A medical diagnosis method and system based on big data analysis

Abstract

The present invention discloses a medical diagnosis method and system based on big data analysis, including: the present invention maps skin lesion features to a high-dimensional kernel space through nonlinear mapping and kernel principal component analysis methods, effectively solving the problem of processing high-dimensional features; using new features obtained by kernel principal component analysis to replace channel weights in a convolutional neural network, it is possible to dynamically adjust the weights of different spatial positions in the feature map, enhance attention to data utility areas, and improve the accuracy of feature extraction; in addition, our invention further improves the accuracy and robustness of medical diagnosis by splicing abnormal temperature areas of a big data analysis model with new features, making full use of the complementarity of multimodal data, and also providing an efficient and reliable solution for medical diagnosis of other diseases.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning and data analysis, and in particular to a medical diagnosis method and system based on big data analysis. Background Art

[0002] With the rapid development of computer vision and machine learning technologies, the field of medical diagnosis is also constantly exploring and applying new technologies to improve the accuracy and efficiency of disease detection. As an important research direction, skin lesion detection and diagnosis has made significant progress in recent years. Traditional skin lesion diagnosis mainly relies on doctors' visual inspection and experience. However, this method has certain subjectivity and limitations. In order to solve these problems, researchers have proposed automatic diagnosis methods based on image processing and machine learning. In particular, the successful application of convolutional neural networks (CNN) in image classification and segmentation tasks has made skin lesion diagnosis methods based on deep learning a research hotspot.

[0003] Gray-Level Co-occurrence Matrix (GLCM) is a classic texture analysis method and is widely used in texture feature extraction of skin lesion areas. GLCM can capture the spatial relationship between the grayscale values ​​of pixels in an image, thereby extracting the texture features of the image. In addition, thermal imaging technology has also been introduced into skin lesion detection. By analyzing the temperature distribution of different parts of the body surface, abnormal temperature areas can be identified to provide auxiliary diagnostic information.

[0004] Although the existing technology has made some significant progress in skin lesion detection and diagnosis, there are still some shortcomings. First, the traditional GLCM-based texture feature extraction method is easily disturbed by image noise and artifacts when facing complex skin lesion images, resulting in a decrease in the accuracy of feature extraction. In addition, the existing thermal imaging analysis method is limited by simple growth conditions during the region growing process, which easily leads to over-segmentation or under-segmentation, affecting the accuracy of temperature data. More importantly, the existing model needs to be trained in advance and the data utility features need to be extracted to ensure the diagnostic accuracy of the model. It can be seen that there are certain limitations. Summary of the invention

[0005] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0006] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a medical diagnosis method based on big data analysis to solve the problems mentioned in the background technology.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a medical diagnosis method based on big data analysis, comprising:

[0009] Collect thermal imaging images of the patient's skin lesion area and body surface, extract texture features of the skin lesion area through a gray-level co-occurrence matrix, and perform region growth on the thermal imaging to obtain temperature data of different parts of the body surface;

[0010] Establishing a big data analysis model, analyzing the texture features of the skin lesion area to obtain different types of skin lesion features, analyzing the temperature data to identify abnormal temperature areas;

[0011] Performing nonlinear mapping on the skin lesion features, mapping the skin lesion features to a high-dimensional kernel space using a kernel principal component analysis method, performing principal component analysis in the high-dimensional kernel space, selecting key components obtained by the analysis as new features, and using the new features to replace channel weights in a convolutional neural network;

[0012] On the basis of the convolutional neural network, a spatial attention mechanism is added to splice the abnormal temperature areas of the big data analysis model with the new features to output the medical diagnosis results.

[0013] As a preferred solution of the medical diagnosis method based on big data analysis described in the present invention, the texture features of the skin lesion area are extracted by using the gray level co-occurrence matrix, including:

[0014] The skin lesion area image is converted into a grayscale image, the grayscale image is segmented, a grayscale co-occurrence matrix is ​​generated, the grayscale co-occurrence matrix is ​​initialized, each pixel point in the lesion area is traversed, and grayscale value pairs of adjacent pixels are calculated according to the selected direction and spacing, and each calculated grayscale value pair is stored to obtain each element in the grayscale co-occurrence matrix;

[0015] Among them, each element represents the probability of occurrence of a gray value pair.

[0016] As a preferred solution of the medical diagnosis method based on big data analysis described in the present invention, the thermal imaging is subjected to regional growth to obtain temperature data of different parts of the body surface, including:

[0017] The thermal image is converted into a grayscale image, and the seed point is marked based on the boundary between the light and dark grayscale.

[0018] Starting from the marked seed point, pixels adjacent to the seed point and satisfying the growth condition are added to the region;

[0019] The growth condition, i.e., the standard deviation between pixels;

[0020] The newly added pixel is used as the region boundary point to determine whether its adjacent pixels also meet the growth conditions;

[0021] If the conditions are met, it is also added to the region, otherwise the regional growth ends;

[0022] Traverse each pixel in the segmented area, read the corresponding temperature value, calculate the average temperature, maximum temperature and temperature distribution characteristics of each area, and obtain the temperature data of different parts of the body surface.

[0023] As a preferred solution of the medical diagnosis method based on big data analysis described in the present invention, wherein: a big data analysis model is established, the texture features of the skin lesion area are analyzed to obtain different types of skin lesion features, the temperature data is analyzed, and the abnormal temperature area is identified, including:

[0024] Creating classification training labels, the big data analysis model is trained by a random forest classifier;

[0025] Through the random forest classifier, several decision trees are created. Each decision tree takes a part of the texture features and temperature data and can be repeated to construct a sample set.

[0026] The sample set is trained using the classification training labels to output different types of skin lesion features and abnormal temperature areas.

[0027] As a preferred solution of the medical diagnosis method based on big data analysis described in the present invention, wherein: nonlinear mapping is performed on the skin lesion characteristics, and the skin lesion characteristics are mapped to a high-dimensional kernel space using a kernel principal component analysis method, including:

[0028] Select a Gaussian kernel function to map the texture features of the skin lesion area to a high-dimensional space and calculate the kernel matrix of the skin lesion feature data;

[0029] The kernel matrix is ​​centralized so that the mean of the eigenvalues ​​in the high-dimensional space is 0.

[0030] As a preferred solution of the medical diagnosis method based on big data analysis described in the present invention, principal component analysis is performed in the high-dimensional kernel space, and key components obtained by the analysis are selected as new features, including:

[0031] Perform eigenvalue decomposition on the centralized kernel matrix to obtain the eigenvalues ​​and corresponding eigenvectors;

[0032] The first m principal components are selected according to the size of the eigenvalue, and the texture features of the original skin lesion area are projected through the screened m principal components to obtain the texture features of the new skin lesion area, and the texture features of the new skin lesion area are the principal components in the high-dimensional space.

[0033] As a preferred solution of the medical diagnosis method based on big data analysis described in the present invention, a spatial attention mechanism is added on the basis of the convolutional neural network, and the abnormal temperature area of ​​the big data analysis model is spliced ​​with the new features, including:

[0034] Perform multi-scale global average pooling on the feature map output by the convolutional layer to generate global average pooling feature maps of different scales;

[0035] Splicing the multi-scale global average pooling feature maps on the channel weights to form a spliced ​​feature map;

[0036] Perform a 1×1 convolution operation on the concatenated feature map to compress the number of channels and generate a single-channel attention feature map;

[0037] Perform Sigmoid activation on the attention feature map of a single channel to obtain the attention weight of each spatial position;

[0038] Multiply the attention weights by the original convolutional feature map element-wise to generate a weighted feature map;

[0039] The abnormal temperature region is copied and expanded to have the same number of channels as the attention feature map, and the expanded abnormal temperature region and the weight-adjusted feature map are concatenated twice on the channel weights to form a new feature map, and the threshold of the abnormal temperature region is output;

[0040] Perform a 3×3 convolution operation on the spliced ​​feature map to compress the number of channels and perform ReLU activation to generate a fused spliced ​​feature map;

[0041] The fused feature map is processed by a multi-layer perceptron, including a single fully connected layer, using Softmax activation to output the probabilities of different types of skin lesions;

[0042] The obtained skin lesion probability and the threshold of the abnormal temperature area are used as input and applied to the smart medical diagnosis platform or the distributed cloud medical diagnosis platform to obtain the medical diagnosis results.

[0043] In a second aspect, the present invention provides a medical diagnosis system based on big data analysis, comprising:

[0044] The patient image acquisition module is used to acquire thermal imaging images of the patient's skin lesion area and body surface, extract the texture features of the skin lesion area through the gray-level co-occurrence matrix, and perform region growth on the thermal imaging to obtain temperature data of different parts of the body surface;

[0045] A big data analysis module is used to establish a big data analysis model, analyze the texture features of the skin lesion area, obtain different types of skin lesion features, analyze the temperature data, and identify abnormal temperature areas;

[0046] A kernel principal component analysis module, used to perform nonlinear mapping on the skin lesion features, map the skin lesion features to a high-dimensional kernel space using a kernel principal component analysis method, perform principal component analysis in the high-dimensional kernel space, select key components obtained by the analysis as new features, and use the new features to replace channel weights in a convolutional neural network;

[0047] The convolutional neural network processing module is used to add a spatial attention mechanism based on the convolutional neural network, splice the abnormal temperature areas of the big data analysis model with the new features, and output the medical diagnosis results.

[0048] In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: the processor implements any step of the above method when executing the computer program.

[0049] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: the computer program implements any step of the above method when executed by a processor.

[0050] Compared with the prior art, the beneficial effects of the invention are as follows: the present invention maps the skin lesion characteristics to a high-dimensional kernel space through nonlinear mapping and kernel principal component analysis methods, thereby effectively solving the problem of processing high-dimensional features; secondly, by using the new features obtained by kernel principal component analysis to replace the channel weights in the convolutional neural network, the weights of different spatial positions in the feature map can be dynamically adjusted, the focus on the data utility area can be enhanced, and the accuracy of feature extraction can be improved; in addition, our invention further improves the accuracy and robustness of medical diagnosis by splicing the abnormal temperature area of ​​the big data analysis model with the new features, making full use of the complementarity of multimodal data, and also providing an efficient and reliable solution for the medical diagnosis of other diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0052] Figure 1 This is an overall flow chart of a medical diagnosis method based on big data analysis according to an embodiment of the present invention;

[0053] Figure 2 A comparison chart of diagnostic accuracy of different classifiers of a medical diagnosis method based on big data analysis according to an embodiment of the present invention;

[0054] Figure 3 A comparison chart of data utility attention of a medical diagnosis method based on big data analysis according to an embodiment of the present invention;

[0055] Figure 4 This is a comparison chart of the robustness of a medical diagnosis method based on big data analysis according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0057] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0058] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0059] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0060] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0061] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0062] Example 1

[0063] Reference Figure 1 , which is the first embodiment of the present invention, and provides a medical diagnosis method based on big data analysis, comprising:

[0064] S1. Collect thermal imaging images of the patient's skin lesion area and body surface, extract the texture features of the skin lesion area through the gray-level co-occurrence matrix, and perform region growth on the thermal imaging to obtain temperature data of different parts of the body surface;

[0065] Further, the skin lesion area image is converted into a grayscale image, the grayscale image is segmented, a grayscale co-occurrence matrix is ​​generated, the grayscale co-occurrence matrix is ​​initialized, each pixel in the lesion area is traversed, and grayscale value pairs of adjacent pixels are calculated according to the selected direction and spacing, and each calculated grayscale value pair is stored to obtain each element in the grayscale co-occurrence matrix;

[0066] Among them, each element represents the probability of occurrence of a gray value pair;

[0067] Specifically, the gray-level co-occurrence matrix can capture the texture features in the image and generate a matrix containing the probability of occurrence of different gray-level value pairs by analyzing the gray-level relationship between adjacent pixels;

[0068] It should be noted that the texture features of the extracted skin lesion area can be used to train the channel weights in the convolutional neural network to achieve a dynamic adjustment effect, so as to enhance the ability to distinguish different types of skin lesions;

[0069] Furthermore, the thermal image is converted into a grayscale image, and the seed point is marked based on the boundary between the light and dark grayscale.

[0070] Furthermore, starting from the marked seed point, pixels adjacent to the seed point and satisfying the growth condition are added to the region;

[0071] Among them, growth condition, i.e., standard deviation between pixels;

[0072] Furthermore, the newly added pixel is used as the region boundary point to determine whether its adjacent pixels also meet the growth conditions;

[0073] Furthermore, if the conditions are met, it is also added to the region, otherwise the region growth ends;

[0074] Furthermore, each pixel in the segmented area is traversed, the corresponding temperature value is read, the average temperature, maximum temperature and temperature distribution characteristics of each area are calculated, and the temperature data of different parts of the body surface are obtained;

[0075] Specifically, texture features reflect the specific morphology of skin lesions, while temperature data from different parts of the body surface provide the physiological status of the lesion area;

[0076] It should be noted that after the thermal image is converted into a grayscale image, the seed point is marked at the junction of the deep and shallow grayscale, and then the region is grown to accurately divide the different temperature areas on the body surface. This provides high-purity temperature data for discovering potential abnormalities in skin lesion areas, so that the big data analysis model only needs to read the temperature data for analysis, and no pre-processing of the temperature data is required;

[0077] S2. Establish a big data analysis model to analyze the texture features of the skin lesion area, obtain the characteristics of different types of skin lesions, analyze the temperature data, and identify abnormal temperature areas;

[0078] Further, create classification training labels;

[0079] Specifically, the classification training labels provide the model with learning signals that can identify texture features and temperature data during processing;

[0080] Among them, the big data analysis model is trained by random forest classifier;

[0081] Preferably, the random forest classifier can process high-dimensional and complex data compared to other classifiers. In practical applications, the texture features of skin lesion areas and the temperature data of different parts of the body surface are both high-dimensional and complex data, and can improve the generalization ability and robustness of the model;

[0082] Furthermore, a random forest classifier is used to create several decision trees, each of which takes a portion of the texture features and temperature data and can be repeated to construct a sample set;

[0083] Specifically, each decision tree is a distributed combination, and each decision tree can be taken out separately for training data;

[0084] Furthermore, the sample set is trained using classification training labels to output different types of skin lesion features and abnormal temperature areas;

[0085] S3, nonlinearly mapping the skin lesion features, mapping the skin lesion features to a high-dimensional kernel space using a kernel principal component analysis method, performing principal component analysis in the high-dimensional kernel space, selecting the key components obtained by the analysis as new features, and using the new features to replace the channel weights in the convolutional neural network;

[0086] Furthermore, a Gaussian kernel function is selected to map the texture features of the skin lesion area into a high-dimensional space, and a kernel matrix of the skin lesion feature data is calculated;

[0087] Specifically, let X = {x1, x2, x3, …, x i} is the texture feature data set of the skin lesion area, each x i , represents a texture feature vector, and the selected Gaussian kernel function is:

[0088]

[0089] Among them, x i is the i-th texture feature vector, Ω is the range interval of [i, j]; σ is the bandwidth parameter of the Gaussian kernel function; k(x i ,x j ) is represented as a Gaussian function, which represents the i-th texture feature vector x in the skin lesion area calculated by the Gaussian kernel function. i and the jth texture feature vector x j The similarity between

[0090] Specifically, the element K in the kernel matrix K ij , obtained by the Gaussian kernel function:

[0091]

[0092] Among them, ||xi -x j || represents the Euclidean distance between the i-th and j-th texture feature vectors;

[0093] Furthermore, the kernel matrix is ​​centered so that the mean of the eigenvalues ​​in the high-dimensional space is 0;

[0094] Specifically, let 1 be the all-one vector matrix and H be the centering matrix, defined as:

[0095]

[0096] Where I is the identity matrix; T represents the transpose of the matrix;

[0097] Specifically, the centralized kernel matrix K c The calculation is as follows:

[0098]

[0099] Furthermore, the eigenvalue decomposition is performed on the centralized kernel matrix to obtain the eigenvalues ​​and corresponding eigenvectors;

[0100] Further, the first m principal components are selected according to the size of the eigenvalues, and the texture features of the original skin lesion area are projected through the screened m principal components to obtain the texture features of the new skin lesion area, and the texture features of the new skin lesion area are the principal components in the high-dimensional space;

[0101] Specifically, for K c Perform eigenvalue decomposition and get:

[0102] K c =VΛV T

[0103] Where V is the eigenvector matrix; Λ is the eigenvalue diagonal matrix;

[0104] Specifically, the original texture features are projected into the new principal component space to obtain the new texture features of the skin lesion area:

[0105]

[0106] Among them, k i is the centralized kernel matrix K c The i-th column, z i is the principal component eigenvector in the new high-dimensional space;

[0107] Specifically, the new feature vector captures the key variation information in the original feature data in high-dimensional space, and the principal component feature vector is used for the channel weight of the convolutional neural network;

[0108] It should be noted that replacing the channel weights in the convolutional neural network with the new features obtained by kernel principal component analysis can dynamically adjust the weights of different spatial positions in the feature map, enhance the focus on the data utility area, and improve the accuracy of feature extraction;

[0109] S4. Add a spatial attention mechanism based on the convolutional neural network to splice the abnormal temperature area of ​​the big data analysis model with the new features to output the medical diagnosis results;

[0110] Furthermore, multi-scale global average pooling is performed on the feature map output by the convolutional layer to generate global average pooling feature maps of different scales;

[0111] Specifically, let the feature map output by the convolutional layer be F, whose dimension is H×W×C, where H is the height, W is the width, and C is the number of channels. Multi-scale global average pooling is performed on the feature map to generate global average pooling feature maps of different scales, and we get:

[0112]

[0113] Among them, s represents the scale; H s and W s are the height and width at this scale respectively; Represents global average pooling feature maps of different scales;

[0114] Furthermore, the multi-scale global average pooling feature maps are concatenated on the channel weights to form a concatenated feature map, which is expressed as:

[0115]

[0116] Among them, α s is the weight parameter of each weight scale, S represents the total number of scales; F concat It is represented as the concatenation result of multi-scale global average pooling feature maps;

[0117] Furthermore, a 1×1 convolution operation is performed on the concatenated feature map to compress the number of channels and generate a single-channel attention feature map, which is expressed as:

[0118]

[0119] in, It is represented by the weight of 1×1 convolution kernel, σ is the Sigmoid activation function; F att Represents the attention feature map after Sigmoid activation;

[0120] Furthermore, the attention feature map of a single channel is Sigmoid activated to obtain the attention weight of each spatial position, which is expressed as:

[0121] F weighted =F⊙F att

[0122] Among them, ⊙ represents the element-wise multiplication operation; F weighted Represented as the result of element-wise multiplication with the original feature map;

[0123] Furthermore, the attention weights are element-wise multiplied with the original convolutional feature map to generate a weighted feature map;

[0124] Furthermore, the abnormal temperature region is copied and expanded to have the same number of channels as the attention feature map, and the expanded abnormal temperature region and the feature map after weight adjustment are concatenated twice on the channel weight to form a new feature map, and the threshold of the abnormal temperature region is output;

[0125] Specifically, the secondary splicing process is expressed as:

[0126]

[0127] F final =[F weighted ∥T expanded ]

[0128] in, represents the Kronecker product of tensors, ∥ represents the concatenation operation, T represents the wield abnormal temperature area; T expanded Represents the abnormal temperature area after replication and expansion; F final Represented as a concatenated feature map;

[0129] It should be noted that the purpose of replicating the abnormal temperature region is to ensure that the dimensions of all feature maps remain consistent during the secondary stitching process;

[0130] Furthermore, a 3×3 convolution operation is performed on the spliced ​​feature map to compress the number of channels and perform ReLU activation to generate a fused spliced ​​feature map, which is expressed as:

[0131]

[0132] Among them, F fused Represents the fused and spliced ​​feature map;

[0133] Furthermore, the fused feature map is processed by a multi-layer perceptron, including a single fully connected layer, using Softmax activation to output the probability P of different types of skin lesions, expressed as:

[0134] P = Softmax(W fc F fused )

[0135] Among them, W fc is the weight matrix of the fully connected layer;

[0136] Furthermore, the obtained skin lesion probability and the threshold value of the abnormal temperature area are used as input and applied to a smart medical diagnosis platform or a distributed cloud medical diagnosis platform to obtain a medical diagnosis result.

[0137] Furthermore, this embodiment also provides a medical diagnosis system based on big data analysis, including:

[0138] The patient image acquisition module is used to acquire thermal imaging images of the patient's skin lesion area and body surface, extract the texture features of the skin lesion area through the gray-level co-occurrence matrix, and perform region growth on the thermal imaging to obtain temperature data of different parts of the body surface;

[0139] A big data analysis module is used to establish a big data analysis model, analyze the texture features of the skin lesion area, obtain different types of skin lesion features, analyze the temperature data, and identify abnormal temperature areas;

[0140] A kernel principal component analysis module, used to perform nonlinear mapping on the skin lesion features, map the skin lesion features to a high-dimensional kernel space using a kernel principal component analysis method, perform principal component analysis in the high-dimensional kernel space, select key components obtained by the analysis as new features, and use the new features to replace channel weights in a convolutional neural network;

[0141] The convolutional neural network processing module is used to add a spatial attention mechanism based on the convolutional neural network, splice the abnormal temperature areas of the big data analysis model with the new features, and output the medical diagnosis results.

[0142] This embodiment also provides a computer device, which is applicable to the medical diagnosis method based on big data analysis, including:

[0143] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the medical diagnosis method based on big data analysis as proposed in the above embodiment.

[0144] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0145] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the medical diagnosis method based on big data analysis proposed in the above embodiment is implemented.

[0146] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0147] Example 2

[0148] Reference Figures 2 to 4 , which is the second embodiment of the present invention, and which provides a medical diagnosis method based on big data analysis, comprising: verifying the beneficial effects of the solution of the present invention by using MATLAB software;

[0149] Select the acquisition equipment of the patient's skin lesion area and body surface thermal imaging image (using Canon EOS R5 camera, resolution set to 45MP, ISO set to 100, shutter speed set to 1 / 100 second, standard lens; infrared thermal imaging device FLIR T660, resolution set to 640×480 pixels, temperature range of [-40℃, +150℃], accuracy of ±0.1℃); carry out in a standardized laboratory environment, room temperature controlled at 40℃±1℃, relative humidity controlled between 40% and 60%, image acquisition after the patient's skin is clean and dry, during the acquisition process, the patient remains still, 0.5 meters away from the camera, and the infrared thermal imaging device is 0.5 meters away from the skin; 5 skin lesion area images and 3 body surface thermal imaging images at different angles are collected for each patient; the traditional diagnostic method is: experienced dermatologists perform visual inspection and diagnosis, record the diagnostic results and opinions, and perform skin biopsy on some patients, observe tissue sections under a microscope, and record pathological diagnosis results; finally, the results of the traditional diagnostic method and the data obtained by the results of the diagnostic method of the present invention are applied together to the intelligent medical diagnosis platform IBM Watson Health;

[0150] In order to verify the superiority of the present invention, especially the preference of random forest classifier (RF) in big data analysis model, this experiment also compared the performance of other commonly used classifiers (support vector machine SVM, k-nearest neighbor algorithm KNN and decision tree DT), and obtained Table 1;

[0151] Table 1

[0152]

[0153] Table 1 shows that in the first 10 experiments, the diagnostic accuracy of the present invention can always maintain a high accuracy rate; however, this does not mean that there will be no decline or deviation in subsequent experiments. Therefore, the experimental data was increased, and skin lesion area images and body surface thermal imaging images of 100 patients were collected, and the results were as follows: Figure 2 As shown; it can be seen in Figure 2 The average diagnostic accuracy of the present invention is about 95%, while the diagnostic accuracy of the SVM classifier closest to the present invention is about 90%, which is about 5% higher.

[0154] Secondly, through Figure 3 It can be seen that after the present invention uses the new features to replace the channel weights in the convolutional neural network, the attention to the data utility in the convolutional neural network increases significantly (median Q1 and quartile Q3), and the accuracy of feature extraction can be achieved without the need to train and extract the data utility features in advance;

[0155] Finally, through Figure 4, the abnormal temperature areas of the big data analysis model are spliced ​​with new features, further improving the robustness of medical diagnosis.

[0156] Those skilled in the art will appreciate that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program codes. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0157] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0158] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0159] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0160] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0161] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A medical diagnosis method based on big data analysis, characterized in that: include: Collect thermal imaging images of the patient's skin lesion area and body surface, extract texture features of the skin lesion area through a gray-level co-occurrence matrix, and perform region growth on the thermal imaging to obtain temperature data of different parts of the body surface; The thermal imaging is subjected to region growth to obtain temperature data of different parts of the body surface, including: The thermal image is converted into a grayscale image, and the seed point is marked based on the boundary between the light and dark grayscale. Starting from the marked seed point, pixels adjacent to the seed point and satisfying the growth condition are added to the region; The growth condition, i.e., the standard deviation between pixels; The newly added pixel is used as the region boundary point to determine whether its adjacent pixels also meet the growth conditions; If the conditions are met, it is also added to the region, otherwise the regional growth ends; Traverse each pixel in the segmented area, read the corresponding temperature value, calculate the average temperature, maximum temperature and temperature distribution characteristics of each area, and obtain the temperature data of different parts of the body surface; Establishing a big data analysis model, analyzing the texture features of the skin lesion area to obtain different types of skin lesion features, analyzing the temperature data to identify abnormal temperature areas; The step of establishing a big data analysis model, analyzing the texture features of the skin lesion area to obtain different types of skin lesion features, and analyzing the temperature data to identify abnormal temperature areas includes: Creating classification training labels, the big data analysis model is trained by a random forest classifier; Through the random forest classifier, several decision trees are created. Each decision tree takes a part of the texture features and temperature data and can be repeated to construct a sample set. Using the classification training labels to train the sample set, and outputting different types of skin lesion features and abnormal temperature areas; Performing nonlinear mapping on the skin lesion features, mapping the skin lesion features to a high-dimensional kernel space using a kernel principal component analysis method, performing principal component analysis in the high-dimensional kernel space, selecting key components obtained by the analysis as new features, and using the new features to replace channel weights in a convolutional neural network; Add a spatial attention mechanism based on the convolutional neural network, combine the abnormal temperature area of ​​the big data analysis model with the new features, and output the medical diagnosis results; The spatial attention mechanism is added on the basis of the convolutional neural network to splice the abnormal temperature area of ​​the big data analysis model with the new features, including: Perform multi-scale global average pooling on the feature map output by the convolutional layer to generate global average pooling feature maps of different scales; Splicing the multi-scale global average pooling feature maps on the channel weights to form a spliced ​​feature map; Perform a 1×1 convolution operation on the concatenated feature map to compress the number of channels and generate a single-channel attention feature map; Perform Sigmoid activation on the attention feature map of a single channel to obtain the attention weight of each spatial position; Multiply the attention weights by the original convolutional feature map element-wise to generate a weighted feature map; The abnormal temperature region is copied and expanded to have the same number of channels as the attention feature map, and the expanded abnormal temperature region and the weight-adjusted feature map are concatenated twice on the channel weights to form a new feature map, and the threshold of the abnormal temperature region is output; Perform a 3×3 convolution operation on the spliced ​​feature map to compress the number of channels and perform ReLU activation to generate a fused spliced ​​feature map; The fused feature map is processed by a multi-layer perceptron, including a single fully connected layer, using Softmax activation to output the probabilities of different types of skin lesions; The obtained skin lesion probability and the threshold of the abnormal temperature area are used as input and applied to the smart medical diagnosis platform or the distributed cloud medical diagnosis platform to obtain the medical diagnosis results.

2. The medical diagnosis method based on big data analysis according to claim 1, characterized in that: The texture features of the skin lesion area are extracted through the gray level co-occurrence matrix, including: The skin lesion area image is converted into a grayscale image, the grayscale image is segmented, a grayscale co-occurrence matrix is ​​generated, the grayscale co-occurrence matrix is ​​initialized, each pixel point in the lesion area is traversed, and grayscale value pairs of adjacent pixels are calculated according to the selected direction and spacing, and each calculated grayscale value pair is stored to obtain each element in the grayscale co-occurrence matrix; Among them, each element represents the probability of occurrence of a gray value pair.

3. The medical diagnosis method based on big data analysis according to claim 1, characterized in that: The skin lesion features are nonlinearly mapped, and the skin lesion features are mapped to a high-dimensional kernel space using a kernel principal component analysis method, including: Select a Gaussian kernel function to map the texture features of the skin lesion area to a high-dimensional space and calculate the kernel matrix of the skin lesion feature data; The kernel matrix is ​​centralized so that the mean of the eigenvalues ​​in the high-dimensional space is 0.

4. The medical diagnosis method based on big data analysis according to claim 3, characterized in that: Performing principal component analysis in the high-dimensional kernel space and selecting key components obtained from the analysis as new features include: Perform eigenvalue decomposition on the centralized kernel matrix to obtain the eigenvalues ​​and corresponding eigenvectors; The first m principal components are selected according to the size of the eigenvalue, and the texture features of the original skin lesion area are projected through the screened m principal components to obtain the texture features of the new skin lesion area, and the texture features of the new skin lesion area are the principal components in the high-dimensional space.

5. A medical diagnosis system based on big data analysis, based on the medical diagnosis method based on big data analysis according to any one of claims 1 to 4, characterized in that: include: The patient image acquisition module is used to acquire thermal imaging images of the patient's skin lesion area and body surface, extract the texture features of the skin lesion area through the gray-level co-occurrence matrix, and perform region growth on the thermal imaging to obtain temperature data of different parts of the body surface; A big data analysis module is used to establish a big data analysis model, analyze the texture features of the skin lesion area, obtain different types of skin lesion features, analyze the temperature data, and identify abnormal temperature areas; A kernel principal component analysis module, used to perform nonlinear mapping on the skin lesion features, map the skin lesion features to a high-dimensional kernel space using a kernel principal component analysis method, perform principal component analysis in the high-dimensional kernel space, select key components obtained by the analysis as new features, and use the new features to replace channel weights in a convolutional neural network; The convolutional neural network processing module is used to add a spatial attention mechanism based on the convolutional neural network, splice the abnormal temperature areas of the big data analysis model with the new features, and output the medical diagnosis results.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images