TWI881857B - Sarcopenia determination system and method - Google Patents

Abstract

A sarcopenia determination system comprises: an image capture device, which captures the two-dimensional image changes of a target test subject within a time period to generate two-dimensional image data; and a computer device, which is communicatively connected to the image capture device, and the computer device comprises: a receiving module, which receives the two-dimensional image data from the image capture device; and an analysis module, which is communicatively connected to the receiving module, and the analysis module generates three-dimensional image data based on the two-dimensional image data, and generates sarcopenia determination data based on the three-dimensional image data, and the sarcopenia determination data indicates whether the target test subject suffers from sarcopenia.

Description

Sarcopenia determination system and method

The present invention relates to a sarcopenia diagnosis system and method, and in particular to a sarcopenia diagnosis system and method that allow users to identify or diagnose sarcopenia directly at home without having to go to a hospital or clinic.

At present, the clinical diagnosis methods for sarcopenia are relatively inaccurate, and some of them are inconvenient for patients. For example, if physical function tests or imaging examinations are required, patients need to go to the hospital. This is a considerable burden for people with sarcopenia, those with limited mobility, or those who are older. In view of this, a sarcopenia diagnosis system and method is needed that allows users to identify or diagnose sarcopenia directly at home without going to a hospital or clinic.

In order to solve the above problems, one of the concepts of the present invention is to provide a sarcopenia diagnosis system and method that can directly identify or diagnose sarcopenia at home.

Based on the above-mentioned concept, the present invention provides a sarcopenia judgment system, comprising: an image capture device, capturing two-dimensional image changes of a target tester within a time period to generate two-dimensional image data; and a computer device, communicatively connected to the image capture device, the computer device comprising: a receiving module, receiving the two-dimensional image data from the image capture device; and an analysis module, communicatively connected to the receiving module, the analysis module generates three-dimensional image data based on the two-dimensional image data, and generates sarcopenia judgment data based on the three-dimensional image data, the sarcopenia judgment data indicating whether the target tester suffers from sarcopenia.

In a preferred embodiment of the present invention, the two-dimensional image data indicates image changes of at least one of the test actions of the target tester, including balance action, one-legged standing action, in-place stepping action, and straight-line walking action, during the time period.

In a preferred embodiment of the present invention, the two-dimensional image data includes a first two-dimensional image data; wherein the analysis module identifies a plurality of key points from the first two-dimensional image data, and the analysis module generates the three-dimensional image data at least based on the plurality of key points.

In a preferred embodiment of the present invention, the analysis module respectively maps the plurality of key points to a three-dimensional coordinate system to generate a plurality of first three-dimensional coordinate points; wherein the analysis module generates a first three-dimensional image data based on the plurality of first three-dimensional coordinate points, and the analysis module makes the three-dimensional image data include the first three-dimensional image data.

In a preferred embodiment of the present invention, the two-dimensional image data includes a second two-dimensional image data; wherein the analysis module identifies a plurality of key points from the second two-dimensional image data; wherein the analysis module selects a plurality of important points from the plurality of key points, and the analysis module generates the three-dimensional image data at least based on the plurality of important points.

In a preferred embodiment of the present invention, the analysis module respectively corresponds the plurality of important points to a three-dimensional coordinate system to generate a plurality of second three-dimensional coordinate points; wherein the analysis module generates a second three-dimensional image data based on the plurality of second three-dimensional coordinate points, and the analysis module makes the three-dimensional image data include the second three-dimensional image data.

In a preferred embodiment of the present invention, the analysis module has a first deep learning model, and the analysis module uses the first deep learning model to generate the three-dimensional image data based on the two-dimensional image data.

In a preferred embodiment of the present invention, the analysis module has a second deep learning model; wherein the analysis module uses the second deep learning model to generate the sarcopenia judgment data based on the three-dimensional image data.

In a preferred embodiment of the present invention, the second deep learning model is a transformer neural network model.

In a preferred embodiment of the present invention, the sarcopenia judgment system further includes a database storing a plurality of three-dimensional image training data, wherein the plurality of three-dimensional image training data respectively have a data time length; wherein the computer device accesses the database, the computer device makes the data time length of each of the plurality of three-dimensional image training data the same as each other, and the computer device uses the plurality of three-dimensional image training data to train the analysis module so that the analysis module can generate the sarcopenia judgment data based on the three-dimensional image data.

In a preferred embodiment of the present invention, the plurality of three-dimensional image training data includes a first training data and a second training data, the first training data includes a first final image data, a first data time length of the first training data is less than a second data time length of the second training data; wherein the computer device generates a plurality of first image filling data according to the first final image data , the plurality of first image filling data are all equal to the first last image data; wherein the computer device connects the plurality of first image filling data to the first last image data of the first training data, and makes the first training data include the plurality of first image filling data, so that the first data time length of the first training data is equal to the second data time length of the second training data.

In a preferred embodiment of the present invention, the plurality of three-dimensional image training data respectively have a label data, and the plurality of label data respectively indicate that the corresponding three-dimensional image training data is a sarcopenia training data or a non-sarcopenia training data.

In a preferred embodiment of the present invention, the plurality of three-dimensional image training data includes a second training data having a second data time length; wherein the three-dimensional image data includes a second final image data, and the three-dimensional image data has a third data time length, and the third data time length is less than the second data time length; wherein the computer device generates a plurality of The computer device connects the plurality of second image filling data to the second last image data of the three-dimensional image data, and makes the three-dimensional image data include the plurality of second image filling data, so that the third data time length of the three-dimensional image data is equal to the second data time length of the second training data.

According to the purpose of the present invention, a sarcopenia determination method is provided, which is applied to a sarcopenia determination system. The sarcopenia determination system includes an image capture device and a computer device, the computer device is communicatively connected to the image capture device, and the computer device includes a receiving module and an analysis module, the analysis module is communicatively connected to the receiving module, and the sarcopenia determination method includes the following steps: The device captures the two-dimensional image changes of a target tester within a time period to generate two-dimensional image data; the receiving module receives the two-dimensional image data from the image capture device; the analysis module generates three-dimensional image data based on the two-dimensional image data; and the analysis module generates sarcopenia judgment data based on the three-dimensional image data, and the sarcopenia judgment data indicates whether the target tester suffers from sarcopenia.

In a preferred embodiment of the present invention, the two-dimensional image data indicates image changes of at least one of the test actions of the target tester, including balance action, one-legged standing action, in-place stepping action, and straight-line walking action, during the time period.

In a preferred embodiment of the present invention, the two-dimensional image data includes a first two-dimensional image data; wherein the sarcopenia determination method includes the following steps: the analysis module identifies a plurality of key points from the first two-dimensional image data; and the analysis module generates the three-dimensional image data at least based on the plurality of key points.

In a preferred embodiment of the present invention, the sarcopenia determination method further includes the following steps: the analysis module respectively maps the plurality of key points to a three-dimensional coordinate system to generate a plurality of first three-dimensional coordinate points; the analysis module generates a first three-dimensional image data based on the plurality of first three-dimensional coordinate points; and the analysis module causes the three-dimensional image data to include the first three-dimensional image data.

In a preferred embodiment of the present invention, the two-dimensional image data includes a second two-dimensional image data; wherein the sarcopenia determination method includes the following steps: the analysis module identifies a plurality of key points from the second two-dimensional image data; the analysis module selects a plurality of important points from the plurality of key points; and the analysis module generates the three-dimensional image data at least based on the plurality of important points.

In a preferred embodiment of the present invention, the sarcopenia determination method further includes the following steps: the analysis module respectively corresponds the plurality of important points to a three-dimensional coordinate system to generate a plurality of second three-dimensional coordinate points; the analysis module generates a second three-dimensional image data based on the plurality of second three-dimensional coordinate points; and the analysis module causes the three-dimensional image data to include the second three-dimensional image data.

In a preferred embodiment of the present invention, the analysis module has a first deep learning model; wherein the sarcopenia judgment method includes the following steps: the analysis module uses the first deep learning model to generate the three-dimensional image data based on the two-dimensional image data.

In a preferred embodiment of the present invention, the analysis module has a second deep learning model; wherein the sarcopenia judgment method includes the following steps: the analysis module uses the second deep learning model to generate the sarcopenia judgment data based on the three-dimensional image data.

In a preferred embodiment of the present invention, the second deep learning model is a transformer neural network model.

In a preferred embodiment of the present invention, the sarcopenia determination system further includes a database, the computer device accesses the database, the database stores a plurality of three-dimensional image training data, and the plurality of three-dimensional image training data respectively have a data time length; wherein the sarcopenia determination method further includes the following steps: the computer device makes the data time length of each of the plurality of three-dimensional image training data the same as each other; and the computer device uses the plurality of three-dimensional image training data to train the analysis module so that the analysis module can generate the sarcopenia determination data based on the three-dimensional image data.

In a preferred embodiment of the present invention, the plurality of three-dimensional image training data includes a first training data and a second training data, the first training data includes a first final image data, and a first data time length of the first training data is less than a second data time length of the second training data; wherein the sarcopenia determination method further includes the following steps: the computer device generates a plurality of training data according to the first final image data The first image filling data is a plurality of first image filling data, and the plurality of first image filling data are all equal to the first last image data; and the computer device connects the plurality of first image filling data to the first last image data of the first training data, and makes the first training data include the plurality of first image filling data, so that the first data time length of the first training data is equal to the second data time length of the second training data.

In a preferred embodiment of the present invention, the plurality of three-dimensional image training data respectively have a label data, and the plurality of label data respectively indicate that the corresponding three-dimensional image training data is a sarcopenia training data or a non-sarcopenia training data.

In a preferred embodiment of the present invention, the plurality of three-dimensional image training data includes a second training data having a second data time length; wherein the three-dimensional image data includes a second final image data, and the three-dimensional image data also has a third data time length, and the third data time length is less than the second data time length; wherein the sarcopenia determination method further includes the following steps: the computer device calculates the second final image data according to the second final image data; The image data generates a plurality of second image filling data, and the plurality of second image filling data are all equal to the second last image data; and the computer device connects the plurality of second image filling data to the second last image data of the three-dimensional image data, and makes the three-dimensional image data include the plurality of second image filling data, so that the third data time length of the three-dimensional image data is equal to the second data time length of the second training data.

The above aspects and other aspects of the invention will become more apparent from the following non-limiting detailed description of specific embodiments and with reference to the accompanying drawings.

Please refer to the first figure, which illustrates a system architecture diagram of a specific embodiment of the sarcopenia determination system of the present invention. As shown in the first figure, the sarcopenia determination system 100 may include a computer device 110 and an image capture device 120, and the computer device 110 may be communicatively connected to the image capture device 120. The computer device 110 includes a receiving module 112 and an analysis module 114, and the analysis module 114 is communicatively connected to the receiving module 112. The computer device 110 may be, for example, a mobile device, a server, a computer, a laptop, etc., but is not limited thereto. The computer device 110 may also be other devices with data processing and data calculation functions. The image capture device 120 may be, for example, a mobile device, a camera, a computer, a laptop, etc., but not limited thereto. The image capture device 120 may also be other devices with image capture and/or image storage functions. Preferably, the computer device 110 may have one or more processors, and implement the functions of the computer device 110 and each module included therein in a manner of hardware and software collaboration. Preferably, the image capture device 120 may have one or more processors, and implement the functions of the image capture device 120 (such as image capture function, image storage function, image transmission function, etc., but not limited thereto) in a manner of hardware and software collaboration.

In the embodiment shown in the first figure, the image capture device 120 can detect and capture the two-dimensional image changes of the target test subject within a time period, so as to generate corresponding two-dimensional image data according to the two-dimensional image changes of the target test subject within the time period. After the image capture device 120 generates the two-dimensional image data, the receiving module 112 of the computer device 110 can receive the two-dimensional image data from the image capture device 120. Then, the analysis module 114 of the computer device 110 can generate three-dimensional image data according to the two-dimensional image data. Then, the analysis module 114 of the computer device 110 can perform data processing and/or calculation and/or judgment according to the three-dimensional image data, and then generate sarcopenia judgment data. The sarcopenia determination data indicates whether the target tester suffers from sarcopenia.

Preferably, the sarcopenia determination data indicates whether the target tester suffers from sarcopenia in the form of a probability value. For example, the sarcopenia determination data may include a probability data indicating a probability value that the target tester suffers from sarcopenia. Preferably, the image capture device 120 is a mobile device. It should be understood that most users or patients have mobile devices, and mobile devices are easy to carry, so it is convenient for users to use the sarcopenia determination system 100 to detect and determine sarcopenia at home or anywhere. In addition, since the analysis model of the sarcopenia judgment system 100 will not have subjective errors (for example, if the judgment is made by medical personnel, there will be subjective errors in the judgment of short time or whether the movement is smooth), the accuracy of the sarcopenia judgment system 100 in judging sarcopenia can be greatly improved compared to traditional methods.

Preferably, the analysis module 114 may include (or store) a first deep learning model, and the analysis module 114 uses the first deep learning model to generate three-dimensional image data based on the two-dimensional image data. Preferably, the first deep learning model is a neural network model (which may be referred to as a first type of neural network model). Preferably, the analysis module 114 may include (or store) a second deep learning model, and the analysis module 114 uses the second deep learning model to generate sarcopenia judgment data based on the three-dimensional image data. Preferably, the second deep learning model is a neural network model (which may be referred to as a second type of neural network model). Preferably, the second deep learning model is a transformer neural network model.

Preferably, the target tester uses the sarcopenia determination system 100 to determine whether the target tester suffers from sarcopenia when performing at least one of the test actions of balancing, standing on one foot, stepping on the spot, and walking in a straight line. That is, the image capture device 120 captures the target tester's movement changes or image changes to generate two-dimensional image data when the target tester performs one or more of the test actions of balancing, standing on one foot, stepping on the spot, and walking in a straight line. Therefore, the two-dimensional image data can indicate the image changes of the target tester performing at least one of the test actions of balancing, standing on one foot, stepping on the spot, and walking in a straight line.

Preferably, the image capture device 120 may include a data storage unit that can store the two-dimensional image data generated by the image capture device 120. In this way, the user only needs to carry the image capture device 120 to perform action tests (such as at least one of the balance action, one-legged standing action, stepping action, and straight walking action, but not limited to this) at any time and collect the corresponding two-dimensional image data. Then, the two-dimensional image data in the data storage unit of the image capture device 120 is transmitted (wired transmission or wireless transmission) to the computer device 110, and the computer device 110 can analyze and judge the two-dimensional image data and generate corresponding sarcopenia judgment data.

In a specific embodiment, the two-dimensional image data generated by the image capture device 120 may include a plurality of two-dimensional image data, and the plurality of two-dimensional image data includes a first two-dimensional image data. The analysis module 114 may identify a plurality of key points from the first two-dimensional image data, and the analysis module 114 may generate three-dimensional image data at least based on the plurality of key points.

In a specific embodiment, the analysis module 114 may correspond the plurality of key points to a three-dimensional coordinate system, thereby generating a plurality of first three-dimensional coordinate points. The plurality of first three-dimensional coordinate points are respectively associated with or correspond to one of the plurality of key points (for example, the plurality of first three-dimensional coordinate points may respectively indicate the three-dimensional position of one of the plurality of key points in the three-dimensional coordinate system). Then, the analysis module 114 may generate first three-dimensional image data according to the plurality of first three-dimensional coordinate points, and the analysis module 114 may make the three-dimensional image data include the first three-dimensional image data. The first three-dimensional image data includes the plurality of first three-dimensional coordinate points, or the first three-dimensional image data indicates the plurality of first three-dimensional coordinate points. Preferably, the analysis module 114 can use the same method to convert multiple two-dimensional image data in the two-dimensional image data into corresponding three-dimensional image data respectively, and the analysis module 114 can also make the three-dimensional image data include the multiple three-dimensional image data.

Preferably, the key points may correspond to specific parts of the target tester in the two-dimensional image data. For example, the key points may correspond to one or more of the center head, upper torso, center torso, bottom torso, right shoulder, left shoulder, right elbow, left elbow, right hand, left hand, right hip, left hip, right knee, left knee, right foot, left foot, and virtual centroid (or body centroid) of the target tester in the two-dimensional image data, but are not limited thereto.

In a specific embodiment, the two-dimensional image data generated by the image capture device 120 may include a plurality of two-dimensional image data, and the plurality of two-dimensional image data includes second two-dimensional image data. The analysis module 114 may identify a plurality of key points from the second two-dimensional image data, and the analysis module 114 may then select a plurality of important points from the plurality of key points, and the analysis module 114 may generate three-dimensional image data at least based on the plurality of important points.

In a specific embodiment, the analysis module 114 may correspond the plurality of important points to a three-dimensional coordinate system, thereby generating a plurality of second three-dimensional coordinate points. The plurality of second three-dimensional coordinate points are respectively associated with or correspond to one of the plurality of important points (for example, the plurality of second three-dimensional coordinate points may respectively indicate the three-dimensional position of one of the plurality of important points in the three-dimensional coordinate system). Then, the analysis module 114 may generate second three-dimensional image data according to the plurality of second three-dimensional coordinate points, and the analysis module 114 may make the three-dimensional image data include the second three-dimensional image data. The second three-dimensional image data includes the plurality of second three-dimensional coordinate points, or the second three-dimensional image data indicates the plurality of second three-dimensional coordinate points. Preferably, the analysis module 114 can use the same method to convert multiple two-dimensional image data in the two-dimensional image data into corresponding three-dimensional image data respectively, and the analysis module 114 can also make the three-dimensional image data include the multiple three-dimensional image data.

Preferably, the important points may correspond to specific parts of the target tester in the two-dimensional image data. For example, the important points may correspond to the center head, upper torso, center torso, bottom torso, right shoulder, left shoulder, right hip, left hip, right knee, left knee, and virtual centroid (or body centroid) of the target tester in the two-dimensional image data, but the present invention is not limited thereto.

It should be understood that since some key points do not have much significance in the analysis, removing some key points and only analyzing and judging based on important points will not significantly reduce the accuracy of the judgment, and this approach can even improve the accuracy of the judgment. In addition, since the number of important points is lower than the number of key points, this approach of only analyzing and judging based on important points will reduce the time required for analysis.

In a specific embodiment, the sarcopenia determination system 100 may further include a database 130, the computer device 110 is communicatively connected to the database 130, and the computer device 110 can access the database 130. The database 130 can store a plurality of three-dimensional image training data, and the plurality of three-dimensional image training data respectively have a data time length. The computer device 110 can make the data time length of each of the plurality of three-dimensional image training data the same as each other, and the computer device 110 can use the plurality of three-dimensional image training data to train the analysis module 114, so that the analysis module 114 can generate sarcopenia determination data according to the three-dimensional image data. Preferably, the computer device uses a plurality of three-dimensional image training data to train the second deep learning model of the analysis module 114, so that the second deep learning model of the analysis module 114 can generate sarcopenia judgment data based on the three-dimensional image data.

Preferably, the database 130 may have one or more processors, and implement the functions of the database 130 in a manner of hardware and software working in coordination. Preferably, a plurality of three-dimensional image training data respectively have a label data, and the plurality of label data respectively indicate that the corresponding three-dimensional image training data is sarcopenia training data or non-sarcopenia training data. Preferably, if the label data of a three-dimensional image training data indicates that the three-dimensional image training data is sarcopenia training data, it means that the three-dimensional image training data is training data measured from a patient suffering from sarcopenia. If the label data of a three-dimensional image training data indicates that the three-dimensional image training data is non-sarcopenia training data, it means that the three-dimensional image training data is training data measured from a target subject who does not suffer from sarcopenia.

In a specific embodiment, the plurality of three-dimensional image training data includes first training data and second training data, and the first training data includes first and last image data. The first data time length of the first training data is less than the second data time length of the second training data. The first data time indicates the video time length of the first training data, and the second data time indicates the video time length of the second training data.

The computer device 110 may generate a plurality of first image filling data according to the first last image data (preferably, the computer device 110 uses the first last image data to generate a plurality of first image filling data based on the first data time length being less than the second data time length). The plurality of first image filling data are all equal to the first last image data (that is, the plurality of first image filling data and the first last image data are all the same image). The computer device 110 may connect the plurality of first image filling data to the first last image data of the first training data, and make the first training data include the plurality of first image filling data, so that the first data time length of the first training data is equal to the second data time length of the second training data. In short, the computer device 110 may use the plurality of first image filling data to fill the first training data, so as to increase the first data time length of the first training data, and make the first data time length of the first training data equal to the second data time length of the second training data.

Preferably, the computer device 110 can use the above-mentioned data filling method to adjust the data time length of each of the plurality of three-dimensional image training data. In this way, the plurality of three-dimensional image training data can have the same data time length. Preferably, the data time length of the second training data is greater than or equal to the original data time length of the remaining three-dimensional image training data. That is, the computer device 110 uses the three-dimensional image training data with the longest data time length as a benchmark to adjust the data time length of the remaining three-dimensional image training data so that the data time lengths of all three-dimensional image training data are the same. However, it should be understood that the computer device 110 can not only use the 3D image training data with the longest data time length as the benchmark for the video time length, but can use other 3D image training data as the benchmark as needed, or can define a benchmark time length as the time length benchmark for other 3D image training data.

In a specific embodiment, the plurality of three-dimensional image training data includes second training data, and the second training data has a second data time length. The three-dimensional image data captured by the image capture device 120 includes the second last image data, and the three-dimensional image data has a third data time length. The third data time length is less than the second data time length.

The computer device 110 may generate a plurality of second image filling data according to the second last image data of the three-dimensional image data (preferably, the computer device 110 uses the first last image data to generate a plurality of first image filling data based on the third data time length being less than the second data time length). The plurality of second image filling data are all equal to the second last image data (that is, the plurality of second image filling data and the second last image data are the same image). The computer device 110 may attach a plurality of second image filling data to the second last image data of the three-dimensional image data, and make the three-dimensional image data include a plurality of second image filling data, so that the third data time length of the three-dimensional image data is equal to the second data time length of the second training data. In short, the computer device 110 may use a plurality of second image filling data to fill the three-dimensional image data generated by the image capture device 120, so as to increase the first data time length of the three-dimensional image data, and make the third data time length of the three-dimensional image data equal to the second data time length of the second training data.

Preferably, the data time length of the second training data is greater than or equal to the original data time length of the remaining three-dimensional image training data. That is, the computer device 110 uses the three-dimensional image training data with the longest data time length as a reference to adjust the data time length of the three-dimensional image data generated by the image capture device 120, so that the third data time length of the three-dimensional image data is the same as the data time length of all three-dimensional image training data.

Please refer to the second figure, which illustrates a schematic diagram of a specific embodiment of two-dimensional image data and key points. As shown in the embodiment of the second figure, the sarcopenia determination system can identify a plurality of key points (such as key points 212, 214, 216 in the figure) from the two-dimensional image data 210. Please refer to the third figure, which illustrates a schematic diagram of a specific embodiment of key points and important points. As shown in the embodiment of the third figure, after the sarcopenia determination system identifies a plurality of key points (such as key points 311~319 in the third figure) from the two-dimensional image data, the sarcopenia determination system can further select a plurality of important points 321~330 from the plurality of key points. Among them, important point 321 corresponds to the target tester's left shoulder, important point 322 corresponds to the target tester's upper torso, important point 323 corresponds to the target tester's center of head, important point 324 corresponds to the target tester's right shoulder, important point 325 corresponds to the target tester's center of torso, important point 326 corresponds to the target tester's left hip, important point 327 corresponds to the target tester's bottom torso, important point 328 corresponds to the target tester's right hip, important point 329 corresponds to the target tester's left knee, and important point 330 corresponds to the target tester's right knee.

Please refer to FIG. 4A and FIG. 4B, which respectively illustrate schematic diagrams of different specific embodiments of three-dimensional image data. As shown in the embodiment of FIG. 4A, the sarcopenia determination system corresponds a plurality of key points to a three-dimensional coordinate system to generate a plurality of three-dimensional coordinate points (for example, the three-dimensional coordinate points 412, 414, 416, etc. in FIG. 4A). The sarcopenia determination system generates three-dimensional image data 410 based on these three-dimensional coordinate points. As shown in the embodiment of FIG. 4B, the sarcopenia determination system corresponds a plurality of important points to a three-dimensional coordinate system to generate a plurality of three-dimensional coordinate points (for example, the three-dimensional coordinate points 422, 424, 426, etc. in FIG. 4A). The sarcopenia determination system generates three-dimensional image data 420 based on the three-dimensional coordinate points.

Please refer to the fifth figure, which illustrates a flow chart of a specific embodiment of the sarcopenia determination method according to the present invention. As shown in the embodiment of the fifth figure, the sarcopenia determination method 500 can be applied to a sarcopenia determination system. Among them, the sarcopenia determination system includes an image capture device and a computer device, the computer device can be communicatively connected to the image capture device, and the computer device includes a receiving module and an analysis module, and the analysis module is communicatively connected to the receiving module. The sarcopenia determination method 500 starts at step 510, and the image capture device captures the two-dimensional image changes of the target tester within a time period to generate two-dimensional image data. Then, step 520 is executed, and the receiving module receives the two-dimensional image data from the image capture device. Next, step 530 is executed, and the analysis module generates three-dimensional image data according to the two-dimensional image data. Next, step 540 is executed, and the analysis module generates sarcopenia determination data according to the three-dimensional image data. The sarcopenia determination data indicates whether the target test subject suffers from sarcopenia.

In a specific embodiment, the two-dimensional image data indicates image changes of at least one of the test actions of the target tester, including balance action, one-legged standing action, in-place stepping action, and straight-line walking action, during the time period.

In a specific embodiment, the two-dimensional image data includes a first two-dimensional image data. The sarcopenia determination method 500 further includes the following steps: the analysis module identifies a plurality of key points from the first two-dimensional image data; and the analysis module generates three-dimensional image data based on at least the plurality of key points.

In a specific embodiment, the sarcopenia determination method 500 further includes the following steps: the analysis module respectively maps a plurality of key points to a three-dimensional coordinate system to generate a plurality of first three-dimensional coordinate points; the analysis module generates first three-dimensional image data based on the plurality of first three-dimensional coordinate points; and the analysis module makes the three-dimensional image data include the first three-dimensional image data.

In a specific embodiment, the two-dimensional image data includes second two-dimensional image data. The sarcopenia determination method 500 further includes the following steps: the analysis module identifies a plurality of key points from the second two-dimensional image data; the analysis module selects a plurality of important points from the plurality of key points; and the analysis module generates three-dimensional image data based on at least the plurality of important points.

In a specific embodiment, the sarcopenia determination method 500 further includes the following steps: the analysis module corresponds a plurality of important points to a three-dimensional coordinate system respectively to generate a plurality of second three-dimensional coordinate points; the analysis module generates second three-dimensional image data based on the plurality of second three-dimensional coordinate points; and the analysis module makes the three-dimensional image data include the second three-dimensional image data.

In a specific embodiment, the analysis module has a first deep learning model. The sarcopenia determination method 500 further includes the following steps: the analysis module uses the first deep learning model to generate three-dimensional image data according to the two-dimensional image data.

In a specific embodiment, the analysis module has a second deep learning model. The sarcopenia determination method 500 further includes the following steps: the analysis module uses the second deep learning model to generate sarcopenia determination data based on the three-dimensional image data. Preferably, the second deep learning model is a transformer neural network model.

In a specific embodiment, the sarcopenia determination system further includes a database, and the computer device of the sarcopenia determination system accesses the database, and the database stores a plurality of three-dimensional image training data. The plurality of three-dimensional image training data respectively have a data time length. The sarcopenia determination method 500 further includes the following steps: the computer device makes the data time length of each of the plurality of three-dimensional image training data the same as each other; and the computer device uses the plurality of three-dimensional image training data to train the analysis module so that the analysis module can generate sarcopenia determination data based on the three-dimensional image data.

In a specific embodiment, the plurality of three-dimensional image training data includes first training data and second training data, and the first training data includes first last image data. The first data time length of the first training data is less than the second data time length of the second training data. The sarcopenia determination method 500 further includes the following steps: a computer device generates a plurality of first image filling data according to the first last image data, and the plurality of first image filling data are all equal to the first last image data; and the computer device connects the plurality of first image filling data to the first last image data of the first training data, and makes the first training data include the plurality of first image filling data, so that the first data time length of the first training data is equal to the second data time length of the second training data.

In a specific embodiment, a plurality of three-dimensional image training data respectively have label data, and the plurality of label data respectively indicate that the corresponding three-dimensional image training data are sarcopenia training data or non-sarcopenia training data.

In a specific embodiment, the plurality of three-dimensional image training data includes second training data, the second training data has a second data time length, the three-dimensional image data includes second final image data, the three-dimensional image data has a third data time length, and the third data time length is less than the second data time length. The sarcopenia determination method 500 further includes the following steps: a computer device generates a plurality of second image filling data according to the second last image data, and the plurality of second image filling data are all equal to the second last image data; and a computer device connects the plurality of second image filling data to the second last image data of the three-dimensional image data, and makes the three-dimensional image data include the plurality of second image filling data, so that the third data time length of the three-dimensional image data is equal to the second data time length of the second training data.

So far, the sarcopenia determination system and method of the present invention have been described by the above description and drawings. However, it should be understood that the above embodiments are not only executable alone, but one or more of the above embodiments can be selectively combined and executed according to needs. In addition, the specific embodiments of the present invention are only for illustrative purposes, and various changes can be made without departing from the scope and spirit of the patent application of the present invention, and all should be included in the patent scope of the present invention. Therefore, the specific embodiments described in this specification are not used to limit the present invention, and the true scope and spirit of the present invention are disclosed in the following patent application scope.

100: Sarcopenia diagnosis system 110: Computer device 112: Receiving module 114: Analysis module 120: Image acquisition device 130: Database 210: Two-dimensional image data 212~216: Key points 311~319: Key points 321~330: Important points 410: Three-dimensional image data 412~416: Three-dimensional coordinate points 420: Three-dimensional image data 422~426: Three-dimensional coordinate points 500: Sarcopenia diagnosis method 510~540: Steps

The first figure is a system architecture diagram of a specific embodiment of the sarcopenia diagnosis system of the present invention.

The second figure is a schematic diagram of a specific embodiment of two-dimensional image data and key points.

The third figure is a schematic diagram of a specific embodiment of key points and important points.

FIG. 4A is a schematic diagram of a specific embodiment of three-dimensional image data.

FIG. 4B is a schematic diagram of a specific embodiment of three-dimensional image data.

Figure 5 is a flow chart of a specific embodiment of the sarcopenia diagnosis method of the present invention.

100: Sarcopenia diagnosis system

110:Computer device

112: Receiving module

114:Analysis module

120: Image capture device

130: Database

Claims (24)

A sarcopenia diagnosis system includes: an image capture device that captures a two-dimensional image change of a target test subject within a time period to generate a two-dimensional image data; and a computer device that is communicatively connected to the image capture device, the computer device includes: a receiving module that receives the two-dimensional image data from the image capture device; and an analysis module that is communicatively connected to the receiving module, the analysis module generates a three-dimensional image data based on the two-dimensional image data, and generates a sarcopenia diagnosis data based on the three-dimensional image data, the sarcopenia diagnosis data indicating whether the target test subject suffers from sarcopenia. The two-dimensional image data includes a second two-dimensional image data; the analysis module identifies a plurality of key points from the second two-dimensional image data; the analysis module selects a plurality of important points from the plurality of key points, and the analysis module generates the three-dimensional image data at least based on the plurality of important points; the plurality of important points correspond to the head center, upper trunk, trunk center, trunk bottom, right shoulder, left shoulder, right hip, left hip, right knee and left knee of the target tester. As in claim 1, the sarcopenia diagnosis system, wherein the two-dimensional image data indicates the image changes of at least one of the test actions of the target test subject during the time period, including balance action, one-legged standing action, in-situ stepping action, and straight-line walking action. As in claim 1, the sarcopenia diagnosis system, wherein the two-dimensional image data includes a first two-dimensional image data; wherein the analysis module identifies a plurality of key points from the first two-dimensional image data, and the analysis module generates the three-dimensional image data at least based on the plurality of key points. As in claim 3, the sarcopenia diagnosis system, wherein the analysis module maps the plurality of key points to a three-dimensional coordinate system respectively to generate a plurality of first three-dimensional coordinate points; wherein the analysis module generates a first three-dimensional image data according to the plurality of first three-dimensional coordinate points, and the analysis module causes the three-dimensional image data to include the first three-dimensional image data. As in claim 1, the sarcopenia diagnosis system, wherein the analysis module corresponds the plurality of important points to a three-dimensional coordinate system respectively to generate a plurality of second three-dimensional coordinate points; wherein the analysis module generates a second three-dimensional image data according to the plurality of second three-dimensional coordinate points, and the analysis module causes the three-dimensional image data to include the second three-dimensional image data. As in claim 1, the sarcopenia diagnosis system, wherein the analysis module has a first deep learning model, and the analysis module uses the first deep learning model to generate the three-dimensional image data based on the two-dimensional image data. As in claim 1, the sarcopenia diagnosis system, wherein the analysis module has a second deep learning model; wherein the analysis module uses the second deep learning model to generate the sarcopenia diagnosis data based on the three-dimensional image data. As in claim 7, the sarcopenia diagnosis system, wherein the second deep learning model is a transformer neural network model. The sarcopenia judgment system of claim 1 further comprises a database, wherein the database stores a plurality of three-dimensional image training data, wherein the plurality of three-dimensional image training data respectively have a data time length; wherein the computer device accesses the database, the computer device makes the data time length of each of the plurality of three-dimensional image training data the same as each other, and the computer device uses the plurality of three-dimensional image training data to train the analysis module, so that the analysis module can generate the sarcopenia judgment data according to the three-dimensional image data. The sarcopenia determination system of claim 9, wherein the plurality of three-dimensional image training data comprises a first training data and a second training data, the first training data comprises a first final image data, a first data time length of the first training data is less than a second data time length of the second training data; wherein the computer device generates a plurality of first image filling data according to the first final image data The plurality of first image filling data are all equal to the first last image data; wherein the computer device connects the plurality of first image filling data to the first last image data of the first training data, and makes the first training data include the plurality of first image filling data, so that the first data time length of the first training data is equal to the second data time length of the second training data. As in claim 9, the sarcopenia determination system, wherein the plurality of three-dimensional image training data each has a label data, and the plurality of label data each indicates that the corresponding three-dimensional image training data is a sarcopenia training data or a non-sarcopenia training data. A sarcopenia determination system as claimed in claim 9, wherein the plurality of three-dimensional image training data includes a second training data, the second training data has a second data time length; wherein the three-dimensional image data includes a second final image data, the three-dimensional image data also has a third data time length, and the third data time length is less than the second data time length; wherein the computer device generates a plurality of three-dimensional image data according to the second final image data. A plurality of second image filling data, the plurality of second image filling data are all equal to the second last image data; wherein the computer device connects the plurality of second image filling data to the second last image data of the three-dimensional image data, and makes the three-dimensional image data include the plurality of second image filling data, so that the third data time length of the three-dimensional image data is equal to the second data time length of the second training data. A sarcopenia determination method is applied to a sarcopenia determination system. The sarcopenia determination system includes an image capture device and a computer device. The computer device is communicatively connected to the image capture device. The computer device includes a receiving module and an analyzing module. The analyzing module is communicatively connected to the receiving module. The sarcopenia determination method includes the following steps: capturing an image by the image capture device; The target tester's two-dimensional image changes within a time period are used to generate two-dimensional image data; the receiving module receives the two-dimensional image data from the image capture device; the analysis module generates three-dimensional image data based on the two-dimensional image data; and the analysis module generates sarcopenia judgment data based on the three-dimensional image data, wherein the sarcopenia judgment data indicates whether the target tester suffers from sarcopenia. The two-dimensional image data includes a second two-dimensional image data; the sarcopenia determination method includes the following steps: the analysis module identifies a plurality of key points from the second two-dimensional image data; the analysis module selects a plurality of important points from the plurality of key points; and the analysis module generates the three-dimensional image data at least based on the plurality of important points; the plurality of important points correspond to the head center, upper trunk, trunk center, trunk bottom, right shoulder, left shoulder, right hip, left hip, right knee and left knee of the target tester. As in claim 13, the sarcopenia diagnosis method, wherein the two-dimensional image data indicates the image changes of at least one of the test actions of the target test subject during the time period, including balance action, one-legged standing action, in-situ stepping action, and straight-line walking action. As in claim 13, the sarcopenia determination method, wherein the two-dimensional image data includes a first two-dimensional image data; wherein the sarcopenia determination method includes the following steps: the analysis module identifies a plurality of key points from the first two-dimensional image data; and the analysis module generates the three-dimensional image data at least based on the plurality of key points. The sarcopenia determination method of claim 15 further comprises the following steps: the analysis module maps the plurality of key points to a three-dimensional coordinate system respectively to generate a plurality of first three-dimensional coordinate points; the analysis module generates a first three-dimensional image data according to the plurality of first three-dimensional coordinate points; and the analysis module causes the three-dimensional image data to include the first three-dimensional image data. The sarcopenia determination method of claim 13 further comprises the following steps: the analysis module maps the plurality of important points to a three-dimensional coordinate system to generate a plurality of second three-dimensional coordinate points; the analysis module generates a second three-dimensional image data according to the plurality of second three-dimensional coordinate points; and the analysis module causes the three-dimensional image data to include the second three-dimensional image data. As in claim 13, the sarcopenia determination method, wherein the analysis module has a first deep learning model; wherein the sarcopenia determination method comprises the following steps: the analysis module uses the first deep learning model to generate the three-dimensional image data according to the two-dimensional image data. As in claim 13, the sarcopenia determination method, wherein the analysis module has a second deep learning model; wherein the sarcopenia determination method comprises the following steps: the analysis module uses the second deep learning model to generate the sarcopenia determination data based on the three-dimensional image data. As in claim 19, the sarcopenia diagnosis method, wherein the second deep learning model is a transformer neural network model. As in claim 13, the sarcopenia determination method, wherein the sarcopenia determination system further comprises a database, the computer device accesses the database, the database stores a plurality of three-dimensional image training data, and the plurality of three-dimensional image training data respectively have a data time length; wherein the sarcopenia determination method further comprises the following steps: the computer device makes the data time length of each of the plurality of three-dimensional image training data the same as each other; and the computer device uses the plurality of three-dimensional image training data to train the analysis module so that the analysis module can generate the sarcopenia determination data according to the three-dimensional image data. The method for determining sarcopenia of claim 21, wherein the plurality of three-dimensional image training data comprises a first training data and a second training data, the first training data comprises a first final image data, a first data time length of the first training data is less than a second data time length of the second training data; wherein the method for determining sarcopenia further comprises the following steps: the computer device generates a plurality of three-dimensional image training data according to the first final image data A plurality of first image filling data, the plurality of first image filling data are all equal to the first last image data; and the computer device connects the plurality of first image filling data to the first last image data of the first training data, and makes the first training data include the plurality of first image filling data, so that the first data time length of the first training data is equal to the second data time length of the second training data. As in the sarcopenia determination method of claim 21, the plurality of three-dimensional image training data each have a label data, and the plurality of label data each indicate that the corresponding three-dimensional image training data is a sarcopenia training data or a non-sarcopenia training data. As in claim 21, the sarcopenia determination method, wherein the plurality of three-dimensional image training data includes a second training data, the second training data has a second data time length; wherein the three-dimensional image data includes a second final image data, the three-dimensional image data also has a third data time length, and the third data time length is less than the second data time length; wherein the sarcopenia determination method further includes the following steps: The computer device calculates the second training data according to the second The last image data generates a plurality of second image filling data, and the plurality of second image filling data are all equal to the second last image data; and the computer device connects the plurality of second image filling data to the second last image data of the three-dimensional image data, and makes the three-dimensional image data include the plurality of second image filling data, so that the third data time length of the three-dimensional image data is equal to the second data time length of the second training data.