A dynamic detection method and stress analysis system for pelvic stability based on dynamic DR

Abstract

This invention relates to the field of medical imaging technology, specifically to a dynamic detection method and stress analysis system for pelvic stability based on dynamic DR (Digital Radiography). The method includes acquiring static CT or MRI images of the patient's pelvis based on the user's vital signs, adapting parameters of a three-dimensional infrared motion capture system, a force platform, and pressure sensors, and establishing a pelvic coordinate system and defining key point geometric features. Under load, the method collects the user's data to be tested; after preprocessing the collected user data, it reconstructs a three-dimensional model of the user's pelvis; based on the reconstructed three-dimensional pelvic model, it performs biomechanical analysis to generate pelvic stability indices; and automatically generates a pelvic stability risk index based on the pelvic stability indices, outputting clinical decision recommendations. This addresses the problem that traditional pelvic detection methods still rely on static images, failing to effectively quantify dynamic stress distribution and stability assessment, resulting in inaccurate assessment results.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical imaging technology, in particular to a dynamic DR-based pelvic stability dynamic detection method and stress analysis system. BACKGROUND

[0002] With the rapid development of medical imaging technology, the application of orthopedic biomechanics analysis in clinical diagnosis and treatment is becoming increasingly widespread. However, traditional pelvic detection methods still rely on static images, which cannot effectively quantify dynamic stress distribution and stability evaluation, resulting in inaccurate evaluation results.

[0003] There are already some invention patents to solve the problem of dynamic evaluation of pelvic stability and stress analysis. For example:

[0004] CN112699725A discloses a stair walking dynamic body stability evaluation method. The method includes setting up a three-dimensional infrared motion capture system, building a force platform, measuring the body morphology index of the testee, pasting reflective Mark points, synchronously collecting kinematics and dynamics related indexes using the three-dimensional infrared motion capture system and the force platform, using the supporting three-dimensional infrared motion capture system processing software to perform naming, intercepting and supplementing operations on the obtained raw data, then importing the preliminarily processed data into three-dimensional virtual simulation software to establish a static model. This method introduces the stability margin index, which can more reasonably reflect the body stability under different motion states than the traditional evaluation index of body stability. However, this method still has the problem of insufficient measurement accuracy and precision of the three-dimensional infrared motion capture system in actual application.

[0005] CN118447186A proposes a non-invasive bone reconstruction prediction system. The system includes an image acquisition and preprocessing module, an optical three-dimensional motion capture module, a ground reaction force acquisition module, a musculoskeletal dynamics analysis module, a bone mechanics analysis module, and a bone reconstruction prediction module. The system first obtains the marker point position data of human body motion from the optical motion capture module, and analyzes the ground reaction force data of the human body during the body motion process from the ground reaction force acquisition module; inverse dynamics analysis is performed through the input kinematics data and ground reaction force to solve the joint force and muscle force under the corresponding motion; the stress and strain energy are obtained by simulating the bone stress through the bone stress analysis module; the morphology and density distribution of the bone within a period of time are predicted through the bone reconstruction prediction module. This system can accurately calculate the morphology and density distribution of the bone under the corresponding motion state, simulate the adjustment process of the mechanical environment on the bone mass, and predict the process and result of human bone reconstruction. However, the precision and stability of data acquisition of this system still need to be improved.

[0006] The prior art at least has the following disadvantages:

[0007] 1. The traditional pelvic detection method relies on static images, which cannot effectively quantify dynamic stress distribution and stability evaluation, resulting in inaccurate evaluation results;

[0008] 2. The existing hip joint activity evaluation method has the problems of strong subjectivity, complicated operation of objective evaluation tool and low evaluation accuracy;

[0009] 3. In the actual application of the existing pelvic stability evaluation method, the measurement accuracy and precision of the three-dimensional infrared motion capture system are insufficient, which affects the accuracy of the evaluation results;

[0010] 4. The current non-invasive bone reconstruction prediction system still needs to be improved in terms of data acquisition accuracy and stability, which affects the accuracy of bone reconstruction and mechanical analysis.

[0011] Therefore, the present application provides a pelvic stability dynamic detection method and stress analysis system based on dynamic DR to solve the above problems. SUMMARY

[0012] In view of the above situation, in order to overcome the defects of the prior art, the present application provides a pelvic stability dynamic detection method and stress analysis system based on dynamic DR, which solves the problem that the traditional pelvic detection method still relies on static images, which cannot effectively quantify dynamic stress distribution and stability evaluation, resulting in inaccurate evaluation results.

[0013] In order to achieve the above purpose, the technical scheme adopted by the present application is:

[0014] In one aspect, a dynamic detection method for pelvic stability based on dynamic DR, the dynamic detection method comprising: S100, based on the physical information of a user, adapting the parameters of a three-dimensional infrared motion capture system, a force platform and a pressure sensor to obtain static CT or MRI image data of the patient's pelvis, and establishing a pelvic coordinate system and defining key point geometric features; S200, under a weight-bearing state, collecting the user's to-be-detected data, including: synchronously collecting pelvic frontal and lateral projection dynamic images by orthogonal dual-plane DR, collecting surface marker point motion trajectory data by an infrared motion capture system, and collecting ground reaction force data by a force platform; S300, after preprocessing the collected user data, reconstructing the user's pelvic three-dimensional model, including aligning the dynamic DR projection and the static image data by an adaptive registration algorithm, preliminarily reconstructing the three-dimensional pelvic model by a deep learning algorithm, after iterative optimization deformation correction, using a generative adversarial network for detail enhancement; S400, based on the reconstructed user pelvic three-dimensional model, performing biomechanical analysis to generate pelvic stability indicators, including establishing a biomechanical model containing bones, ligaments and muscles based on the finite element method, applying load according to the ground reaction force data, and real-time calculating joint contact stress, bone displacement vector and ligament tension; S500, automatically generating a pelvic stability risk index according to the pelvic stability indicators, and outputting a clinical decision suggestion.

[0015] A further improvement of the present application is that in step S200, the collecting of the to-be-detected data of the user under a weight-bearing state specifically includes: orthogonal dual-plane DR acquisition, in the weight-bearing standing / walking state of the patient, synchronously collecting dynamic images of the user's frontal and lateral positions by an orthogonally arranged X-ray tube-detector pair, wherein the frontal position projection focuses on the pubis-sacrum midline, and the lateral position projection covers the anterior iliac crest to the ischial tuberosity; motion capture data synchronization, surface setting infrared reflective marker points on both sides of the iliac crest, L5 spinous process and pubic symphysis, tracking three-dimensional coordinates of the marker points by a multi-camera infrared system, and realizing time synchronization with DR images through a hardware trigger signal; and mechanical data acquisition, embedding a piezoelectric sensor array in the force platform to synchronously record the vertical / shear ground reaction force and the pressure center trajectory.

[0016] A further improvement of the present application is that in step S300, the preprocessing of the collected user data includes: geometric correction of the DR image based on the flat panel detector distortion model, and automatic identification of the bony landmarks: acetabular roof, ischial major notch and sacroiliac joint surface by a U-Net neural network; removing high-frequency noise of the motion data by a zero-phase Butterworth low-pass filter, and compensating for skin sliding error based on the rigid body assumption; normalizing the dynamic load of the collected mechanical signals.

[0017] A further improvement in this application is that, in step S300, the user's pelvic 3D model is reconstructed, including: multimodal registration: coarse registration of the dynamic DR projection and the CT static template using the ICP algorithm; optimization of the deformation field using the Demons differential homeomorphism algorithm; deep learning reconstruction: the input layer consists of the registered biplane DR image and the coordinates of motion capture markers, a 3D CNN network generates the initial point cloud of the pelvis, and the output layer is a voxelized model containing cortical / cancellous bone partitions; GAN detail optimization: the generator adds residual blocks to the U-Net structure, outputs high-resolution bone microstructures, the discriminator is PatchGAN to judge the local anatomical realism, and the expression for the adversarial loss function is:

[0018]

[0019] In expression (1), x represents the input data, i.e., the registered low-resolution model, y represents the real high-resolution pelvic model, i.e., the training label, z represents the random noise vector, G(x,z) represents the augmented model output by the generator, and D(·) represents the discriminator's probability output of realism, with a value range of 0-1. This represents the mathematical expectation operator.

[0020] A further improvement of this application is that, in step S400, the biomechanical analysis based on the reconstructed three-dimensional model of the user's pelvis is performed to generate pelvic stability indices, including: finite element modeling: determining the element types and material properties of bones, ligaments and muscles; dynamic load application: converting the force table GRF data into hip joint contact forces, and mapping muscle forces to finite element nodes after inverse dynamics calculation using OpenSim.

[0021] Stability index calculation:

[0022] The expression for the stress concentration factor is:

[0023] In expression (2), σ max σ represents the maximum Von Mises stress in the finite element model of the pelvis. nom This represents the average stress across the entire pelvic region.

[0024] The expression for displacement deformation is:

[0025] In expression (3), Represents the displacement vector of the sacral node. Represents the displacement vector of the pubic symphysis node;

[0026] The expression for the ligament tension warning threshold is:

[0027] T lig> 0.8 x UTS (4), in expression (4), T lig represents the real-time tension value of the ligament, and UTS represents the ultimate tensile strength of the ligament.

[0028] A further improvement of the present application is that in step S500, the multi-index weighted score model is constructed, including: determining a multi-index weighted score model, whose expression is:

[0029]

[0030] In expression (5), w1 represents a stress concentration factor weight, w2 represents a displacement deformation weight, w3 represents a ligament tension utilization rate weight, and w1+w2+w3=1 is satisfied;

[0031] Based on the multi-index weighted score model, a pelvic stability risk index is determined, and risk classification is performed, and corresponding clinical decision suggestions are given according to different risk classifications.

[0032] On the other hand, a stress analysis system of a pelvic stability dynamic detection method based on dynamic DR includes: a multi-modal acquisition module: integrating an orthogonal dual-plane DR device, an infrared motion capture system and a force platform, for synchronously acquiring dynamic images, motion trajectories and ground reaction forces; a data processing module: performing image distortion correction, motion data filtering and force signal amplification; a model reconstruction module: configured with an adaptive registration algorithm and a generative adversarial network, outputting a dynamic three-dimensional pelvic model; a biomechanical analysis module: calculating stress distribution and stability indicators based on the finite element method; a decision output module: generating a stability risk index and a clinical advice report.

[0033] A further improvement of the present application is that the multi-modal acquisition module includes: a DR synchronous control unit for controlling the synchronous acquisition of the orthogonal dual-plane DR device, ensuring that the dynamic images of the frontal position and the lateral position can be accurately corresponded; an infrared marker point layout and tracking unit responsible for laying out infrared reflective marker points at specified body surface positions, and tracking the three-dimensional coordinates of these marker points in real time using a multi-camera infrared system; a mechanical data acquisition and conversion unit embedded with a piezoelectric sensor array of the force platform, for accurately recording the ground reaction force and the center of pressure trajectory, and converting them into digital signals that can be used for subsequent analysis; the data processing module includes: an image preprocessing submodule, performing geometric correction and automatic identification of bony landmarks to improve the accuracy and reliability of the DR images; a motion data filtering and compensation submodule using a zero-phase Butterworth low-pass filter to remove high-frequency noise, and compensating for skin sliding errors based on the rigid body assumption; a mechanical signal normalization processing submodule for normalizing the collected mechanical signals to ensure the consistency and comparability of the data.

[0034] Further improvements of the present application are that the model reconstruction module adopts optimized algorithms and model parameters when performing adaptive registration and deep learning reconstruction, to improve the reconstruction accuracy and detail performance of the pelvic three-dimensional model; the detail optimization part of the generative adversarial network continuously adjusts the parameters of the generator and the discriminator, so that the output pelvic model is closer to the real anatomical structure; the biomechanical analysis module establishes a biomechanical model containing bones, ligaments and muscles based on the finite element method, and calculates joint contact stress, bone displacement vector and ligament tension and other stability indicators in real time; the decision output module considers multiple factors such as stress concentration factor, displacement deformation and ligament tension utilization rate according to a multi-index weighted scoring model, automatically generates a pelvic stability risk index, and gives corresponding clinical decision suggestions according to the risk classification.

[0035] The beneficial effects of the present application are:

[0036] 1. Improve the accuracy and objectivity of pelvic stability evaluation: Through dynamic DR technology, combined with infrared motion capture system and force platform, the image and motion data of the pelvis in dynamic state can be collected in real time, so as to effectively quantify the dynamic stress distribution and stability evaluation, and improve the accuracy and objectivity of the evaluation results.

[0037] 2. Simplify the detection process and improve the detection efficiency: The pelvic stability dynamic detection method based on dynamic DR provided by the present application simplifies the detection process through automatic and intelligent data processing and model reconstruction technology, reduces the tediousness of manual operation, and improves the detection efficiency.

[0038] 3. Provide comprehensive pelvic stability evaluation indicators: The present application calculates multiple stability indicators including stress concentration factor, displacement deformation and ligament tension through the biomechanical analysis module, and provides comprehensive pelvic stability evaluation basis for doctors.

[0039] 4. Provide strong support for clinical decision making: The present application automatically generates a pelvic stability risk index according to a multi-index weighted scoring model, and gives corresponding clinical decision suggestions according to the risk classification, which provides strong support for doctors to develop personalized treatment plans. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 The present application is a schematic flow chart of a pelvic stability dynamic detection method based on dynamic DR;

[0041] Figure 2 The present application is a schematic structural diagram of a stress analysis system of a pelvic stability dynamic detection method based on dynamic DR. DETAILED DESCRIPTION

[0042] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0043] The technical solution will be described in detail below with reference to specific embodiments.

[0044] Example 1

[0045] refer to Figure 1 A dynamic detection method for pelvic stability based on dynamic radiography (DR), the dynamic detection method comprising:

[0046] S100: Based on the user's vital signs information, adapt the parameters of the three-dimensional infrared motion capture system, force platform and pressure sensor to obtain static CT or MRI image data of the patient's pelvis, and establish a pelvic coordinate system and define the geometric features of key points.

[0047] S200: Under load, collect the user's test data, including: synchronously collect dynamic images of the pelvis in frontal and lateral positions using orthogonal dual-plane DR, collect motion trajectory data of body surface markers using an infrared motion capture system, and collect ground reaction force data using a force measuring table.

[0048] S300: After preprocessing the collected user data, the user's pelvic 3D model is reconstructed, including aligning the dynamic DR projection with the static image data through an adaptive registration algorithm, using a deep learning algorithm to initially reconstruct the 3D pelvic model, and after iterative optimization and deformation correction, using a generative adversarial network for detail enhancement.

[0049] S400 performs biomechanical analysis based on the reconstructed three-dimensional model of the user's pelvis, and generates pelvic stability indices. This includes establishing a biomechanical model containing bones, ligaments, and muscles based on the finite element method, applying loads according to ground reaction force data, and calculating joint contact stress, bone displacement vector, and ligament tension in real time.

[0050] S500: Automatically generate a pelvic stability risk index based on the pelvic stability indicators and output clinical decision recommendations.

[0051] Specifically, in step S200, collecting the user's data to be tested under load includes:

[0052] Orthogonal dual-plane DR acquisition: In the patient's weight-bearing standing / walking state, dynamic images of the user's frontal and lateral positions are acquired simultaneously through orthogonally arranged X-ray tubes and detectors. The sampling rate is ≥15 frames / second and the X-ray pulse width is ≤5ms to eliminate motion blur. The frontal projection focuses on the pubic symphysis-sacral midline, and the lateral projection covers the anterior edge of the acetabulum to the ischial tuberosity.

[0053] Motion capture data synchronization involves placing infrared reflective markers on both sides of the iliac crest of the pelvis, the L5 spinous process, and the pubic symphysis. An 8-camera infrared system (sampling rate 200Hz) is used to track the three-dimensional coordinates of the markers, and the time synchronization with the DR image is achieved through hardware trigger signals (error ≤1ms).

[0054] For mechanical data acquisition, a piezoelectric sensor array (spatial resolution ≤2mm) is embedded in the force measuring platform. The charge amplifier converts the mechanical signal into a voltage signal (range 0-2000N, accuracy ±1.5%FS), and simultaneously records the ground reaction force (GRF) in the vertical / shear direction and the trajectory of the pressure center.

[0055] In one embodiment of this application, step S300, the preprocessing of the collected user data, includes:

[0056] Geometric correction of DR images based on a flat panel detector distortion model is expressed as follows: in, k1 and k2 represent distortion coefficients, and bony landmarks are automatically identified through the U-Net neural network: acetabular roof, greater sciatic notch, and sacroiliac joint surface;

[0057] A zero-phase Butterworth low-pass filter (cutoff frequency 15Hz) is used to remove high-frequency noise from the motion data, and skin slippage error is compensated based on the rigid body assumption. The expression is as follows: α represents the soft tissue damping coefficient, and θ represents the angle between the acceleration and the normal of the marker point.

[0058] The dynamic load of the acquired mechanical signals is normalized, F norm =F raw / BW, where BW is body weight.

[0059] Specifically, in step S300, the user's pelvic 3D model is reconstructed, including:

[0060] Multimodal registration: The dynamic DR projection and the CT static template are coarsely registered using the ICP algorithm; the deformation field is optimized using the Demons differential homeomorphism algorithm.

[0061] Deep learning reconstruction: The input layer consists of the registered biplane DR image and the coordinates of motion capture markers. A 3D CNN network generates the initial point cloud of the pelvis. The output layer is a voxelized model containing cortical bone / cancellous bone partitions.

[0062] GAN detail optimization: The generator adds residual blocks to the U-Net structure to output high-resolution bone microstructures. The discriminator uses PatchGAN to judge the realism of local anatomy. The expression for the adversarial loss function is:

[0063]

[0064] In expression (1), x represents the input data, i.e., the registered low-resolution model, y represents the real high-resolution pelvic model, i.e., the training label, z represents the random noise vector, G(x,z) represents the augmented model output by the generator, and D(·) represents the discriminator's probability output of realism, with a value range of 0-1. This represents the mathematical expectation operator.

[0065] In one embodiment of this application, in step S400, the biomechanical analysis based on the reconstructed three-dimensional model of the user's pelvis to generate pelvic stability indices includes:

[0066] Finite element modeling: Determining the element types and material properties of bones, ligaments, and muscles;

[0067] Dynamic load application: The force table GRF data is converted into hip joint contact force, and the muscle force is mapped to the finite element nodes after inverse dynamic calculation using OpenSim;

[0068] Stability index calculation:

[0069] The expression for the stress concentration factor is:

[0070] In expression (2), σ max σ represents the maximum Von Mises stress in the finite element model of the pelvis. nom This represents the average stress across the entire pelvic region.

[0071] The expression for displacement deformation is:

[0072] In expression (3), Represents the displacement vector of the sacral node. Represents the displacement vector of the pubic symphysis node;

[0073] The expression for the ligament tension warning threshold is:

[0074] T lig >0.8×UTS(4), in expression (4), T lig The value represents the real-time tension of the ligament, while UTS represents the ultimate tensile strength of the ligament.

[0075] Specifically, in step S500, constructing the multi-index weighted scoring model includes:

[0076] The multi-index weighted scoring model is defined as follows:

[0077]

[0078] In expression (5), w1 represents the stress concentration factor weight, w2 represents the displacement deformation weight, and w3 represents the ligament tension utilization weight, and satisfies w1+w2+w3=1.

[0079] Based on a multi-indicator weighted scoring model, a pelvic stability risk index is determined and risk is classified. Corresponding clinical decision-making recommendations are given according to different risk classifications.

[0080] Example 2

[0081] like Figure 2 As shown, a stress analysis system for implementing a dynamic detection method for pelvic stability based on dynamic DR includes:

[0082] Multimodal acquisition module: integrates orthogonal dual-plane DR equipment, infrared motion capture system and force measuring platform, used to simultaneously acquire dynamic images, motion trajectory and ground reaction force;

[0083] Data processing module: performs image distortion correction, motion data filtering, and force signal amplification;

[0084] Model reconstruction module: Configures adaptive registration algorithm and generative adversarial network to output dynamic 3D pelvic model;

[0085] Biomechanical analysis module: Calculates stress distribution and stability indices based on the finite element method;

[0086] Decision output module: Generates stability risk index and clinical recommendation report.

[0087] Specifically, the multimodal acquisition module includes:

[0088] The DR synchronization control unit is used to control the synchronous acquisition of the orthogonal dual-plane DR equipment to ensure that the dynamic images of the front and side views can be accurately matched.

[0089] The infrared marker deployment and tracking unit is responsible for deploying infrared reflective markers at designated body surface locations and using a multi-camera infrared system to track the three-dimensional coordinates of these markers in real time.

[0090] The mechanical data acquisition and conversion unit, with a piezoelectric sensor array embedded in the force measuring platform, is used to accurately record the ground reaction force and the trajectory of the pressure center, and convert them into digital signals that can be used for subsequent analysis.

[0091] The data processing module includes:

[0092] The image preprocessing submodule performs geometric correction and automatic identification of bony landmarks to improve the accuracy and reliability of DR images;

[0093] The motion data filtering and compensation submodule uses a zero-phase Butterworth low-pass filter to remove high-frequency noise and compensates for skin sliding errors based on the rigid body assumption.

[0094] The mechanical signal normalization processing submodule normalizes the acquired mechanical signals to ensure data consistency and comparability.

[0095] In one embodiment of this application, the model reconstruction module employs optimized algorithms and model parameters when performing adaptive registration and deep learning reconstruction to improve the reconstruction accuracy and detail representation of the 3D pelvic model; the detail optimization part of the generative adversarial network continuously adjusts the parameters of the generator and discriminator to make the output pelvic model closer to the real anatomical structure.

[0096] The biomechanical analysis module establishes a biomechanical model including bone, ligaments and muscles based on the finite element method, and calculates stability indicators such as joint contact stress, bone displacement vector and ligament tension in real time.

[0097] The decision output module automatically generates a pelvic stability risk index based on a multi-index weighted scoring model, comprehensively considering factors such as stress concentration factor, displacement deformation, and ligament tension utilization rate, and provides corresponding clinical decision recommendations according to the risk level.

[0098] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0099] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0100] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0101] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0102] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0103] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0104] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0105] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A dynamic detection method of pelvic stability based on dynamic DR, characterized in that, The dynamic detection method comprises: S100, based on the user's sign information, adapt the parameters of the three-dimensional infrared motion capture system, the force platform and the pressure sensor to obtain the static CT or MRI image data of the patient's pelvis, establish the pelvis coordinate system and define the key point geometric characteristics; S200, in the weight-bearing state, collect the user's data to be detected, including: synchronously collecting the pelvis front and side projection dynamic images by orthogonal double-plane DR, collecting the surface marker point motion trajectory data by the infrared motion capture system, and collecting the ground reaction force data by the force platform; S300, after preprocessing the collected user data, reconstruct the user's pelvis three-dimensional model, including aligning the dynamic DR projection and the static image data by the self-adaptive registration algorithm, preliminarily reconstructing the three-dimensional pelvis model by the deep learning algorithm, after iterative optimization deformation correction, using the generative adversarial network for detail enhancement; S400, based on the reconstructed user pelvis three-dimensional model, perform biomechanical analysis to generate pelvis stability indicators, including establishing a biomechanical model containing bones, ligaments and muscles based on the finite element method, applying load according to the ground reaction force data, and real-time calculating joint contact stress, bone displacement vector and ligament tension; S500, automatically generate a pelvis stability risk index according to the pelvis stability indicators, and output clinical decision suggestions; In step S300, the collected user data is preprocessed, including: geometric correction of DR image based on flat panel detector distortion model, expression is: wherein, k1, k2 represent distortion coefficients, and bone landmarks are automatically identified by U-Net neural network: acetabular roof, greater sciatic notch and sacroiliac joint surface; Zero-phase Butterworth low-pass filter is used to remove the high-frequency noise of motion data, and the skin sliding error is compensated based on the rigid body assumption, which is expressed as: α represents the soft tissue damping coefficient, and θ represents the normal angle between acceleration and marker point. The dynamic load of the collected mechanical signal is normalized, F norm = F raw / BW, BW is the body weight; In step S300, the pelvis three-dimensional model of the user is reconstructed, including: Multi-modal registration: coarsely register the dynamic DR projection and the CT static template by ICP algorithm; and optimize the deformation field by Demons differential homeomorphism algorithm; Deep learning reconstruction: the input layer is the registered double-plane DR image and the motion capture marker point coordinates, the 3D CNN network generates the initial point cloud of the pelvis, and the output layer is a voxelized model containing cortical bone / cancellous bone partition; GAN detail optimization: the generator is a U-Net structure with a residual block, which outputs high-resolution bone microstructure; the discriminator is a PatchGAN that judges the local anatomical authenticity, and the expression of the adversarial loss function is: In expression (1), x represents the input data, i.e. the registered low resolution model, y represents the real high resolution pelvis model, i.e. the training label, z represents a random noise vector, G(x,z) represents the enhanced model output by the generator, D(·) represents the probability output of the discriminator on the authenticity, ranging from 0 to 1, denotes the mathematical expectation operator; In step S400, based on the reconstructed user pelvis three-dimensional model, perform biomechanical analysis to generate pelvis stability indicators, including: Finite element modeling: determine the unit type and material properties of bones, ligaments and muscles; Dynamic load application: convert the force platform GRF data into hip joint contact force, and map the muscle force to the finite element nodes after inverse dynamics calculation by OpenSim; Stability index calculation: The expression of the stress concentration factor is: In expression (2), σ max denotes the maximum Von Mises stress of the pelvic finite element model, σ nom denotes the global average stress of the pelvis; The expression of the displacement deformation amount is: In expression (3), denotes the sacral node displacement vector, denotes the pubic symphysis node displacement vector; The expression of the ligament tension early warning threshold is: T lig > 0.8 x UTS (4), in expression (4), T lig represents the ligament real-time tension value, and UTS represents the ligament ultimate tensile strength; In step S500, a multi-index weighted scoring model is constructed, including: Determine the multi-index weighted scoring model, and its expression is: ， In expression (5), w1 represents the stress concentration factor weight, w2 represents the displacement deformation amount weight, and w3 represents the ligament tension utilization rate weight, and w1+w2+w3=1 is satisfied; Based on the multi-index weighted scoring model, determine the pelvis stability risk index and perform risk classification, and give corresponding clinical decision suggestions according to different risk classifications.

2. The dynamic DR-based pelvic stability dynamic detection method according to claim 1, characterized in that, In step S200, the user's to-be-detected data is collected in a weight-bearing state, specifically including: Orthogonal dual-plane DR acquisition, in the patient's weight-bearing standing / walking state, the dynamic images of the user's front and side positions are synchronously acquired by the orthogonally arranged X-ray tube-detector pair, wherein the front position projection focuses on the pubic symphysis-sacrum midline, and the side position projection covers the front edge of the acetabulum to the ischial tuberosity; Motion capture data synchronization, infrared reflective marker points are arranged on both sides of the iliac crest, L5 spinous process and pubic symphysis, a multi-camera infrared system is used to track the three-dimensional coordinates of the marker points, and time synchronization with the DR image is realized through a hardware trigger signal; And mechanical data acquisition, a piezoelectric sensor array is embedded in the force platform to synchronously record the vertical / shear ground reaction force and the pressure center trajectory.

3. A stress analysis system for implementing the dynamic detection method of the stability of the pelvis based on dynamic DR according to any one of claims 1-2, characterized in that, Including: A multi-modal acquisition module: integrating an orthogonal dual-plane DR device, an infrared motion capture system and a force platform, for synchronously acquiring dynamic images, motion trajectories and ground reaction forces; A data processing module: performing image distortion correction, motion data filtering and force signal amplification; A model reconstruction module: configured with adaptive registration algorithms and a generative adversarial network, outputting a dynamic three-dimensional pelvic model; A biomechanical analysis module: calculating stress distribution and stability indicators based on the finite element method; A decision output module: generating a stability risk index and a clinical recommendation report.

4. The stress analysis system for dynamic detection of pelvic stability based on dynamic DR according to claim 3, characterized in that, The multi-modal acquisition module includes: A DR synchronous control unit for controlling the synchronous acquisition of the orthogonal dual-plane DR device to ensure that the dynamic images of the front and side positions can be accurately corresponded; An infrared marker point arrangement and tracking unit responsible for arranging infrared reflective marker points at specified body surface locations and using a multi-camera infrared system to track the three-dimensional coordinates of these marker points in real time; A mechanical data acquisition and conversion unit embedding a piezoelectric sensor array in the force platform for accurately recording the ground reaction force and the pressure center trajectory and converting them into digital signals that can be used for subsequent analysis; The data processing module includes: An image preprocessing submodule that performs geometric correction and automatic identification of bony landmarks to improve the accuracy and reliability of the DR image; A motion data filtering and compensation submodule that uses a zero-phase Butterworth low-pass filter to remove high-frequency noise and compensates for skin sliding errors based on the rigid body assumption; A mechanical signal normalization processing submodule that normalizes the collected mechanical signals to ensure data consistency and comparability.

5. The stress analysis system for dynamic detection of pelvic stability based on dynamic DR according to claim 3, characterized in that, When performing adaptive registration and deep learning reconstruction, the model reconstruction module uses optimized algorithms and model parameters to improve the reconstruction accuracy and detail performance of the three-dimensional pelvic model; the detail optimization part of the generative adversarial network adjusts the parameters of the generator and discriminator to make the output pelvic model closer to the real anatomical structure; The biomechanical analysis module establishes a biomechanical model containing bones, ligaments and muscles based on the finite element method, and calculates joint contact stress, bone displacement vector and ligament tension in real time; The decision output module automatically generates a pelvic stability risk index according to a multi-index weighted score model by comprehensively considering multiple factors such as stress concentration factors, displacement deformation amounts, and ligament tension utilization rates, and gives corresponding clinical decision suggestions according to risk classification.

Images