal intensity in the image, and the density distribution corresponding to the femoral head necrotic tissue is calculated by combining the known relationship between the X-ray attenuation coefficient and tissue density with the image voxel value. At the same time, the morphological characteristics corresponding to the femoral head necrotic tissue are inverted according to the relationship between T1 and T2 relaxation time and tissue morphology to obtain the image characteristics of the femoral head necrotic tissue, including the femoral head necrotic tissue structure, femoral head necrotic tissue morphology and femoral head necrotic tissue density.

[0088] In an embodiment of the present invention, by combining the physical property data set of medical images of the femoral head and the texture feature set of the femoral head necrotic tissue area for analysis, the grayscale and signal intensity in the image are inferred based on the previously established relationship between the grayscale value and tissue density, the magnetic resonance signal and the hydrogen proton density and relaxation time. For example, in X-ray images, higher grayscale values ​​may correspond to higher density bone tissue. By analyzing the grayscale distribution, the structure of the trabeculae in the necrotic tissue can be understood. By using the known relationship between the X-ray attenuation coefficient and tissue density, combined with the voxel value in the CT image, each voxel is Calculations are performed to obtain the density distribution of the femoral head necrotic tissue. For example, based on the attenuation coefficient formula and voxel value, the tissue density corresponding to each voxel is calculated, and then a density distribution map is drawn. For the T1 and T2 relaxation times in the MRI image, based on the correspondence between the T1 and T2 relaxation times and tissue morphology in medical research, the morphological characteristics of the femoral head necrotic tissue, such as the size and shape of the necrotic area, are inverted. These tissue structures, density distributions and morphological characteristics are sorted out, and finally the femoral head necrotic tissue image characteristics including the femoral head necrotic tissue structure, femoral head necrotic tissue morphology and femoral head necrotic tissue density are obtained.

[0089] Furthermore, the multi-scale texture feature analysis of the segmented images of the femoral head necrotic tissue area at different scales based on wavelet transform is performed. Specifically, at a small scale, the fine texture features corresponding to the femoral head tissue, including the microstructural texture corresponding to the trabeculae, are extracted through the high-frequency sub-band coefficients corresponding to the wavelet transform, while at a large scale, the fractal dimension is used to calculate and analyze the texture roughness and self-similarity corresponding to the entire femoral head.

[0090] Furthermore, the femoral head treatment feature analysis module includes the following functions:

[0091] Based on the corresponding femoral head pain level and femoral head motion limitation range in the femoral head necrosis treatment data file, the corresponding femoral head necrosis treatment plan is divided into femoral head treatment stages to obtain the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage;

[0092] In an embodiment of the present invention, the femoral head pain degree score and femoral head motion restriction range data of each patient are extracted from the femoral head necrosis treatment data file. The pain degree score adopts a 1-10 point system, and the patient makes a subjective evaluation based on his or her own feelings. The motion restriction range is calculated by measuring the flexion, abduction and internal rotation angles of the hip joint and comparing them with the normal motion angle range. For example, the femoral head pain degree between 1-3 points and the femoral head motion restriction range between 10%-20% are in the early stage, including the hip joint flexion angle reduced from the normal 120°-140° to 100°-120°, the abduction angle reduced from the normal 30°-45° to 25°-35°, and The internal rotation angle decreased from the normal 30°-40° to 20°-30°; the mid-term stage was when the femoral head pain level was between 4-6 points and the range of restricted femoral head motion expanded to 20%-40%, including the hip flexion angle decreased to 80°-100°, the abduction angle decreased to 20°-25°, and the internal rotation angle decreased to 10°-20°; the late stage was when the femoral head pain level was between 7-10 points and the range of restricted femoral head motion expanded by more than 40%, including the hip flexion angle less than 80°, the abduction angle less than 20°, and the internal rotation angle less than 10°. Finally, the corresponding femoral head necrosis treatment sub-cases were obtained under each femoral head symptom stage.

[0093] Preferably, based on the femoral head medical history and drug use history corresponding to the patient's basic information, the femoral head necrosis treatment and allergy probability analysis is performed on the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage to obtain the femoral head necrosis treatment probability and the femoral head necrosis treatment allergy probability;

[0094] In an embodiment of the present invention, the patient's previous medical history of the femoral head is collected from the patient's basic information, including whether there has been any fracture, dislocation, etc., as well as the history of drug use, such as whether corticosteroids and other drugs that may cause femoral head necrosis have been used. For treatment sub-cases under each femoral head symptom stage, big data analysis methods and probability statistical models are used for analysis. For example, a database containing a large amount of patient data is established to analyze the treatment effects and allergic conditions of patients with a specific medical history or who have used specific drugs when receiving different treatment sub-cases. For early treatment sub-cases, the ratio of the number of patients with a history of fracture and who have used corticosteroids who are cured when receiving the treatment sub-case to the total number of people is counted to obtain the probability of femoral head necrosis treatment; the ratio of the number of people who have allergic reactions to the total number of people is counted to obtain the probability of femoral head necrosis treatment allergy. The same analysis is performed on the mid-term and late-stage treatment sub-cases, and finally the necrosis treatment probability and allergy probability corresponding to the treatment sub-cases under each symptom stage are obtained, thereby obtaining the femoral head necrosis treatment probability and the femoral head necrosis treatment allergy probability.

[0095] Preferably, based on the density of femoral head necrotic tissue, the femoral head necrotic tissue attenuation evaluation and analysis are performed on the femoral head necrotic tissue treatment sub-cases corresponding to each femoral head symptom stage to obtain the femoral head necrotic tissue treatment attenuation efficiency;

[0096] In an embodiment of the present invention, the initial density data of the necrotic tissue of the femoral head of the patient is obtained by using medical imaging technology (such as CT scanning). For each treatment case under each femoral head symptom stage, the initial density of the necrotic tissue is recorded before the treatment begins. During the treatment, CT scans are performed at regular intervals (e.g., once a month) to obtain real-time density data of necrotic tissue. After the treatment, the final density of necrotic tissue is recorded. , by calculating the density difference , and obtain the actual attenuation of necrotic tissue. At the same time, through the analysis of a large amount of clinical data and medical research, the theoretical maximum attenuation of femoral head necrotic tissue under different symptom stages is determined. , using the formula Calculate the attenuation efficiency of femoral head necrosis for each treatment case. For example, in the early treatment case, the initial density is 1.2g / cm³, the final density is 1.0g / cm³, and the theoretical maximum attenuation is 0.3g / cm³. The treatment attenuation efficiency is 1.2-1.0 / 0.3=66.7, which is the final treatment attenuation efficiency of femoral head necrosis.

[0097] Preferably, the probability of treating femoral head necrosis, the probability of allergic reaction to treating femoral head necrosis and the attenuation efficiency of treating femoral head necrosis tissue are taken as basic features to obtain the basic features of treating femoral head necrosis.

[0098] In an embodiment of the present invention, after completing the previous calculations, the femoral head necrosis treatment probability, femoral head necrosis treatment allergy probability and femoral head necrotic tissue treatment attenuation efficiency corresponding to the treatment sub-cases obtained under each femoral head symptom stage are sorted out, and the three characteristic values ​​of each treatment sub-case are recorded in a table to form a data set. For example, the treatment probability of early treatment sub-case 1 is 80%, the allergy probability is 5%, and the treatment attenuation efficiency is 70%; the treatment probability of mid-term treatment sub-case 2 is 60%, the allergy probability is 10%, and the treatment attenuation efficiency is 50%, etc. These data are combined together to constitute the basic characteristics of femoral head necrosis treatment. These basic characteristics can be used for subsequent treatment effect prediction and analysis, and finally the basic characteristics of femoral head necrosis treatment are obtained.

[0099] Furthermore, the femoral head symptom staging is specifically based on the femoral head pain level between 1-3 points and the femoral head movement limitation range between 10%-20% as the early stage, including the hip flexion angle from the normal 120°-140° to 100°-120°, the abduction angle from the normal 30°-45° to 25°-35°, and the internal rotation angle from the normal 30°-40° to 20°-30°; the femoral head pain level between 4-6 points and The mid-stage is when the restricted range of femoral head motion expands to between 20%-40%, including the hip flexion angle reduced to 80°-100°, the abduction angle reduced to 20°-25°, and the internal rotation angle reduced to 10°-20°; the late stage is when the femoral head pain level is between 7-10 points and the restricted range of femoral head motion expands by more than 40%, including the hip flexion angle less than 80°, the abduction angle less than 20°, and the internal rotation angle less than 10°.

[0100] Furthermore, the necrotic tissue attenuation evaluation and analysis of the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage based on the femoral head necrosis tissue density includes:

[0101] Based on the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage, the tissue density treatment simulation of the corresponding femoral head necrosis tissue density is performed to generate the density attenuation process of the femoral head necrosis tissue under each treatment effect;

[0102] In an embodiment of the present invention, by targeting different femoral head symptom stages, such as early, middle and late stages, corresponding femoral head necrosis treatment sub-cases are determined respectively, and these treatment sub-cases contain information such as specific treatment methods, treatment cycles and drug dosages. Medical imaging technology (such as CT scanning) is used to obtain the initial density data of femoral head necrosis tissue in each stage, and a three-dimensional femoral head necrosis tissue model is constructed. In a computer simulation environment, based on the treatment parameters in the treatment sub-case, the mechanism of action of drugs on necrotic tissue, the effect of physical therapy on tissue, etc. are simulated. For example, for drug treatment, based on the composition and action principle of the drug, the interaction between drug molecules and necrotic tissue cells is simulated, leading to apoptosis and decomposition of tissue cells, thereby causing changes in tissue density. According to the treatment cycle, the entire treatment process is divided into multiple time steps, and the tissue density data is updated in each time step, and finally the density attenuation process corresponding to femoral head necrosis tissue under each treatment action is generated.

[0103] Preferably, actual attenuation statistics are performed on the density attenuation process of the femoral head necrosis tissue under various treatments to obtain the actual tissue density attenuation of the femoral head necrosis tissue under various treatments;

[0104] In the embodiment of the present invention, after completing the previous density decay process simulation, the tissue density data of each time step is analyzed, and for each treatment case corresponding to the density decay process, the tissue density value at the initial moment is recorded. and tissue density at the end of treatment , by calculating the difference between the two , the overall density attenuation of femoral head necrotic tissue under the treatment is obtained. In order to more accurately calculate the actual attenuation, the tissue density values ​​can be recorded at multiple key time points during the treatment process, the density differences between adjacent time points can be calculated, and these differences can be accumulated. For example, three key time points are selected during the treatment cycle. , and the corresponding tissue density values ​​are 、 、 , then the actual attenuation , such statistics are performed on the density attenuation process under each treatment effect, and finally the actual tissue density attenuation of femoral head necrosis tissue under each treatment effect is obtained.

[0105] Preferably, the theoretical maximum attenuation of the femoral head necrotic tissue is obtained, and based on the theoretical maximum attenuation of the femoral head necrotic tissue, the actual attenuation of the tissue density corresponding to the femoral head necrotic tissue under various treatment effects is quantitatively calculated to obtain the treatment attenuation efficiency of the femoral head necrotic tissue.

[0106] In an embodiment of the present invention, a large amount of medical research and clinical data are analyzed to determine the theoretical maximum attenuation of femoral head necrotic tissue. The theoretical maximum attenuation refers to the maximum density attenuation value that femoral head necrotic tissue can reach under ideal treatment conditions. It is usually related to factors such as the initial range and nature of the necrotic tissue. It is assumed that the theoretical maximum attenuation of femoral head necrotic tissue under a certain symptom stage is determined by research. , for the actual attenuation of tissue density under each treatment effect obtained previously ( represents different treatment sub-cases), using the formula Perform quantitative calculation of necrotic tissue attenuation. For example, the actual attenuation of tissue density under a certain treatment sub-case is 0.5g / cm³, and the theoretical maximum attenuation under the symptom stage is 1g / cm³. Then the treatment attenuation efficiency of femoral head necrotic tissue corresponding to the treatment sub-case is 50%. Such calculation is performed on the actual attenuation under all treatment effects, and finally the treatment attenuation efficiency of femoral head necrotic tissue under each treatment effect is obtained.

[0107] Furthermore, the treatment effect prediction and analysis module includes the following functions:

[0108] Conduct treatment effect correlation mining analysis on the imaging characteristics of femoral head necrosis tissue and the various femoral head necrosis characteristic factors within the basic characteristics of femoral head necrosis treatment to obtain the correlation between each femoral head necrosis characteristic factor and the treatment effect;

[0109] In an embodiment of the present invention, a large amount of data on patients with femoral head necrosis is collected, wherein the imaging features of femoral head necrosis tissue are obtained through medical imaging equipment (such as magnetic resonance imaging, MRI, X-ray, etc.), including information such as trabecular structure, size and position of necrotic area; the basic characteristics of femoral head necrosis treatment cover factors such as the probability of femoral head necrosis treatment, the probability of allergic reaction to femoral head necrosis treatment, and the attenuation efficiency of femoral head necrosis tissue treatment. An association rule mining algorithm, such as the Apriori algorithm, is used to process data on these characteristic factors and treatment effects (such as the degree of pain relief, joint function recovery, etc.). Taking the Apriori algorithm as an example, the minimum support and minimum confidence thresholds are set, and the frequent item sets are found by scanning the data multiple times, thereby obtaining the correlation between each femoral head necrosis characteristic factor and the treatment effect. For example, it is found that when the patient is less than 50 years old and undergoes surgical treatment, the confidence that the treatment effect is a significant recovery of joint function is higher. This is an association relationship, and finally the correlation between each femoral head necrosis characteristic factor and the treatment effect is obtained.

[0110] Preferably, based on the correlation between each characteristic factor of femoral head necrosis and the treatment effect, the imaging characteristics of femoral head necrosis tissue and each characteristic factor of femoral head necrosis within the basic characteristics of femoral head necrosis treatment are evaluated for correlation, and the characteristic correlation factor between each characteristic factor of femoral head necrosis and the treatment effect is obtained;

[0111] In an embodiment of the present invention, the correlation between each characteristic factor of femoral head necrosis and the treatment effect is evaluated by using the Pearson correlation coefficient method based on the previously obtained correlation relationship. For each characteristic factor of femoral head necrosis (such as the size of the necrotic area) and the treatment effect index (such as the degree of pain relief), the Pearson correlation coefficient between them is calculated. The calculation of the Pearson correlation coefficient is based on the mean, standard deviation and covariance of the two. If the calculated Pearson correlation coefficient of the size of the necrotic area and the degree of pain relief is -0.6, it means that the two are negatively correlated, that is, the larger the necrotic area, the lower the degree of pain relief. Such calculations are performed on all imaging features of femoral head necrosis tissue and characteristic factors within the basic treatment characteristics to obtain characteristic correlation factors between each characteristic factor of femoral head necrosis and the treatment effect. These factors quantify the degree of correlation between the characteristic factors and the treatment effect, and finally the characteristic correlation factors between each characteristic factor of femoral head necrosis and the treatment effect are obtained.

[0112] Preferably, causal inference of the treatment effect is performed on the imaging characteristics of femoral head necrosis tissue and each femoral head necrosis characteristic factor within the basic characteristics of femoral head necrosis treatment to obtain the number of potential causal relationships between each femoral head necrosis characteristic factor and the treatment effect;

[0113] In an embodiment of the present invention, by using a causal inference algorithm, such as the Granger causality test, the causal relationship between the imaging characteristics of femoral head necrosis tissue and each characteristic factor within the basic treatment characteristics and the treatment effect is inferred. Taking the Granger causality test as an example, the time series data of each characteristic factor (such as the treatment cycle) and the treatment effect (such as the recovery of joint function) are input into the test model. Through test analysis, if it is found that the change in the treatment cycle statistically precedes the change in the joint function recovery, and this change is significant, then it is considered that the treatment cycle has a potential causal relationship with the joint function recovery. Such causal inference is performed on all characteristic factors, and the specific number of potential causal relationships between each characteristic factor and the treatment effect is counted, and finally the number of potential causal relationships between each femoral head necrosis characteristic factor and the treatment effect is obtained.

[0114] Preferably, based on the characteristic correlation factors and the potential causal relationship number between each characteristic factor of femoral head necrosis and the treatment effect, the imaging characteristics of femoral head necrosis tissue and the basic characteristics of femoral head necrosis treatment are screened for the characteristics affecting the treatment effect, so as to obtain the significant characteristics affecting the treatment effect of femoral head necrosis;

[0115] In an embodiment of the present invention, by setting screening criteria, the characteristic correlation factors and potential causal relationship numbers between each characteristic factor of femoral head necrosis and the treatment effect are comprehensively considered. For the characteristic correlation factor, the larger the absolute value, the stronger the correlation; for the potential causal relationship number, the larger the number, the greater the possibility of the causal influence of the characteristic factor on the treatment effect. For example, the characteristic factor with an absolute value of the characteristic correlation factor greater than 0.5 and a potential causal relationship number greater than 2 is set as a significant influencing feature. According to this standard, the imaging features of femoral head necrosis tissue and all characteristic factors within the basic treatment features are screened. If the characteristic correlation factor of the necrotic area size is -0.7 and the potential causal relationship number is 3, which meets the screening criteria, the necrotic area size is determined as a significant feature affecting the treatment effect of femoral head necrosis. Through such screening, a group of features that have a significant impact on the treatment effect are obtained, and finally a significant feature affecting the treatment effect of femoral head necrosis is obtained.

[0116] Preferably, a convolutional neural network is used to construct a femoral head necrosis treatment effect prediction model, and the significant features affecting the femoral head necrosis treatment effect prediction model is input into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans.

[0117] In an embodiment of the present invention, a femoral head necrosis treatment effect prediction model is constructed by using Python programming language and deep learning framework (such as TensorFlow or PyTorch), and a convolutional neural network with a pyramid network architecture is built. In the shallow layer of the network, a 3x3 convolution kernel is used to perform a convolution operation on the input image data and treatment feature data to extract detailed features such as tiny trabecular changes and early necrotic area boundaries, and then the features are reduced in dimension through a 1x1 pooling window. In the deep network, a 5x5 convolution kernel is used to extract overall features such as joint morphology and macroscopic distribution of necrotic areas, and a 1x1 pooling window is also used for dimensionality reduction. When training the model, the convolution kernel is used. Energy constraints are introduced, the activation energy and information transfer energy of each neuron are calculated and included in the optimization objective function, and the prediction error is minimized through the optimizer (such as stochastic gradient descent). At the same time, the cross-validation method is used to divide the data set into multiple subsets, and one of the subsets is used as the validation set in turn, and the remaining subsets are used as the training set. The model is trained and verified multiple times, and the model parameters with the best performance are selected. The screened out significant feature data that affects the treatment effect of femoral head necrosis is input into the trained model. The model processes and predicts the input data at different network layers, and finally outputs the corresponding femoral head necrosis treatment effects under different treatment plans, such as the degree of pain relief, joint function recovery and other prediction results.

[0118] Furthermore, the femoral head necrosis treatment effect prediction model is specifically a pyramid network architecture that extracts treatment effect features corresponding to multiple scales. In the shallow network, the architecture uses a 3x3 corresponding convolution kernel and a 1x1 pooling window to extract the detailed features corresponding to the femoral head, including tiny trabecular changes and the boundaries of early necrotic areas, while in the deep network, the architecture uses a 5x5 corresponding convolution kernel and a 1x1 pooling window to extract the overall features corresponding to the femoral head, including joint morphology and the macroscopic distribution of necrotic areas. By introducing energy constraints, the activation energy and information transfer energy of each neuron in the network are included in the optimization target to focus on the prediction error corresponding to the model. At the same time, cross-validation is used to improve the corresponding prediction performance of the model, so as to predict the corresponding femoral head necrosis treatment effect at different network layers.

[0119] Furthermore, the present invention also provides a method for predicting and analyzing the therapeutic effect of femoral head necrosis based on big data, which is implemented based on the above-mentioned platform for predicting and analyzing the therapeutic effect of femoral head necrosis based on big data. The method for predicting and analyzing the therapeutic effect of femoral head necrosis based on big data includes:

[0120] Obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and treatment plans for femoral head necrosis. Perform data standardization cleaning and medical file integration on the multi-source medical data on the treatment of femoral head necrosis to generate a femoral head necrosis treatment data file.

[0121] Perform femoral head feature analysis on the corresponding femoral head medical images in the femoral head necrosis treatment data file to obtain the femoral head necrosis tissue image features, including femoral head necrosis tissue structure, femoral head necrosis tissue morphology, and femoral head necrosis tissue density;

[0122] Based on the corresponding femoral head clinical symptoms in the femoral head necrosis treatment data file, the corresponding femoral head necrosis treatment plan is divided into femoral head treatment stages to obtain the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage; based on the patient's basic information and femoral head necrosis tissue density, the femoral head necrosis treatment characteristics of the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage are analyzed to obtain the basic characteristics of femoral head necrosis treatment;

[0123] The imaging features of femoral head necrosis tissue and the basic characteristics of femoral head necrosis treatment are screened for features that affect the treatment effect, so as to obtain the significant features that affect the treatment effect of femoral head necrosis; a femoral head necrosis treatment effect prediction model is constructed, and the significant features that affect the treatment effect of femoral head necrosis are input into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis, so as to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans.

[0124] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0125] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A big data-based platform for predicting and analyzing the therapeutic effect of femoral head necrosis, characterized in that: Includes the following modules: A femoral head necrosis file generation module is used to obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and a treatment plan for femoral head necrosis, wherein the basic patient information includes the patient's corresponding age, gender, previous medical history of the femoral head, and medication history; the clinical symptoms of the femoral head include the degree of femoral head pain and the range of limited femoral head movement; the medical images of the femoral head include X-ray images, CT images, and MRI images corresponding to the femoral head; the treatment plan for femoral head necrosis includes treatment methods, drug dosages, and surgical records; and the multi-source medical data for the treatment of femoral head necrosis is cleaned according to data standards and medical file integration is performed to generate a femoral head necrosis treatment data file; The femoral head image feature analysis module is used to perform femoral head feature analysis on the corresponding femoral head medical images in the femoral head necrosis treatment data file to obtain the femoral head necrosis tissue image features, including the femoral head necrosis tissue structure, femoral head necrosis tissue morphology and femoral head necrosis tissue density; The femoral head treatment feature analysis module is used to perform femoral head treatment stage on the corresponding femoral head necrosis treatment plan based on the corresponding femoral head clinical symptoms in the femoral head necrosis treatment data file, so as to obtain the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage; based on the patient's basic information and femoral head necrosis tissue density, the femoral head necrosis treatment feature analysis is performed on the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage to obtain the basic characteristics of femoral head necrosis treatment; The femoral head treatment feature analysis module includes the following functions: Based on the corresponding femoral head pain level and femoral head motion limitation range in the femoral head necrosis treatment data file, the corresponding femoral head necrosis treatment plan is divided into femoral head treatment stages to obtain the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage; Based on the corresponding femoral head medical history and drug use history in the patient's basic information, the femoral head necrosis treatment and allergy probability analysis is performed on the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage to obtain the femoral head necrosis treatment probability and femoral head necrosis treatment allergy probability; Based on the density of femoral head necrotic tissue, the necrotic tissue attenuation evaluation and analysis were performed on the femoral head necrotic tissue treatment cases corresponding to each femoral head symptom stage to obtain the femoral head necrotic tissue treatment attenuation efficiency; The femoral head necrosis treatment probability, femoral head necrosis treatment allergy probability and femoral head necrosis tissue treatment attenuation efficiency are used as basic features to obtain the femoral head necrosis treatment basic features; The treatment effect prediction and analysis module is used to screen the treatment effect-influencing features of the femoral head necrosis tissue imaging features and the basic treatment features of the femoral head necrosis to obtain the significant features that affect the treatment effect of the femoral head necrosis; construct a femoral head necrosis treatment effect prediction model, and input the significant features that affect the treatment effect of the femoral head necrosis into the femoral head necrosis treatment effect prediction model to perform treatment effect prediction analysis to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans; The treatment effect prediction and analysis module includes the following functions: Conduct treatment effect correlation mining analysis on the imaging characteristics of femoral head necrosis tissue and the various femoral head necrosis characteristic factors within the basic characteristics of femoral head necrosis treatment to obtain the correlation between each femoral head necrosis characteristic factor and the treatment effect; Based on the correlation between each characteristic factor of femoral head necrosis and the treatment effect, the correlation between the imaging characteristics of femoral head necrosis tissue and each characteristic factor of femoral head necrosis within the basic characteristics of femoral head necrosis treatment was evaluated to obtain the characteristic correlation factor between each characteristic factor of femoral head necrosis and the treatment effect; Causal inference of the therapeutic effect was performed on the imaging characteristics of femoral head necrosis tissue and each characteristic factor of femoral head necrosis within the basic characteristics of femoral head necrosis treatment, and the number of potential causal relationships between each characteristic factor of femoral head necrosis and the therapeutic effect was obtained; Based on the characteristic correlation factors and potential causal relationship numbers between each characteristic factor of femoral head necrosis and the treatment effect, the imaging characteristics of femoral head necrosis tissue and the basic characteristics of femoral head necrosis treatment were screened for the characteristics affecting the treatment effect, so as to obtain the significant characteristics affecting the treatment effect of femoral head necrosis; A convolutional neural network was used to construct a femoral head necrosis treatment effect prediction model, and the significant features affecting the femoral head necrosis treatment effect were input into the femoral head necrosis treatment effect prediction model for treatment effect prediction analysis, so as to predict and output the corresponding femoral head necrosis treatment effects under different treatment plans; The femoral head necrosis treatment effect prediction model is specifically a pyramid network architecture that extracts treatment effect features corresponding to multiple scales. In the shallow network, the architecture uses a 3x3 corresponding convolution kernel and a 1x1 pooling window to extract the detailed features corresponding to the femoral head, including tiny trabecular changes and the boundaries of early necrotic areas. In the deep network, the architecture uses a 5x5 corresponding convolution kernel and a 1x1 pooling window to extract the overall features corresponding to the femoral head, including joint morphology and the macroscopic distribution of necrotic areas. By introducing energy constraints, the activation energy and information transfer energy of each neuron in the network are included in the optimization target to focus on the prediction error corresponding to the model. At the same time, cross-validation is used to improve the corresponding prediction performance of the model, so as to predict the corresponding femoral head necrosis treatment effect at different network layers.

2. The big data-based femoral head necrosis treatment effect prediction and analysis platform according to claim 1 is characterized in that: The femoral head necrosis file generation module includes the following functions: Obtain multi-source medical data on the treatment of femoral head necrosis, including basic patient information, clinical symptoms of the femoral head, medical images of the femoral head, and a treatment plan for femoral head necrosis, wherein the basic patient information includes the patient's age, gender, previous medical history of the femoral head, and medication history; the clinical symptoms of the femoral head include the degree of femoral head pain and the range of limited femoral head movement; the medical images of the femoral head include X-ray images, CT images, and MRI images corresponding to the femoral head; and the treatment plan for femoral head necrosis includes treatment methods, medication dosage, and surgical records; The multi-source medical data on the treatment of femoral head necrosis was cleaned to remove noise, duplicate data, and abnormal data. For image data, labeling errors were corrected and image coding was standardized using the unified medical image transmission standard DICOM. For text data, natural language processing was used to perform normalization and delete duplicate medical examination records to obtain multi-source standardized data on the treatment of femoral head necrosis. The multi-source data on the treatment of femoral head necrosis corresponding to the same patient in the multi-source standard data on the treatment of femoral head necrosis are correlated and matched and integrated to generate a data file for the treatment of femoral head necrosis.

3. The big data-based femoral head necrosis treatment effect prediction and analysis platform according to claim 1 is characterized in that: The femoral head image feature analysis module includes the following functions: An in-depth analysis of the physical characteristics of the femoral head medical images corresponding to the femoral head necrosis treatment data files was performed. For X-ray images, the attenuation law of X-rays when passing through the femoral head tissue was studied. The relationship between the grayscale value and tissue density in the image was analyzed based on the differences in the degree of X-ray absorption by different tissues. For CT images, the femoral head tissue characteristics reflected by the voxel values ​​at different levels were deeply analyzed. For MRI images, the hydrogen proton density and relaxation time corresponding to the magnetic resonance signal and the femoral head tissue were studied to obtain a femoral head medical image physical characteristics dataset. Perform femoral head necrotic tissue area segmentation on femoral head medical images to generate femoral head necrotic tissue area segmentation images; The multi-scale texture feature analysis of the femoral head necrotic tissue area segmentation images at different scales was performed based on wavelet transform to obtain the texture feature set of the femoral head necrotic tissue area. Based on the physical property dataset of femoral head medical images, the texture feature set of the femoral head necrotic tissue region was analyzed to invert the tissue structure corresponding to the femoral head necrotic tissue according to the corresponding grayscale and signal intensity in the image. The density distribution corresponding to the femoral head necrotic tissue was calculated by combining the known relationship between the X-ray attenuation coefficient and tissue density with the image voxel value. At the same time, the morphological characteristics corresponding to the femoral head necrotic tissue were inverted according to the relationship between T1 and T2 relaxation times and tissue morphology to obtain the image characteristics of the femoral head necrotic tissue, including the femoral head necrotic tissue structure, femoral head necrotic tissue morphology and femoral head necrotic tissue density.

4. The big data-based femoral head necrosis treatment effect prediction and analysis platform according to claim 3 is characterized in that: The multi-scale texture feature analysis of the femoral head necrotic tissue area segmentation image at different scales based on wavelet transform is specifically to extract the subtle texture features corresponding to the femoral head tissue at a small scale, including the microstructural texture corresponding to the trabeculae, by using the high-frequency sub-band coefficients corresponding to the wavelet transform, and to calculate and analyze the texture roughness and self-similarity corresponding to the entire femoral head at a large scale using the fractal dimension.

5. The big data-based femoral head necrosis treatment effect prediction and analysis platform according to claim 1 is characterized in that: The femoral head symptom staging is specifically as follows: the early stage is when the femoral head pain level is between 1-3 points and the range of restricted femoral head motion is between 10%-20%, including the hip flexion angle decreasing from the normal 120°-140° to 100°-120°, the abduction angle decreasing from the normal 30°-45° to 25°-35°, and the internal rotation angle decreasing from the normal 30°-40° to 20°-30°; the middle stage is when the femoral head pain level is between 4-6 points and the range of restricted femoral head motion expands to 20%-40%, including the hip flexion angle decreasing to 80°-100°, the abduction angle decreasing to 20°-25°, and the internal rotation angle decreasing to 10°-20°; the late stage is when the femoral head pain level is between 7-10 points and the range of restricted femoral head motion expands by more than 40%, including the hip flexion angle less than 80°, the abduction angle less than 20°, and the internal rotation angle less than 10°.

6. The big data-based femoral head necrosis treatment effect prediction and analysis platform according to claim 1 is characterized in that: The necrotic tissue attenuation evaluation and analysis of the femoral head necrosis treatment sub-cases corresponding to each femoral head symptom stage based on the femoral head necrosis tissue density includes: Based on the corresponding femoral head necrosis treatment sub-cases under each femoral head symptom stage, the tissue density treatment simulation of the corresponding femoral head necrosis tissue density is performed to generate the density attenuation process of the femoral head necrosis tissue under each treatment effect; The actual attenuation amount of the density attenuation process of the femoral head necrosis tissue under each treatment effect is statistically analyzed to obtain the actual attenuation amount of the tissue density under each treatment effect; The theoretical maximum attenuation of femoral head necrotic tissue is obtained, and based on the theoretical maximum attenuation of femoral head necrotic tissue, the actual attenuation of tissue density corresponding to femoral head necrotic tissue under various treatment effects is quantitatively calculated to obtain the treatment attenuation efficiency of femoral head necrotic tissue.

Images