An Automatic Measurement Method for Distal Femoral Anatomical Parameters Based on a 3D CT Model of the Knee Joint

Abstract

This invention discloses an automatic measurement method for distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint. The method includes: acquiring knee CT medical image data and preprocessing it; automatically segmenting the bone structures in the knee CT image data using a deep learning segmentation network to obtain a three-dimensional segmentation result of the femur; performing three-dimensional reconstruction based on the three-dimensional segmentation result to generate a three-dimensional bone structure model of the femur; establishing a unified three-dimensional spatial coordinate system for the knee joint; performing anatomical feature analysis on the three-dimensional bone structure model of the femur; extracting key anatomical feature points of the medial condyle, lateral condyle, and patellar surface region of the femur; determining relevant anatomical axes; calculating knee joint anatomical parameters; and outputting the anatomical parameters. This invention enables automatic analysis of knee joint bone structures, improves the efficiency and accuracy of parameter measurement, reduces manual intervention, and can be used for preoperative planning of the knee joint, distal femoral anatomical assessment, and orthopedic digital image analysis.

Description

Technical Field

[0001] This invention relates to the fields of medical image processing and computer vision technology, and in particular to an automatic measurement method for distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint. Background Technology

[0002] The knee joint is one of the most important weight-bearing joints in the human body, playing a vital role in sports and daily activities. With the increasing aging of the population, the incidence of degenerative diseases such as knee osteoarthritis is rising year by year. For patients with severe knee joint disease, total knee arthroplasty (TKA) is a common and effective treatment method.

[0003] During knee replacement surgery, surgeons typically need to select an appropriately sized prosthesis based on the patient's knee joint anatomy and perform preoperative planning. Traditional preoperative planning methods rely primarily on two-dimensional medical imaging and the surgeon's experience for measurement and judgment. This approach suffers from problems such as low measurement efficiency, large human error, and difficulty in fully considering individual differences.

[0004] With the development of medical imaging technology, 3D medical images such as CT scans can provide more accurate bone structure information, laying the foundation for 3D modeling and anatomical parameter measurement of the knee joint. However, the existing 3D modeling and parameter measurement processes often require a lot of manual operation, such as manually segmenting bone structures, manually marking anatomical points, and manually measuring anatomical distances, which is not only inefficient but also results in poor stability and repeatability.

[0005] In recent years, deep learning technology has made significant progress in the field of medical image segmentation. By training neural network models, it is possible to automatically segment organs or tissue structures in medical images, providing a new technical means for the automatic analysis of three-dimensional medical images.

[0006] Therefore, utilizing medical image automatic segmentation technology and three-dimensional geometric analysis methods to achieve automatic extraction of knee joint bone structure and automatic measurement of anatomical parameters is of great significance for improving preoperative planning efficiency, assisting in personalized prosthesis design, and enhancing surgical precision. Summary of the Invention

[0007] In view of this, the purpose of this invention is to provide an automatic measurement method for distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint, which can solve the problems of manual operation, low efficiency, and poor stability of measurement results in the existing technology for knee joint bone structure segmentation and anatomical parameter measurement.

[0008] The present invention solves the technical problem by adopting the following technical solution:

[0009] An automatic measurement method for distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint includes the following steps:

[0010] Step 1: Acquire knee joint CT medical image data and preprocess the image data;

[0011] Step 2: Use a deep learning segmentation network to automatically segment the bone structure in the knee joint CT image data to obtain the three-dimensional segmentation result of the femur.

[0012] Step 3: Perform three-dimensional reconstruction based on the three-dimensional segmentation results to generate a three-dimensional bone structure model of the femur;

[0013] Step 4: Establish a unified three-dimensional coordinate system for the knee joint;

[0014] Step 5: Perform anatomical feature analysis on the three-dimensional bone structure model of the femur in a three-dimensional spatial coordinate system, extract key anatomical feature points of the medial condyle, lateral condyle and patellar surface region of the femur, and determine the relevant anatomical axes;

[0015] Step 6: Calculate the anatomical parameters of the knee joint based on key anatomical feature points and anatomical axes;

[0016] Step 7: Output anatomical parameters to provide a basis for personalized knee prosthesis design or preoperative planning for total knee replacement surgery.

[0017] Furthermore, in step 1, the method for preprocessing the image data is as follows:

[0018] Represent the raw data of knee joint CT medical images as a three-dimensional grayscale volume. voxel coordinates In this stage, image preprocessing is performed on the original volumetric data, and grayscale values ​​are truncated to reduce the impact of abnormal noise, resulting in preprocessed volumetric data. This is used for subsequent automatic segmentation of bone structures.

[0019] Furthermore, the deep learning segmentation network in step 2 is a medical image segmentation network based on the U-Net structure, which extracts multi-scale features through the encoder and decoder structure to achieve automatic segmentation of the femoral structure.

[0020] Furthermore, in step 4, the method for establishing a unified three-dimensional spatial coordinate system for the knee joint includes:

[0021] First, based on the femoral point cloud data, a principal orientation analysis is performed to determine the proximal and distal orientation of the femur, and this orientation is defined as... First, along the axial direction; second, extract the medial and lateral extreme points in the distal femur region and define the medial and lateral direction vectors:

[0022]

[0023] in, and These represent the extreme points on the medial and lateral sides of the femur, respectively.

[0024] Determining the forward and backward directions using the cross product relationship: .

[0025] Furthermore, in step 5, the method for analyzing the anatomical features of the three-dimensional femoral bone structure model in a three-dimensional spatial coordinate system includes:

[0026] First, a point set C of the distal condylar region is extracted by truncating the femur along the longitudinal axis. Second, the point set C is divided into medial condylar regions using a clustering algorithm. and lateral condylar region Subsequently, extreme value searches were performed before and after each region:

[0027]

[0028]

[0029] , These are the anterior and posterior points of the medial condyle, respectively; , These are respectively used as the anterior and posterior points of the lateral condyle;

[0030] The deepest point of the patellar surface is determined in the femoral patellar region using curvature analysis or geometric depression detection methods. .

[0031] Furthermore, the knee joint anatomical parameters in step 6 specifically include: the anteroposterior diameter of the medial femoral condyle; the anteroposterior diameter of the lateral femoral condyle; the width of the medial and lateral femoral condyles; the posterior condyle tangent; the condylar line; and patellar surface related distance parameters.

[0032] The present invention discloses an automatic measurement method for distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint, which has the following beneficial effects:

[0033] (1) Automatic segmentation of knee joint bone structure to improve image processing efficiency. This invention uses a deep learning segmentation network to automatically identify and segment bone structures in knee CT images, accurately extracting the three-dimensional structural information of the femur. This avoids the problem of relying on manual layer-by-layer delineation of bone structure contours in traditional methods, greatly reducing manual intervention, improving image processing efficiency, and improving the consistency and stability of segmentation results.

[0034] (2) Achieve automatic reconstruction of the three-dimensional bone structure of the knee joint and improve the accuracy of structural representation. This invention constructs a three-dimensional bone structure model of the femur by reconstructing the segmentation results in three dimensions, so that the spatial morphology of the bone structure can be fully expressed. Compared with traditional two-dimensional image measurement methods, the three-dimensional model can more realistically reflect the spatial structural features of the bone, thus providing a reliable basis for the accurate measurement of subsequent anatomical parameters.

[0035] (3) Automatic extraction of anatomical features to improve the accuracy of parameter measurement. Under a unified three-dimensional coordinate system, this invention automatically extracts key anatomical feature points of the medial condyle, lateral condyle and patellar surface region by performing geometric feature analysis on key anatomical regions of the femur, and constructs corresponding anatomical axes, thereby realizing automatic calculation of anatomical parameters, reducing subjective errors caused by manual measurement, and improving the objectivity and repeatability of measurement results. Attached Figure Description

[0036] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] To address the problems of low efficiency and unstable measurement accuracy in existing knee joint 3D modeling and anatomical parameter measurement processes that rely on manual operation, this invention proposes a knee joint bone structure analysis method based on automatic medical image processing. This method enables automatic analysis of knee joint bone structure, improves parameter measurement efficiency and accuracy, reduces manual intervention, and can be used for preoperative planning of the knee joint, distal femoral anatomy assessment, and orthopedic digital image analysis.

[0039] refer to Figure 1 This invention provides an automatic measurement method for distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint, comprising the following steps:

[0040] Step 1: Acquire knee joint CT medical image data and preprocess the image data;

[0041] Specifically, the subject's knee joint CT volume data is acquired, which can be a DICOM sequence or an NIfTI format file. The raw volume data of the knee joint CT medical image is then represented as a three-dimensional grayscale volume. voxel coordinates In this stage, image preprocessing is performed on the original volumetric data, and grayscale values ​​are truncated to reduce the impact of abnormal noise, resulting in preprocessed volumetric data. This is used for subsequent automatic segmentation of bone structures.

[0042] Step 2: The bone structure in the knee CT image data is automatically segmented using a deep learning segmentation network to obtain the three-dimensional segmentation result of the femur. The deep learning segmentation network is a medical image segmentation network based on the U-Net structure. It extracts multi-scale features through the encoder and decoder structure to achieve automatic segmentation of the femur structure.

[0043] Specifically, the preprocessed volume data is input into a deep learning segmentation network for inference calculation to obtain a probability map of the femur:

[0044]

[0045] in, For a well-trained segmentation model, The network parameters are then used. The segmentation results are subsequently post-processed, including maximum connected component filtering, hole filling, and morphological smoothing, to obtain complete and continuous femoral segmentation results.

[0046] The voxel spacing was then obtained from the CT header file. Map it to real physical coordinates. ,in This represents element-wise multiplication. This transformation yields a three-dimensional physical model of the femur, providing a basis for subsequent anatomical parameter measurements.

[0047] Step 3: After obtaining the three-dimensional physical space model of the femur, the three-dimensional isosurface extraction algorithm is used to reconstruct the surface of the bone structure to generate a three-dimensional bone structure model of the femur.

[0048] Step 4: Establish a unified three-dimensional coordinate system for the knee joint. Methods include:

[0049] First, based on the femoral point cloud data, a principal orientation analysis is performed to determine the proximal and distal orientation of the femur, and this orientation is defined as... First, along the axial direction; second, extract the medial and lateral extreme points in the distal femur region and define the medial and lateral direction vectors:

[0050]

[0051] in, and These represent the extreme points on the medial and lateral sides of the femur, respectively.

[0052] Determining the forward and backward directions using the cross product relationship: .

[0053] Step 5: Perform anatomical feature analysis on the three-dimensional bone structure model of the femur in a three-dimensional spatial coordinate system, extract key anatomical feature points of the medial condyle, lateral condyle and patellar surface region of the femur, and determine the relevant anatomical axes;

[0054] Methods for anatomical feature analysis of a three-dimensional femoral bone structure model in a three-dimensional spatial coordinate system include:

[0055] First, a point set C of the distal condylar region is extracted by truncating the femur along the longitudinal axis. Second, the point set C is divided into medial condylar regions using a clustering algorithm. and lateral condylar region Subsequently, extreme value searches were performed before and after each region:

[0056]

[0057]

[0058] , These are the anterior and posterior points of the medial condyle, respectively; , These are respectively used as the anterior and posterior points of the lateral condyle;

[0059] The deepest point of the patellar surface is determined in the femoral patellar region using curvature analysis or geometric depression detection methods. .

[0060] Step 6: Calculate knee joint anatomical parameters based on key anatomical feature points and anatomical axes, including:

[0061] (1) Transcondylar line (TEA):

[0062]

[0063] in , The medial and lateral epicondyle points;

[0064] (2) Posterior condylar tangent (PCT):

[0065]

[0066] (3) Anteroposterior diameter of the medial femoral condyle:

[0067]

[0068] (4) Anteroposterior diameter of the lateral femoral condyle:

[0069]

[0070] (5) Width of the femur (medial and lateral sides):

[0071]

[0072] (6) Distance from the patellar surface point to the tangent at the posterior condyle:

[0073]

[0074] Step 7: The calculated anatomical parameters are output in a structured format, including the numerical values ​​of each parameter and their corresponding spatial relationships, and can be visualized and annotated on the 3D model. These anatomical parameters can be used for personalized knee joint prosthesis design and preoperative planning for total knee arthroplasty.

[0075] This invention utilizes a deep learning segmentation network to automatically identify and segment bone structures in knee CT images. It accurately extracts the three-dimensional structural information of the femur, avoiding the reliance on manual layer-by-layer delineation of bone structure contours in traditional methods. This significantly reduces manual intervention, improves image processing efficiency, and enhances the consistency and stability of segmentation results. Furthermore, this invention reconstructs the segmentation results into a three-dimensional femoral bone structure model, allowing for a complete representation of the spatial morphology of the bone structure. Compared to traditional two-dimensional image measurement methods, the three-dimensional model more realistically reflects the spatial structural characteristics of the skeleton, providing a reliable foundation for subsequent accurate measurement of anatomical parameters. Under a unified three-dimensional coordinate system, this invention automatically extracts key anatomical feature points of the medial condyle, lateral condyle, and patellar surface region through geometric feature analysis of key anatomical areas of the femur, and constructs corresponding anatomical axes. This enables automatic calculation of anatomical parameters, reducing subjective errors caused by manual measurement and improving the objectivity and repeatability of measurement results.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic measurement method for distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint, characterized in that, Includes the following steps: Step 1: Acquire knee joint CT medical image data and preprocess the image data; Step 2: Use a deep learning segmentation network to automatically segment the bone structure in the knee joint CT image data to obtain the three-dimensional segmentation result of the femur. Step 3: Perform three-dimensional reconstruction based on the three-dimensional segmentation results to generate a three-dimensional bone structure model of the femur; Step 4: Establish a unified three-dimensional coordinate system for the knee joint; Step 5: Perform anatomical feature analysis on the three-dimensional bone structure model of the femur in a three-dimensional spatial coordinate system, extract key anatomical feature points of the medial condyle, lateral condyle and patellar surface region of the femur, and determine the relevant anatomical axes; Step 6: Calculate the anatomical parameters of the knee joint based on key anatomical feature points and anatomical axes; Step 7: Output anatomical parameters to provide a basis for personalized knee prosthesis design or preoperative planning for total knee replacement surgery.

2. The method for automatically measuring distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint according to claim 1, characterized in that, In step 1, the method for preprocessing the image data is as follows: Represent the raw data of knee joint CT medical images as a three-dimensional grayscale volume. voxel coordinates In this stage, image preprocessing is performed on the original volume data, and grayscale values ​​are truncated to reduce the impact of abnormal noise, resulting in preprocessed volume data. This is used for subsequent automatic segmentation of bone structures.

3. The method for automatically measuring distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint according to claim 2, characterized in that, The deep learning segmentation network in step 2 is a medical image segmentation network based on the U-Net structure. It extracts multi-scale features through the encoder and decoder structure to achieve automatic segmentation of the femoral structure.

4. The method for automatically measuring distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint according to claim 3, characterized in that, Step 4 involves establishing a unified three-dimensional coordinate system for the knee joint, including the following methods: First, based on the femoral point cloud data, a principal orientation analysis is performed to determine the proximal and distal orientation of the femur, and this orientation is defined as... First, along the axial direction; second, extract the medial and lateral extreme points in the distal femur region and define the medial and lateral direction vectors: in, and These represent the extreme points on the medial and lateral sides of the femur, respectively. Determining the forward and backward directions using the cross product relationship: .

5. The method for automatically measuring distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint according to claim 4, characterized in that, Step 5, the method for anatomical feature analysis of the three-dimensional femoral bone structure model in a three-dimensional spatial coordinate system includes: First, a point set C of the distal condylar region is extracted by truncating the femur along the longitudinal axis. Second, the point set C is divided into medial condylar regions using a clustering algorithm. and lateral condylar region Subsequently, extreme value searches were performed before and after each region: ； ； , These are the anterior and posterior points of the medial condyle, respectively; , These are respectively used as the anterior and posterior points of the lateral condyle; The deepest point of the patellar surface is determined in the femoral patellar region using curvature analysis or geometric depression detection methods. .

6. The method for automatically measuring distal femoral anatomical parameters based on a three-dimensional CT model of the knee joint according to claim 5, characterized in that, The knee joint anatomical parameters in step 6 specifically include: the anteroposterior diameter of the medial femoral condyle; the anteroposterior diameter of the lateral femoral condyle; the width of the medial and lateral femoral condyles; the posterior condyle tangent; the transcondylar line; and patellar surface related distance parameters.

Images