Prosthetic socket and its adaptive adjustment method

Abstract

This application relates to the field of intelligent adjustment technology. Specifically, it discloses a prosthetic socket and its adaptive adjustment method. It uses artificial intelligence technology based on deep learning to analyze the pressure values between the residual limb and the socket collected by multiple pressure sensors, captures the pressure correlation pattern between the position of the residual limb and the socket, and introduces the spatial topological features between each pressure sensor to strengthen the feature expression of the pressure correlation pattern between the position of the residual limb and the socket, so as to intelligently evaluate the matching degree between the residual limb and the socket, and then realize the adaptive adjustment of the prosthetic socket. In this way, it can make the prosthetic socket more suitable for the residual limb, improve the comfort and stability of the prosthesis, reduce the pain and injury of the residual limb, and thus improve the quality of life of amputees.

Description

Technical Field

[0001] The present application relates to the field of intelligent adjustment technology, and more specifically, to a prosthetic socket and its adaptive adjustment method. Background Art

[0002] The prosthetic socket is an essential auxiliary device for many amputees. It directly contacts the residual limb, providing a stable and comfortable environment for the residual limb, supporting the body weight, controlling the movement of the prosthesis, and enabling the amputee to walk, stand, and perform daily activities.

[0003] A reliable connection between the residual limb and the socket ensures the realization of the prosthesis function. Specifically, any minor movement between the residual limb and the prosthetic socket will impede the amputee's precise control of the prosthesis, reducing the stability of standing and walking and increasing the patient's energy consumption. However, currently available prosthetic sockets on the market mainly rely on manual adjustment, which is both time-consuming and laborious. Moreover, since the shapes and sizes of the residual limbs of each amputee are different, and the shape of the residual limb changes due to daily activities, this manual adjustment method cannot adapt to the shape changes of the residual limb in a timely manner. For amputees, this undoubtedly increases the difficulty of using the prosthesis and affects their quality of life.

[0004] Therefore, there is a need for a prosthetic socket and its adaptive adjustment method. Summary of the Invention

[0005] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a prosthetic socket and its adaptive adjustment method, which use artificial intelligence technology based on deep learning to analyze the pressure values between the residual limb and the socket collected by multiple pressure sensors, capture the pressure correlation pattern between the positions of the residual limb and the socket, and introduce the spatial topological features between the pressure sensors to strengthen the feature expression of the pressure correlation pattern between the positions of the residual limb and the socket, thereby intelligently evaluating the matching degree between the residual limb and the socket, and further realizing the adaptive adjustment of the prosthetic socket. In this way, the prosthetic socket can be more adapted to the residual limb, improving the comfort and stability of the prosthesis, reducing the pain and injury of the residual limb, and thus enhancing the quality of life of the amputee.

[0006] Correspondingly, according to one aspect of the present application, there is provided an adaptive adjustment method for a prosthetic socket, which includes:

[0007] Obtaining the pressure values between the residual limb and the socket collected by multiple pressure sensors;

[0008] Extracting a pressure correlation pattern from the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a pressure correlation pattern feature vector;

[0009] Extract the spatial topological features between the multiple pressure sensors to obtain a spatial topological feature matrix;

[0010] Use an information fusion device to perform information fusion on the spatial topological feature matrix and the pressure correlation pattern feature vector to obtain a pressure correlation pattern feature vector containing spatial topological features;

[0011] Based on the pressure correlation pattern feature vector containing spatial topological features, determine whether to adjust the socket.

[0012] In the above prosthetic socket adaptive adjustment method, performing pressure correlation pattern extraction on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a pressure correlation pattern feature vector includes: performing data correction on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a corrected pressure input vector; inputting the corrected pressure input vector into a pressure correlation pattern feature extractor between positions based on a multi-scale neighborhood feature extraction network to obtain the pressure correlation pattern feature vector.

[0013] In the above prosthetic socket adaptive adjustment method, performing data correction on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a corrected pressure input vector includes: arranging the pressure values between the residual limb and the socket collected by the multiple pressure sensors into a pressure input vector; performing gamma correction on the pressure input vector to obtain the corrected pressure input vector.

[0014] In the above prosthetic socket adaptive adjustment method, performing gamma correction on the pressure input vector to obtain the corrected pressure input vector includes: correcting the pressure input vector with the following gamma correction formula to obtain the corrected pressure input vector; where the gamma correction formula is:

[0015]

[0016] where, L i is the i-th eigenvalue of the pressure input vector, A, B, C, and D are adjustment parameters with different numerical values, and L' i is the i-th eigenvalue of the corrected pressure input vector.

[0017] In the above prosthetic socket adaptive adjustment method, extracting the spatial topological features between the multiple pressure sensors to obtain a spatial topological feature matrix includes: constructing a spatial topological matrix between the multiple pressure sensors; passing the spatial topological matrix through a spatial topological feature extractor based on a convolutional neural network model to obtain the spatial topological feature matrix.

[0018] In the above adaptive adjustment method of the prosthetic socket, the eigenvalues at each position in the non-diagonal positions of the spatial topology matrix are the spatial distances between the corresponding two pressure sensors, and the eigenvalues at each position in the diagonal positions of the spatial topology matrix are 0.

[0019] In the above adaptive adjustment method of the prosthetic socket, an information fusion device is used to perform information fusion on the spatial topology feature matrix and the pressure correlation pattern feature vector to obtain a pressure correlation pattern feature vector including spatial topology features, including: passing the pressure correlation pattern feature vector through a point convolution layer to obtain a first convolution feature vector; passing the first convolution feature vector through a rectified linear unit based on the ReLU function to obtain a first rectified convolution feature vector; passing the first rectified convolution feature vector through a point convolution layer to obtain a second convolution feature vector; passing the second convolution feature vector through a rectified linear unit based on the Sigmoid function to obtain a second rectified convolution feature vector; fusing the second rectified convolution feature vector with the spatial topology feature matrix to obtain the pressure correlation pattern feature vector including spatial topology features.

[0020] In the above adaptive adjustment method of the prosthetic socket, based on the pressure correlation pattern feature vector including spatial topology features, it is determined whether to adjust the socket, including: passing the pressure correlation pattern feature vector including spatial topology features through a matching degree evaluator based on a classifier to obtain an evaluation result, and the evaluation result is used to represent the matching degree level label between the residual limb and the socket; based on the evaluation result, it is determined whether to adjust the socket.

[0021] In the above adaptive adjustment method of the prosthetic socket, a training step is further included: training the inter-position pressure correlation pattern feature extractor based on the multi-scale neighborhood feature extraction network, the spatial topology feature extractor based on the convolutional neural network model, the information fusion device, and the matching degree evaluator based on the classifier; wherein, the training step includes: obtaining training data, the training data including training pressure values between the residual limb and the socket collected by a plurality of pressure sensors, and true values of the matching degree level labels between the residual limb and the socket; arranging the training pressure values between the residual limb and the socket collected by the plurality of pressure sensors into a training pressure input vector; performing gamma correction on the training pressure input vector to obtain a training corrected pressure input vector; inputting the training corrected pressure input vector into the inter-position pressure correlation pattern feature extractor based on the multi-scale neighborhood feature extraction network to obtain a training pressure correlation pattern feature vector; constructing a training spatial topology matrix between the plurality of pressure sensors, wherein the feature value at each position in the non-diagonal positions of the training spatial topology matrix is the spatial distance between the corresponding two pressure sensors, and the feature value at each position in the diagonal positions is 0; inputting the training spatial topology matrix through the spatial topology feature extractor based on the convolutional neural network model to obtain a training spatial topology feature matrix; using the information fusion device to perform information fusion on the training spatial topology feature matrix and the training pressure correlation pattern feature vector to obtain a training pressure correlation pattern feature vector including training spatial topology features; inputting the training pressure correlation pattern feature vector including training spatial topology features through the matching degree evaluator based on the classifier to obtain a classification loss function value; using the classification loss function value as the loss function value to train the inter-position pressure correlation pattern feature extractor based on the multi-scale neighborhood feature extraction network, the spatial topology feature extractor based on the convolutional neural network model, the information fusion device, and the matching degree evaluator based on the classifier, wherein, in each round of iteration of the training, the training pressure correlation pattern feature vector including training spatial topology features is iteratively optimized.

[0022] According to another aspect of the present application, a prosthetic socket is provided, which includes:

[0023] A pressure monitoring module, configured to obtain pressure values between the residual limb and the socket collected by a plurality of pressure sensors;

[0024] A pressure correlation pattern extraction module, configured to perform pressure correlation pattern extraction on the pressure values between the residual limb and the socket collected by the plurality of pressure sensors to obtain a pressure correlation pattern feature vector;

[0025] A sensor spatial topology feature extraction module, which is used to extract the spatial topology features between the multiple pressure sensors to obtain a spatial topology feature matrix;

[0026] An information fusion module, which is used to use an information fusion device to perform information fusion on the spatial topology feature matrix and the pressure correlation pattern feature vector to obtain a pressure correlation pattern feature vector containing spatial topology features;

[0027] A socket adjustment analysis module, which is used to determine whether to adjust the socket based on the pressure correlation pattern feature vector containing spatial topology features.

[0028] Compared with the prior art, the prosthetic socket and its adaptive adjustment method provided by the present application use artificial intelligence technology based on deep learning to analyze the pressure values between the residual limb and the socket collected by multiple pressure sensors, capture the pressure correlation pattern between the position of the residual limb and the socket, and introduce the spatial topology features between the pressure sensors to strengthen the feature expression of the pressure correlation pattern between the position of the residual limb and the socket, so as to intelligently evaluate the matching degree between the residual limb and the socket, and then realize the adaptive adjustment of the prosthetic socket. In this way, the prosthetic socket can be more suitable for the residual limb, improve the comfort and stability of the prosthetic limb, reduce the pain and injury of the residual limb, and thus improve the quality of life of amputees. Description of the Drawings

[0029] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. They are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0030] Figure 1 It is a flowchart of the adaptive adjustment method of the prosthetic socket according to the embodiment of the present application.

[0031] Figure 2 It is a schematic structural diagram of the adaptive adjustment method of the prosthetic socket according to the embodiment of the present application.

[0032] Figure 3 It is a flowchart of extracting the pressure correlation pattern from the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain the pressure correlation pattern feature vector in the adaptive adjustment method of the prosthetic socket according to the embodiment of the present application.

[0033] Figure 4Flow chart for data correction of the pressure values between the residual limb and the socket collected by the multiple pressure sensors in the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application to obtain a corrected pressure input vector.

[0034] Figure 5 Flow chart for extracting the spatial topological features between the multiple pressure sensors in the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application to obtain a spatial topological feature matrix.

[0035] Figure 6 Flow chart for determining whether to adjust the socket based on the pressure correlation pattern feature vector including spatial topological features in the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application.

[0036] Figure 7 Flow chart for the training step in the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application.

[0037] Figure 8 Block diagram of the prosthetic socket according to an embodiment of the present application. Detailed implementation manners

[0038] Next, embodiments of the present application will be described in more detail with reference to the accompanying drawings, and the above and other objects, features, and advantages of the present application will become more obvious. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein. At the same time, the accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0039] Figure 1 Flow chart of the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application. Figure 2 Schematic architecture diagram of the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application. As Figure 1 and Figure 2As shown in the figure, the adaptive adjustment method of the prosthetic socket according to the embodiment of the present application includes the steps of: S110, obtaining the pressure values between the residual limb and the socket collected by a plurality of pressure sensors; S120, extracting the pressure correlation pattern of the pressure values between the residual limb and the socket collected by the plurality of pressure sensors to obtain a pressure correlation pattern feature vector; S130, extracting the spatial topological features between the plurality of pressure sensors to obtain a spatial topological feature matrix; S140, using an information fusion device to fuse the spatial topological feature matrix and the pressure correlation pattern feature vector to obtain a pressure correlation pattern feature vector including spatial topological features; S150, determining whether to adjust the socket based on the pressure correlation pattern feature vector including spatial topological features.

[0040] As mentioned in the above background technology, the prosthetic socket, as an indispensable auxiliary device for many amputees in their daily lives, is of great importance. To achieve the best function of the prosthesis, the connection between the residual limb and the socket is crucial. However, most of the prosthetic sockets on the current market rely on manual adjustment. This adjustment method is not only time-consuming and laborious, but also difficult to achieve an ideal fitting effect. Moreover, since the shape and size of each amputee's residual limb are unique, and daily activities may cause changes in the shape of the residual limb, the manual adjustment method often cannot adapt to these changes in a timely manner, and may even cause pain and injury to the residual limb, which undoubtedly increases the difficulty for amputees to use the prosthesis.

[0041] To address the above technical problems, the technical concept of the present application is to analyze the pressure values between the residual limb and the socket collected by a plurality of pressure sensors using artificial intelligence technology based on deep learning, capture the pressure correlation pattern between the position of the residual limb and the socket, and introduce the spatial topological features between each pressure sensor to strengthen the feature expression of the pressure correlation pattern between the position of the residual limb and the socket, so as to intelligently evaluate the matching degree between the residual limb and the socket, and then realize the adaptive adjustment of the prosthetic socket. In this way, the prosthetic socket can be more adapted to the residual limb, improve the comfort and stability of the prosthesis, reduce the pain and injury of the residual limb, and thus improve the quality of life of amputees.

[0042] In the above adaptive adjustment method of the prosthetic socket, in step S110, pressure values between the residual limb and the socket collected by multiple pressure sensors are obtained. It should be understood that the fitting degree between the prosthetic socket and the residual limb determines the wearing comfort and functional performance of the prosthetic limb. If the socket and the residual limb are very well-fitted, the pressure distribution between the inner wall of the socket and the residual limb should be uniform. That is to say, during the adaptive adjustment process of the prosthetic socket, the fitting degree between the socket and the residual limb can be evaluated by monitoring the pressure values at various positions between the residual limb and the socket, so as to adjust the structure of the socket targeted, making the socket and the residual limb more fitted, thereby reducing the relative movement between the residual limb and the socket, and avoiding the occurrence of excessive or too small pressure points, which may cause damage to the residual limb or loosening of the prosthetic limb. Based on this, in the technical solution of this application, multiple pressure sensors are arranged at various positions on the inner wall of the prosthetic socket to collect the pressure values between the residual limb and the socket, and then to understand the pressure distribution, so as to provide a basis for the adaptive adjustment of the socket.

[0043] In the above adaptive adjustment method of the prosthetic socket, in step S120, pressure correlation pattern extraction is performed on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a pressure correlation pattern feature vector. Specifically, Figure 3 It is a flowchart of performing pressure correlation pattern extraction on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a pressure correlation pattern feature vector in the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application. As Figure 3 shown, step S120 includes: S121, performing data correction on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a corrected pressure input vector; S122, inputting the corrected pressure input vector into a pressure correlation pattern feature extractor between positions based on a multi-scale neighborhood feature extraction network to obtain the pressure correlation pattern feature vector.

[0044] Specifically, in step S121, data correction is performed on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a corrected pressure input vector. More specifically, Figure 4 It is a flowchart of performing data correction on the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a corrected pressure input vector in the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application. As Figure 4 shown, step S121 includes: S1211, arranging the pressure values between the residual limb and the socket collected by the multiple pressure sensors into a pressure input vector; S1212, performing gamma correction on the pressure input vector to obtain the corrected pressure input vector.

[0045] Specifically, in step S1211, the pressure values between the residual limb and the socket collected by the multiple pressure sensors are arranged as a pressure input vector. It should be understood that considering that the pressure values between the residual limb and the socket collected by the multiple pressure sensors respectively describe the pressure states between a certain position on the inner wall of the prosthetic socket and the residual limb, and there is a certain correlation between the pressure states at each position. For example, when the patient walks, there are usually relatively high pressures at the metatarsal position and the fibular prominence position, while there may be relatively low pressures at the tibia position. That is to say, the areas corresponding to the protruding parts of the residual limb or the areas in closer contact with the inner wall of the socket are usually high-pressure areas; relatively speaking, the areas corresponding to the sunken parts of the residual limb or the areas in looser contact with the inner wall of the socket are usually low-pressure areas. Therefore, in order to obtain the pressure correlation pattern between the positions of the residual limb and the socket, so as to understand the pressure distribution, and thus provide data support for the adaptive adjustment of the prosthetic socket, the pressure values between the residual limb and the socket collected by the multiple pressure sensors are further arranged as a pressure input vector to integrate the pressure information between different positions, facilitating the subsequent analysis of the pressure correlation pattern between the positions of the residual limb and the socket.

[0046] Specifically, in step S1212, gamma correction is performed on the pressure input vector to obtain the corrected pressure input vector. It should be understood that considering that when using pressure sensors to collect pressure data between the residual limb and the socket, due to various factors such as the non-linear response of the sensors, the collected pressure data may have noise and errors. Therefore, in order to eliminate these noises and errors and improve the accuracy of the pressure data, gamma correction is further performed on the pressure input vector to obtain the corrected pressure input vector. Among them, gamma correction is a commonly used data correction method, which compensates for the non-linear response of the sensor by applying a gamma curve, can effectively eliminate the non-linear errors in the pressure input vector, improve the data accuracy, so that the corrected pressure input vector is closer to the real pressure distribution, and can more accurately reflect the pressure distribution state between the residual limb and the socket, in order to better support the subsequent feature extraction and matching degree evaluation processes of the pressure correlation pattern between the positions of the residual limb and the socket.

[0047] In a specific example of the present application, step S1212 includes: correcting the pressure input vector with the following gamma correction formula to obtain the corrected pressure input vector; where the gamma correction formula is:

[0048]

[0049] where, L i is the i-th eigenvalue of the pressure input vector, and A, B, C, and D are adjustment parameters with different numerical values, L' iis the i-th eigenvalue of the corrected pressure input vector.

[0050] Specifically, in step S122, the corrected pressure input vector is input into the inter-position pressure correlation pattern feature extractor based on the multi-scale neighborhood feature extraction network to obtain the pressure correlation pattern feature vector. It should be understood that the multi-scale neighborhood feature extraction network is a deep learning model that can extract data features from different spatial scales (receptive fields), thereby avoiding the information loss problem caused by a single scale in traditional methods and improving the accuracy of feature extraction. In the technical solution of this application, the inter-position pressure correlation pattern feature extractor adopts the multi-scale neighborhood feature extraction network, and performs a sliding convolution operation on the corrected pressure input vector through multiple convolutional kernels with different scales to capture the inter-position pressure correlation features between the residual limb and the socket at different feature space scales, revealing the pressure distribution law and correlation features between the residual limb and the socket, so as to more fully express the spatial position pressure correlation between the residual limb and the socket and provide strong support for subsequent matching degree evaluation.

[0051] In the above adaptive adjustment method of the prosthetic socket, in step S130, the spatial topological features between the multiple pressure sensors are extracted to obtain the spatial topological feature matrix. Specifically, Figure 5 is a flowchart of extracting the spatial topological features between the multiple pressure sensors to obtain the spatial topological feature matrix in the adaptive adjustment method of the prosthetic socket according to an embodiment of this application. As Figure 5 shown, step S130 includes: S131, constructing the spatial topological matrix between the multiple pressure sensors; S132, inputting the spatial topological matrix into the spatial topological feature extractor based on the convolutional neural network model to obtain the spatial topological feature matrix.

[0052] Specifically, in step S131, the spatial topological matrix between the multiple pressure sensors is constructed. It should be understood that considering that the spatial positions of the pressure sensors on the inner wall of the prosthetic socket have an important influence on the collected pressure distribution data. Therefore, in order to more accurately evaluate the fitting degree between the socket and the residual limb, the spatial topological matrix between the multiple pressure sensors is further constructed, thereby introducing the spatial distance features between the pressure sensors to strengthen the feature expression of the inter-position pressure correlation pattern between the residual limb and the socket. Specifically, in the technical solution of this application, the eigenvalue at each non-diagonal position in the spatial topological matrix is the spatial distance between the corresponding two pressure sensors, and the eigenvalue at each diagonal position is 0. In this way, by constructing the spatial topological matrix between the multiple pressure sensors, the spatial correlation between the pressure sensors is fully considered, and the pressure distribution correlation pattern between the residual limb and the socket can be understood more accurately.

[0053] Specifically, in step S132, the spatial topology matrix is passed through a spatial topology feature extractor based on a convolutional neural network model to obtain the spatial topology feature matrix. It should be understood that the spatial topology feature extractor is based on a convolutional neural network architecture and can model the spatial relationships in the spatial topology matrix through convolutional operations and pooling operations, effectively extracting the spatial topology features between each pressure sensor, converting the original spatial topology information into a more representative and interpretable feature representation, and providing more useful input for subsequent data analysis.

[0054] In the above-mentioned adaptive adjustment method of the prosthetic socket, in step S140, an information fusion device is used to fuse the spatial topology feature matrix and the pressure correlation pattern feature vector to obtain a pressure correlation pattern feature vector containing spatial topology features. It should be understood that the pressure correlation pattern feature vector characterizes the pressure correlation pattern between the residual limb and the socket, and the spatial topology feature matrix characterizes the spatial position topology information between each pressure sensor. By fusing the two, the pressure distribution correlation between the residual limb and the socket can be more comprehensively described, improving the characterization ability of the pressure correlation pattern feature vector, and thus providing more accurate data support for subsequent matching degree evaluation and adaptive adjustment. Specifically, the information fusion device is a meta-learning model, which can adaptively learn the relevant features in the spatial topology feature matrix and the pressure correlation pattern feature vector, make full use of the correlation between the two, and correlate and interact the spatial topology features between each pressure sensor with the pressure correlation pattern features between the residual limb and the socket, so as to introduce the spatial topology features into the pressure correlation pattern feature vector between the residual limb and the socket, thereby obtaining the pressure correlation pattern feature vector containing spatial topology features.

[0055] In a specific example of the present application, step S140 includes: passing the pressure correlation pattern feature vector through a point convolutional layer to obtain a first convolutional feature vector; passing the first convolutional feature vector through a rectified linear unit based on the ReLU function to obtain a first rectified convolutional feature vector; passing the first rectified convolutional feature vector through a point convolutional layer to obtain a second convolutional feature vector; passing the second convolutional feature vector through a rectified linear unit based on the Sigmoid function to obtain a second rectified convolutional feature vector; fusing the second rectified convolutional feature vector and the spatial topology feature matrix to obtain the pressure correlation pattern feature vector containing spatial topology features.

[0056] In the above-mentioned adaptive adjustment method of the prosthetic socket, in step S150, based on the pressure correlation pattern feature vector containing spatial topology features, it is determined whether to adjust the socket. Specifically,Figure 6 Flowchart for determining whether to adjust the prosthetic socket based on the pressure correlation pattern feature vector including spatial topological features in the adaptive adjustment method of the prosthetic socket according to an embodiment of the present application. As Figure 6 shown, step S150 includes: S151, passing the pressure correlation pattern feature vector including spatial topological features through a matching degree evaluator based on a classifier to obtain an evaluation result, where the evaluation result is used to represent the matching degree level label between the residual limb and the prosthetic socket; S152, determining whether to adjust the prosthetic socket based on the evaluation result.

[0057] Specifically, in step S151, the pressure correlation pattern feature vector including spatial topological features is passed through a matching degree evaluator based on a classifier to obtain an evaluation result, where the evaluation result is used to represent the matching degree level label between the residual limb and the prosthetic socket. It should be understood that a classifier is a type of algorithm widely used in pattern recognition and data classification, which can classify data according to the feature patterns of the input data. In the technical solution of the present application, by training the classifier, it learns the association relationship between different pressure correlation patterns between the residual limb and the prosthetic socket and the preset labels (the matching degree level labels between the residual limb and the prosthetic socket), and then maps the pressure correlation pattern feature vector including spatial topological features to the corresponding matching degree level labels according to this association relationship to achieve the evaluation of the matching degree between the residual limb and the prosthetic socket.

[0058] Specifically, in step S152, it is determined whether to adjust the prosthetic socket based on the evaluation result. That is to say, if the evaluation result indicates that the matching degree between the residual limb and the prosthetic socket does not reach the preset standard, then the prosthetic socket needs to be adjusted. Specifically, the parameters such as the inner diameter, depth, and shape of the prosthetic socket can be changed to make the prosthetic socket more conform to the shape and size of the residual limb, thereby improving the matching degree between the prosthetic socket and the residual limb. In this way, the use comfort and functional performance of the prosthetic limb can be effectively improved, providing a better quality of life for the residual limb patients.

[0059] It should be understood that before using the above neural network model, it is necessary to train the pressure correlation pattern feature extractor between positions based on the multi-scale neighborhood feature extraction network, the spatial topological feature extractor based on the convolutional neural network model, the information fusion device, and the matching degree evaluator based on the classifier. That is to say, in the adaptive adjustment method of the prosthetic socket of the present application, it also includes a training step of training the pressure correlation pattern feature extractor between positions based on the multi-scale neighborhood feature extraction network, the spatial topological feature extractor based on the convolutional neural network model, the information fusion device, and the matching degree evaluator based on the classifier.

[0060] Figure 7The figure is a flowchart of the training step in the adaptive adjustment method of a prosthetic socket according to an embodiment of the present application. As Figure 7 shown, the training step S200 includes: S210, obtaining training data, where the training data includes training pressure values between the residual limb and the socket collected by a plurality of pressure sensors, and true values of the matching degree level labels between the residual limb and the socket; S220, arranging the training pressure values between the residual limb and the socket collected by the plurality of pressure sensors into a training pressure input vector; S230, performing gamma correction on the training pressure input vector to obtain a training corrected pressure input vector; S240, inputting the training corrected pressure input vector into the inter-position pressure correlation pattern feature extractor based on the multi-scale neighborhood feature extraction network to obtain a training pressure correlation pattern feature vector; S250, constructing a training spatial topology matrix between the plurality of pressure sensors, where the eigenvalue of each position in the non-diagonal positions of the training spatial topology matrix is the spatial distance between the corresponding two pressure sensors, and the eigenvalue of each position in the diagonal positions is 0; S260, passing the training spatial topology matrix through the spatial topology feature extractor based on the convolutional neural network model to obtain a training spatial topology feature matrix; S270, using the information fusion device to perform information fusion on the training spatial topology feature matrix and the training pressure correlation pattern feature vector to obtain a training pressure correlation pattern feature vector including training spatial topology features; S280, passing the training pressure correlation pattern feature vector including training spatial topology features through the matching degree evaluator based on the classifier to obtain a classification loss function value; S290, using the classification loss function value as the loss function value to train the inter-position pressure correlation pattern feature extractor based on the multi-scale neighborhood feature extraction network, the spatial topology feature extractor based on the convolutional neural network model, the information fusion device, and the matching degree evaluator based on the classifier, where in each round of iteration of the training, the training pressure correlation pattern feature vector including training spatial topology features is iteratively optimized.

[0061] In the above technical solution, the training pressure correlation pattern feature vector expresses the multi-scale local neighborhood time series correlation features in the local sample space domain after the training pressure values collected by the pressure sensors are corrected by non-linear response. Thus, after using the information fusion device to perform information fusion on the training spatial topology feature matrix and the training pressure correlation pattern feature vector, the time-space cross-dimensional global sample space domain topology correlation features based on the local sample space domain distance topology correlation features in the global sample space domain can be further extracted.

[0062] However, considering that the training pressure correlation pattern feature vector containing the topological features of the training space includes multi-scale time series correlation based on the time dimension and time series feature space distance topological correlation based on the space dimension, it is desired to improve the iterative convergence speed of the classifier for the training pressure correlation pattern feature vector containing the topological features of the training space under multi-dimensional feature fusion representation.

[0063] Therefore, in the technical solution of the present application, each time the training pressure correlation pattern feature vector containing the topological features of the training space is iteratively classified and regressed by the classifier, the training pressure correlation pattern feature vector containing the topological features of the training space is iteratively optimized according to the following optimization formula to obtain an optimized training pressure correlation pattern feature vector containing the topological features of the training space, where the optimization formula is:

[0064]

[0065] where v i is the i-th eigenvalue of the training pressure correlation pattern feature vector containing the topological features of the training space, φ(v i ) represents the probability function of the eigenvalue, that is, the probability function that maps the eigenvalue v i to the interval [0, 1], and p is the class probability value obtained by the classifier, L is the length of the training pressure correlation pattern feature vector containing the topological features of the training space, α is the weight hyperparameter, exp(·) represents the exponential operation with base e, and v' i is the i-th eigenvalue of the optimized training pressure correlation pattern feature vector containing the topological features of the training space.

[0066] Here, by using the Bernoulli probability likelihood distribution as the authoritative meta-entity to perform high-level class decisions on the eigenvalue v i of the training pressure correlation pattern feature vector containing the topological features of the training space based on the Bernoulli probability distribution (p, 1 - p), and introducing the probability density space mapping assignment relative to the overall feature set, thereby realizing the key insight of the eigenvalue v i of the training pressure correlation pattern feature vector containing the topological features of the training space relative to the class probability mapping, to achieve the matching function correctness of the eigenvalue v i of the training pressure correlation pattern feature vector containing the topological features of the training space with class convergence based on the class probability distribution (p, 1 - p), so as to promote sample iterative convergence through convergence performance superiority and improve the classification training speed.

[0067] In summary, the adaptive adjustment method of the prosthetic socket according to the embodiments of the present application is elucidated. It uses artificial intelligence technology based on deep learning to analyze the pressure values between the residual limb and the socket collected by multiple pressure sensors, captures the pressure correlation pattern between the position of the residual limb and the socket, and introduces the spatial topological features between the pressure sensors to strengthen the feature expression of the pressure correlation pattern between the position of the residual limb and the socket, thereby intelligently evaluating the matching degree between the residual limb and the socket, and further realizing the adaptive adjustment of the prosthetic socket. In this way, the prosthetic socket can be more adapted to the residual limb, improving the comfort and stability of the prosthesis, reducing the pain and injury of the residual limb, and thus enhancing the quality of life of amputees.

[0068] Figure 8 FIG. is a block diagram of a prosthetic socket according to an embodiment of the present application. As Figure 8 shown, the prosthetic socket 100 according to an embodiment of the present application includes: a pressure monitoring module 110 for obtaining the pressure values between the residual limb and the socket collected by multiple pressure sensors; a pressure correlation pattern extraction module 120 for extracting a pressure correlation pattern from the pressure values between the residual limb and the socket collected by the multiple pressure sensors to obtain a pressure correlation pattern feature vector; a sensor spatial topological feature extraction module 130 for extracting the spatial topological features between the multiple pressure sensors to obtain a spatial topological feature matrix; an information fusion module 140 for using an information fusion device to fuse the spatial topological feature matrix and the pressure correlation pattern feature vector to obtain a pressure correlation pattern feature vector including spatial topological features; and a socket adjustment analysis module 150 for determining whether to adjust the socket based on the pressure correlation pattern feature vector including spatial topological features.

[0069] Here, those skilled in the art can understand that the specific operations of the above-mentioned modules in the prosthetic socket have been described in detail in the description of the adaptive adjustment method of the prosthetic socket above with reference to Figures 1 to 7 and therefore, the repeated description thereof will be omitted.

[0070] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, advantages, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, advantages, effects, etc. are essential for each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and easy understanding, rather than limitations, and the above details do not limit the present invention to necessarily adopt the above specific details to implement.

[0071] In the above embodiments, the descriptions of the various embodiments each have their own focuses. For parts that are not elaborated or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0072] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced by the present invention. Any associated drawing marks in the claims should not be regarded as limiting the claimed rights.

[0073] In addition, it is obvious that the word "including" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0074] Finally, it should be noted that the above description has been given for purposes of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A prosthetic socket, characterized in that, Including: A pressure monitoring module for obtaining the pressure values between the residual limb and the socket collected by a plurality of pressure sensors; A pressure correlation pattern extraction module for extracting a pressure correlation pattern from the pressure values between the residual limb and the socket collected by the plurality of pressure sensors to obtain a pressure correlation pattern feature vector, which characterizes the pressure correlation pattern between the positions of the residual limb and the socket; A sensor spatial topology feature extraction module for extracting the spatial topology features between the plurality of pressure sensors to obtain a spatial topology feature matrix, which characterizes the spatial position topology information between the respective pressure sensors; An information fusion module for using an information fusion device to fuse the spatial topology feature matrix and the pressure correlation pattern feature vector to obtain a pressure correlation pattern feature vector including spatial topology features; A socket adjustment analysis module for determining whether to adjust the socket based on the pressure correlation pattern feature vector including spatial topology features; The information fusion module is configured to: Pass the pressure correlation pattern feature vector through a point convolution layer to obtain a first convolution feature vector; Pass the first convolution feature vector through a rectified linear unit based on the ReLU function to obtain a first rectified convolution feature vector; Pass the first rectified convolution feature vector through a point convolution layer to obtain a second convolution feature vector; Pass the second convolution feature vector through a rectified linear unit based on the Sigmoid function to obtain a second rectified convolution feature vector; Fuse the second rectified convolution feature vector with the spatial topology feature matrix to obtain the pressure correlation pattern feature vector including spatial topology features.

Images