A personalized joint skeletal and cartilage morphometry system

Abstract

A personalized joint skeletal and cartilage morphometry system is provided that uses artificial intelligence. [Solution] The personalized joint skeleton and cartilage morphometry system includes a database 11, an image acquisition module 12, an image analysis module 13, a neural network model 14, a model construction module 15 and a calculation module 16. The database stores a dataset. The image acquisition module acquires a plurality of skeleton images of a subject. The image analysis module processes the skeleton images by a statistical shape model and principal component analysis to generate a 3D model of the skeleton and cartilage. The neural network model generates and outputs skeleton motion parameters. The model construction module generates the personalized joint skeleton and cartilage model. The calculation module simulates the relative positions between the 3D models of the skeleton and cartilage during motion and calculates the cartilage shape and cartilage wear value.

Description

[Technical field]

[0001] The present invention relates to a joint bone and cartilage measurement system, and more particularly to a personalized joint bone and cartilage morphology measurement system that applies existing artificial intelligence technology. [Background technology]

[0002] Cartilage is a type of connective tissue that plays a vital role in human joints, reducing shock and friction during joint movement and ensuring smooth joint operation. However, cartilage is easily damaged by aging, overuse of the joint, or trauma, which may lead to osteoarthritis (OA). Osteoarthritis, also known as degenerative joint disease, is a type of arthritis in which the articular cartilage and bones below the joint are damaged, and the most common sites are the distal joints of the fingers, the base of the thumb, the neck, the lower back, the knee, and the hip joint. Among them, knee osteoarthritis (Knee Osteoarthritis, Knee OA) is the most common. In this disease, inflammation and damage to the joints cause bone changes, degeneration of tendons and ligaments, and cartilage wear, resulting in joint pain, swelling, deformation, joint dysfunction, and even a decrease in quality of life. Therefore, early diagnosis and evaluation of treatment effects are crucial for targeted treatment, and detecting cartilage changes is a key factor.

[0003] Traditionally, the detection of cartilage changes mainly relies on traditional clinical evaluation and imaging diagnostic methods, such as X-ray and MRI. Currently, the most commonly used classification method for osteoarthritis is the Kellgren-Lawrence Grading System. This system classifies knee osteoarthritis into five stages based on the degree of joint space reduction (indicating the degree of cartilage wear) and the degree of osteophyte formation based on upright knee joint X-ray images. However, X-ray and MRI detection methods have some limitations. For example, X-ray cannot obtain actual cartilage information and can only estimate the severity of knee joint osteoarthritis based on the joint space, which may lead to differences in evaluation between doctors. On the other hand, MRI has limitations in image quality and makes it difficult to accurately identify the different positions of damaged and normal cartilage. In addition, all of these methods can only obtain information in a stationary state, require expensive and specialized equipment, and are also dependent on the operator's skills.

[0004] Therefore, in order to solve the above-mentioned problems, it is necessary to develop a new joint skeleton and cartilage measurement system. Summary of the Invention [Problem to be solved by the invention]

[0005] The present invention provides a personalized joint skeletal and cartilage morphometry system to solve the above-mentioned problems of the prior art. [Means for solving the problem]

[0006] The present invention provides a personalized joint skeleton and cartilage morphometry system for constructing a personalized joint skeleton and cartilage model of a subject and calculating the cartilage wear value of the subject based on the model. The measurement system includes a database, an image acquisition module, an image analysis module, a neural network model, a model construction module, and a calculation module. The database stores a dataset, which includes a plurality of healthy human joint skeleton and cartilage images, corresponding skeleton and cartilage 3D models, and a plurality of shape parameters. The image acquisition module acquires a plurality of skeleton images of the subject, each of which includes a first skeleton image and a second skeleton image.

[0007] The image analysis module is connected to the image acquisition module and the database, and processes the subject's skeletal image through a statistical shape model trained using the dataset to generate a 3D model of the skeleton and cartilage. Further, a Principal Component Analysis (PCA) is used to obtain a plurality of first skeletal and cartilage shape parameters corresponding to the first skeletal image and a plurality of second skeletal and cartilage shape parameters corresponding to the second skeletal image from the skeletal and cartilage 3D model and shape parameters. Then, a first skeletal and cartilage 3D model and a second skeletal and cartilage 3D model are generated based on the first and second skeletal and cartilage shape parameters.

[0008] The neural network model is connected to the image analysis module, and receives the first and second skeletal and cartilage shape parameters as input, and generates and outputs the first and second skeletal motion parameters based on a machine learning method learned by the neural network model, the neural network model being trained based on the dataset.

[0009] The model construction module is connected to the image analysis module and the neural network model, and generates a personalized articular skeleton and cartilage model based on the first skeleton and cartilage 3D model, the second skeleton and cartilage 3D model, and the first and second skeleton motion parameters.

[0010] The calculation module is connected to the model construction module, and based on the personalized joint skeleton and cartilage model, simulates the relative positions of the first skeleton and cartilage 3D model and the second skeleton and cartilage 3D model while the subject is moving, and further removes the areas where the cartilages interfere with each other in the first and second skeleton and cartilage 3D models during the movement, and calculates the actual cartilage shape and cartilage wear value of the subject.

[0011] The personalized joint skeletal and cartilage morphometry system further includes an image matching module, which is connected to the image analysis module and the database. The image matching module adjusts the angle and position of the first skeletal and cartilage 3D model and the second skeletal and cartilage 3D model, projects them, and uses a single-plane dynamic image matching method to compare the joint skeletal and cartilage images of healthy people in the dataset to calculate the first skeletal motion parameters and the second skeletal motion parameters of the subject.

[0012] Each skeletal image of the subject also includes an X-ray image.

[0013] The first and second skeletal motion parameters include translation parameters and rotation parameters in three directions of the X, Y and Z axes.

[0014] The personalized articular skeletal and cartilage models are visually displayed using color-matching and perspective.

[0015] In summary, the present invention provides a personalized joint skeleton and cartilage morphology measurement system, which combines dynamic X-ray technology with artificial intelligence to realize a technology for reconstructing joint skeleton and cartilage morphology. The system can reconstruct the patient's knee joint skeleton and cartilage model in more detail and accurately calculate the degree and location of cartilage changes. In addition, the personalized joint skeleton and cartilage model of the present invention can be visualized using color correspondence and perspective, providing an intuitive means of understanding the cartilage condition and allowing for a faster and more comprehensive evaluation of the health of the cartilage. The artificial intelligence-based planar X-ray measurement system for joint skeleton and cartilage 3D morphology based on the present invention provides doctors with dynamic and more precise medical information, enabling them to develop personalized treatment plans and evaluate treatment effects for each patient. [Brief description of the drawings]

[0016] [Figure 1] FIG. 1 shows a functional block diagram of a personalized joint skeletal and cartilage morphometry system according to one embodiment of the present invention. [Diagram 2] FIG. 2 shows a procedural flow diagram of a personalized joint skeletal and cartilage morphometry method according to one embodiment of the present invention. [Diagram 3] FIG. 3 shows a functional block diagram of a personalized joint skeletal and cartilage morphometry system according to another embodiment of the present invention. [Figure 4] FIG. 4 shows a procedural flow diagram of a personalized joint skeletal and cartilage morphometry method according to another embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] In order to make the advantages, spirit and features of the present invention easier and clearer to understand, specific embodiments will be described and discussed in detail below with reference to the accompanying drawings. However, it should be noted that these specific embodiments are shown as representative examples of the present invention, and the specific methods, devices, conditions, materials, etc. shown do not limit the present invention or the corresponding embodiments. In addition, the elements in the drawings are shown only to show their relative positions, and are not drawn based on actual proportions. It should be noted in advance that the step numbers of the present invention are used to distinguish different steps, and do not necessarily indicate the order of steps.

[0018] Please refer to Fig. 1. Fig. 1 shows a functional block diagram of a personalized joint skeletal and cartilage morphometry system 1 according to an embodiment of the present invention. As shown in Fig. 1, the personalized joint skeletal and cartilage morphometry system 1 of this embodiment includes a database 11, an image acquisition module 12, an image analysis module 13, a neural network model 14, a model construction module 15, and a calculation module 16.

[0019] The database 11 is for storing a dataset, the dataset including a plurality of healthy human joint skeleton and cartilage images, corresponding skeleton and cartilage 3D models, and a plurality of shape parameters. The image acquisition module 12 is for acquiring a plurality of skeleton images of a subject, each skeleton image including a first skeleton image and a second skeleton image.

[0020] The image analysis module 13 is connected to the image acquisition module 12 and the database 11, and processes the subject's skeletal image using a statistical shape model trained based on the data set to generate a 3D model of the skeleton and cartilage. In addition, the image analysis module 13 uses Principal Components Analysis (PCA) to obtain a plurality of first skeleton and cartilage shape parameters corresponding to the first skeleton image and a plurality of second skeleton and cartilage shape parameters corresponding to the second skeleton image from the 3D model and shape parameters of the skeleton and cartilage. Based on these first and second skeleton and cartilage shape parameters, the image analysis module 13 generates a first skeleton and cartilage 3D model and a second skeleton and cartilage 3D model.

[0021] The neural network model 14 is connected to the image analysis module 13, and receives the first and second skeletal and cartilage shape parameters as input, and generates and outputs the first and second skeletal motion parameters based on the first and second skeletal shape parameters. This neural network model 14 is trained by machine learning based on the data set.

[0022] The model construction module 15 is connected to the image analysis module 13 and the neural network model 14, and generates a personalized joint skeleton and cartilage model based on the first skeleton and cartilage 3D model, the second skeleton and cartilage 3D model, and the first and second skeleton motion parameters.

[0023] The calculation module 16 is connected to the model construction module 15, and simulates the relative positions of the first skeleton and cartilage 3D model and the second skeleton and cartilage 3D model while the subject is moving based on the personalized joint skeleton and cartilage model, and removes the areas where the cartilages interfere with each other in the first and second skeleton and cartilage 3D models during the movement, and calculates the actual cartilage shape and cartilage wear value of the subject.

[0024] In this embodiment, each skeletal image of the subject further includes an X-ray image. The first skeletal motion parameter and the second skeletal motion parameter include translation parameters and rotation parameters in three directions of the X, Y and Z axes.

[0025] Please refer to Figures 1 and 2. Figure 2 shows a procedure flow diagram of a personalized joint skeleton and cartilage morphometry method according to one specific embodiment of the present invention. Each procedure of the personalized joint skeleton and cartilage morphometry method shown in Figure 2 is performed using the personalized joint skeleton and cartilage morphometry system 1 of Figure 1. As shown in Figure 2, the personalized joint skeleton and cartilage morphometry method in this specific embodiment first stores a dataset in a database 11, the dataset including a plurality of healthy human joint skeleton and cartilage images, corresponding 3D models of the skeleton and cartilage, and a plurality of shape parameters. Then, an image acquisition module 12 is used to acquire a plurality of skeletal images of a subject, each skeletal image including a first skeletal image and a second skeletal image. Furthermore, an image analysis module 13 is used to process the skeletal images of the subject using a statistical shape model learned based on the dataset to generate a 3D model of the skeleton and cartilage. Then, a plurality of first skeleton and cartilage shape parameters corresponding to the first skeleton image and a plurality of second skeleton and cartilage shape parameters corresponding to the second skeleton image are obtained from the skeleton and cartilage 3D model and shape parameters by using Principal Component Analysis (PCA), and a first skeleton and cartilage 3D model and a second skeleton and cartilage 3D model are generated based on these first and second skeleton and cartilage shape parameters. Next, the first and second skeleton and cartilage shape parameters are input to a neural network model 14, which generates and outputs first skeleton motion parameters and second skeleton motion parameters based on these shape parameters. This neural network model 14 is trained by machine learning based on the dataset. Then, a model construction module 15 is used to generate a personalized joint skeleton and cartilage model based on the first skeleton and cartilage 3D model, the second skeleton and cartilage 3D model, the first skeleton motion parameters, and the second skeletal motion parameters. Finally, a calculation module 16 is used to simulate the relative positions of the first skeleton and cartilage 3D model and the second skeleton and cartilage 3D model while the subject is moving based on the personalized joint skeleton and cartilage model.Furthermore, the regions where the cartilage interferes with each other in the first and second skeletal and cartilage 3D models during exercise are removed, and the actual cartilage shape and cartilage wear value of the subject are calculated. In this way, this measurement method enables accurate and efficient evaluation of the condition of the joint skeleton and cartilage of the subject.

[0026] Furthermore, the personalized joint skeleton and cartilage model can be generated by adjusting the model shape parameters based on the characteristics of the joint by combining a statistical shape model with principal component analysis (PCA). In practical applications, the personalized joint skeleton and cartilage morphometry system of this specific embodiment can be used to reconstruct a personalized knee joint model. In practice, after completing the construction of a database of knee joint sample models, the model samples from this database can be used as training models to calculate the average model and construct shape variations. The shape variations can be obtained through principal component analysis (PCA), and the eigenvalue (eigenvalue λ i ) and eigenvector (φ i ) is obtained. The new knee joint model is then expressed by adding the model variation to the average model.

[0027]

number

[0028] where b i represents the shape model parameters that belong to that model.

[0029] In practical applications, a skeletal 3D model of the joint can be constructed using artificial intelligence, and a kinematic calculation method combined with dynamic X-rays can be used to obtain the distribution of skeletal distances between the femur and tibia at each time point of the passive flexion and extension of the joint. Furthermore, these skeletal distance distributions can be compared with a cartilage 3D model previously constructed using artificial intelligence, and its thickness can be evaluated to calculate the perforation area, degree of perforation, and perforation volume of the cartilage, and to evaluate the distribution of the cartilage wear status and its severity. In practice, personalized joint skeleton and cartilage models can also be visualized using color correspondence diagrams and perspective methods. This method is expected to not only enable a more comprehensive evaluation of the health status of articular cartilage, but also to utilize the detection of cartilage changes as a new method for evaluating the effectiveness of treatment.

[0030] Moreover, in practice, databases play an important role in machine learning. The database provides basic data required for training, evaluation, and improvement of machine learning models, and allows the machine learning models to learn using data in the database and adjust parameters to improve their performance and accuracy. In this specific embodiment, the database stores data including skeletal images of multiple joints, multiple skeletal 3D models corresponding to the skeletal images, multiple position parameters of the joints, and multiple cartilage 3D models.

[0031] In practical applications, the method for constructing a personalized joint skeleton and cartilage model is carried out by using dynamic X-ray images of different planes and adjusting the model shape parameters using artificial intelligence technology and machine learning. The skeleton part in the reconstructed joint 3D model is compared for similarity with the dynamic X-ray images using an image matching method, and the optimal model shape parameters are identified, so that a personalized joint 3D model is reconstructed. This image matching method is mainly a process of applying geometric transformations to two or more images to spatially align or match them. In addition, by comparing and matching an object in 3D space with a 2D image, it is possible to determine the exact position that the object should have in real space. This includes a method of using operations such as translation, rotation, and zooming of the image to match or overlap specific features or areas between images.

[0032] In this specific embodiment, the personalized joint skeleton and cartilage model further includes a distal femur, a proximal tibia, and cartilage, which can be further determined using image matching to determine the relative positions of the distal femur, the proximal tibia, and the cartilage to obtain an accurate knee joint model.

[0033] Please refer to FIG. 3. FIG. 3 shows a functional block diagram of a personalized joint skeleton and cartilage morphometry system 2 according to another embodiment of the present invention. As shown in FIG. 3, this embodiment is characterized in that it further includes an image counterpoint module 21, which is different from the above embodiment. In this embodiment, the image counterpoint module 21 is connected to the image analysis module 13 and the model construction module 15, and has the function of adjusting the angle and position of the first skeleton and cartilage 3D model and the second skeleton and cartilage 3D model, and projecting them. In addition, the first skeleton motion parameter and the second skeleton motion parameter of the subject are calculated by comparing with the joint skeleton and cartilage image of a healthy person in the data set using a single-plane dynamic image matching method.

[0034] Please refer to Figures 3 and 4. Figure 4 shows a procedure flow diagram of a personalized joint skeleton and cartilage morphometry method based on another specific embodiment of the present invention. The procedure of the personalized joint skeleton and cartilage morphometry method shown in Figure 4 is carried out using the personalized joint skeleton and cartilage morphometry system 2 of Figure 3. This specific embodiment is characterized in that it further includes the following steps, unlike the above-mentioned specific embodiment.

[0035] In this specific embodiment, the following steps are performed after step S3: in step S41, the image counterpoint module 21 is used to adjust the angle and position of the first skeleton and cartilage 3D model and the second skeleton and cartilage 3D model, and project them. Then, the single-plane dynamic image matching method is used to compare with the joint skeleton and cartilage images of a healthy person in the dataset, and the first skeleton motion parameters and the second skeleton motion parameters of the subject are calculated. In practical applications, the database can further include a knee joint sample model database. This database can be used to build personalized knee joint models of healthy people and patients through computed tomography (CT scan) and used as a data source for machine learning. The personalized knee joint model can further include a distal femur, a proximal tibia, and cartilage. First, a CT scan is performed with a suitable joint scanning range (ensuring that the knee joint is located in the center of the image), and the skeletal CT images acquired by the image acquisition module are imported into the visualization software to screen the images. Then, a 3D skeletal solid model is constructed using a 3D modeling module. The 3D skeletal solid model is aligned with the dynamic X-ray image using an image matching method to calculate the 3D kinematic information of the knee joint. This information includes the translation and rotation in three directions of the X, Y, and Z axes.

[0036] In practice, the image alignment method can include a single-plane image matching technique using a graphical user interface (GUI) developed in Matlab. The GUI allows the user to adjust the bones' spatial positions and initially align them to the dynamic radiographs. Based on these initial positions, an initial guess is set for the six degrees of freedom (6DOF) of the 3D skeletal model. The dynamic radiograph parameters and distortion correction parameters obtained during system calibration are combined to generate Digitally Reconstructed Radiographs (DRR). The position of the six DOFs in the model's spatial position is then adjusted, and new DRR images (I DRR ) to generate dynamic X-ray images (I fl ) is calculated. The process uses optimization techniques to determine the best position of the bones in space.

[0037] A vectorized genetic algorithm can be used as the optimization method. The gradient difference (GD) method is used as the similarity index. In this method, I DRR and I fl Apply horizontal and vertical Sobel operators to the four gradient images dI DRR / di, dI DRR / dj, dI fl / di, dI fl Then, the difference between the DRR gradient image and the dynamic X-ray gradient image is calculated to generate the vertical and horizontal gradient difference image I diffV and I diffH This difference-based measurement is given by:

[0038]

number

[0039] It should be noted that in practical applications, the data in the database and the acquisition method are not limited to those described above and can be selected according to the needs of the user. Other modules and their functions included in the personalized joint skeleton and cartilage morphometry system 2 of this specific embodiment are generally the same as the corresponding modules in the above specific embodiment, so detailed description is omitted here. This specific embodiment builds a training model based on joint images and data acquired by CT scan, and reconstructs the patient's knee joint skeleton and cartilage model using machine learning techniques by combining dynamic X-ray technology and artificial intelligence. This process allows the degree and location of cartilage changes to be more accurately grasped. In addition, this specific embodiment provides actual cartilage information and accurate position information that cannot be obtained by conventional X-ray and MRI methods, making it possible to obtain dynamic joint information. In addition, it is possible to reduce time and labor costs without relying on specialized equipment or operator skills.

[0040] In summary, the present invention provides a personalized joint skeleton and cartilage morphology measurement system. The system combines dynamic X-ray technology with artificial intelligence to realize a technique for reconstructing joint skeleton and cartilage morphology, and reconstructs the patient's knee joint skeleton and cartilage model in more detail. It can also accurately calculate the degree and location of cartilage changes. In addition, the personalized joint skeleton and cartilage model of the present invention is visualized using color correspondence and perspective, providing an intuitive understanding of the cartilage condition and enabling a rapid and comprehensive evaluation of the health of the cartilage. The artificial intelligence-based joint skeleton and cartilage 3D morphology planar X-ray measurement system of the present invention provides doctors with dynamic and more precise medical information, enabling them to formulate personalized treatment plans and evaluate treatment effects for each patient.

[0041] The above specific examples are described in detail to more clearly explain the features and spirit of the present invention, and are not intended to limit the scope of the present invention. Rather, the scope of the present invention is intended to include various modifications and equivalent configurations contemplated within the scope of the utility model registration claims of the present invention. Therefore, the interpretation of the utility model registration claims of the present invention should be given the broadest interpretation based on the above description, and should be interpreted to include all possible modifications and equivalent configurations. [Explanation of symbols]

[0042] 1, 2 Personalized Joint Skeletal and Cartilage Morphometry System 11 Database 12 Image Acquisition Module 13 Image Analysis Module 14 Neural Network Model 15 Model Building Module 16 Computational Module 21 Image Alignment Module S1~S6, S41 Procedure

Claims

1. A personalized joint skeleton and cartilage morphometry system for constructing a personalized joint skeleton and cartilage model of a subject and calculating a cartilage wear value of the subject based on the model, the system comprising: a database for storing a dataset, the dataset including a plurality of healthy human joint skeleton and cartilage images, corresponding skeleton and cartilage 3D models, and a plurality of shape parameters; an image acquisition module for acquiring a plurality of skeleton images of the subject, each skeleton image including a first skeleton image and a second skeleton image, connected to the image acquisition module and the database; processing the subject's skeleton images using a statistical shape model learned based on the dataset to generate the skeleton and cartilage 3D models; and performing a principal component analysis (PCA) on the skeleton images.The present invention further comprises an image analysis module that uses a PCA (Computer-Based Algorithm) to obtain a plurality of first skeletal and cartilage shape parameters corresponding to a first skeletal image and a plurality of second skeletal and cartilage shape parameters corresponding to a second skeletal image from the skeleton and cartilage 3D model and shape parameters, and generates a first skeleton and cartilage 3D model and a second skeleton and cartilage 3D model based on the first and second skeletal and cartilage shape parameters, and a neural network model that is connected to the image analysis module and receives the first and second skeletal and cartilage shape parameters as input, and generates and outputs a plurality of first skeletal motion parameters and a plurality of second skeletal motion parameters corresponding to the first and second skeletal and cartilage shape parameters, the neural network model being generated by machine learning based on a dataset. A personalized joint skeleton and cartilage morphometry system, comprising: a model construction module, which is trained and connected to the image analysis module and the neural network model, and which generates a personalized joint skeleton and cartilage model based on a first skeleton and cartilage 3D model, a second skeleton and cartilage 3D model, and the first and second skeleton motion parameters; and a calculation module, which is connected to the model construction module, and which simulates the relative positions of the first skeleton and cartilage 3D model and the second skeleton and cartilage 3D model while the subject is moving based on the personalized joint skeleton and cartilage model, removes areas of cartilage that interfere with each other in the first and second skeleton and cartilage 3D models during movement, and calculates the actual cartilage shape and cartilage wear value of the subject based on the personalized joint skeleton and cartilage model.

2. The personalized joint skeleton and cartilage morphometry system of claim 1, further comprising an image matching module connected to the image analysis module and the model construction module for adjusting the angle and position of the first skeleton and cartilage 3D model and the second skeleton and cartilage 3D model and projecting them, and using a single-plane dynamic image matching method to compare with joint skeleton and cartilage images of a healthy person in the dataset and calculate the first skeletal motion parameters and the second skeletal motion parameters of the subject.

3. 2. The personalized joint skeletal and cartilage morphometry system of claim 1, wherein each skeletal image of the subject further comprises an X-ray image.

4. 2. The personalized joint skeletal and cartilage morphometry system of claim 1, wherein the first skeletal motion parameters and the second skeletal motion parameters further include translation parameters and rotation parameters in three directions of the X, Y and Z axes.

5. 2. The personalized joint skeletal and cartilage morphometry system of claim 1, wherein the personalized joint skeletal and cartilage model is further visualized using color correspondence and perspective.

Images