Method for preparing patellofemoral joint prosthesis

Abstract

The present invention belongs to the field of joint prosthesis preparation, and specifically discloses a patellofemoral joint prosthesis preparation method, including: using a 2D femoral segmentation algorithm to segment the femoral trochlear groove to obtain femoral trochlear groove data; establishing a three-dimensional segmentation model according to the segmented femoral trochlear groove data, and obtaining trochlear surface data through the three-dimensional segmentation model; fitting the femoral trochlear groove surface according to the trochlear surface data, and preparing the patellofemoral joint prosthesis based on the femoral trochlear groove surface; wherein the 2D femoral segmentation algorithm includes: using the slice samples of the femoral trochlear groove as input data sets, batch inputting the 2D femoral segmentation model; using the 2D femoral segmentation model to segment the input data set to obtain the femoral trochlear groove data. The patellofemoral joint prosthesis designed by the above method can effectively improve the good matching between the artificial joint prosthesis and the real femoral trochlear groove.

Description

Technical Field

[0001] The invention belongs to the technical field of joint prosthesis preparation, and in particular relates to a method for preparing a patellofemoral joint prosthesis. Background Art

[0002] Patellofemoral osteoarthritis (PFOA) is a common bone disease mainly caused by patellar cartilage degeneration, wear and tear, and patellofemoral joint degeneration. Among people over 55 years old who show symptoms of patellofemoral osteoarthritis, 24% are women and 11% are men.

[0003] Patellofemoral arthroplasty (PFA) is a type of artificial knee joint unicompartmental surface replacement in which the replacement area is limited to the active area of ​​the patellofemoral joint. The patellofemoral joint prosthesis includes the femoral trochlear prosthesis and the patellar prosthesis. The treatment targets of patellofemoral joint replacement are patients with simple patellofemoral joint diseases who have failed non-surgical and conventional surgical treatments. The replacement area of ​​patellofemoral joint replacement is limited to the patellofemoral joint. It retains most of the physiological structures of the knee joint, such as the tibiofemoral joint, meniscus, and cruciate ligament, and preserves the knee joint's own kinematic mechanism to the greatest extent. Compared with total knee arthroplasty (TKA), it has the advantages of less trauma and faster postoperative recovery. In particular, the kinematics of the patellofemoral joint after total knee arthroplasty is closer to that of a normal knee joint.

[0004] However, whether the patellofemoral joint mechanism can be restored after patellofemoral joint replacement depends largely on the design of the artificial trochlear groove. The femoral trochlear groove is a continuous curve, and because the curve of the trochlear groove has different curvatures at different positions in the trochlear groove, the morphology of the trochlear groove varies.

[0005] At present, the design of patellofemoral joint prosthesis is mainly based on the morphological characteristics of the patellofemoral joint of Western Caucasians and combined with the Chinese knee joint database. The 3D bone model is generated by CT scanning of people of different ages, and the basic height and width of the femoral trochlear groove of the human body are manually measured. The corresponding 3D models of femoral trochlear prosthesis and patellar prosthesis are constructed according to the average structural parameters obtained.

[0006] A total knee prosthesis disclosed in the prior art includes a femoral prosthesis and a patellar prosthesis mounted on the femoral prosthesis, wherein the femoral prosthesis includes an anterior condylar surface provided with a trochlear groove, and the patellar prosthesis cooperates with the trochlear groove. The projection of the line connecting the lowest points of the trochlear groove on the coronal plane is an "S"-shaped curve. The total knee prosthesis designed by the invention can better improve the movement stability of the patellar prosthesis, the patellar trajectory is closer to the natural knee joint trajectory, the probability of postoperative patellar dislocation is reduced, and the wear of the patellar prosthesis is reduced, and the service life of the knee prosthesis is effectively improved. However, the characteristics of the trochlear groove of the artificial joint prosthesis designed by the invention cannot completely match the physiological and anatomical characteristics of the patient's trochlear groove.

[0007] Another example is a patellofemoral joint prosthesis and a manufacturing method thereof disclosed in the prior art, wherein the patellofemoral joint prosthesis includes a femoral trochlear prosthesis and a patellar prosthesis, wherein the femoral trochlear prosthesis includes a medial condyle and a lateral condyle located on both sides of the femoral trochlear prosthesis, and a trochlear groove connecting the medial condyle and the lateral condyle. The medial condyle and the lateral condyle are respectively connected to the trochlear groove in an arc shape with different curvature radii, so that the connection surface formed by the medial condyle, the lateral condyle and the trochlear groove together constitutes a matching curved surface for matching with the patellar prosthesis, the femoral trochlear prosthesis also includes a fixing surface formed on the femoral trochlear prosthesis opposite to the matching curved surface, a fixing assembly is formed on the fixing surface, the patellar prosthesis includes an articular curved surface matched with the matching curved surface, the cross section of the articular curved surface is elliptical in a top view state, and the cross section is asymmetric parabolic in a front view state, and a connecting column is formed on the surface of the patellar prosthesis opposite to the articular curved surface. Although the patellofemoral joint prosthesis designed by them is more suitable for Chinese people, the artificial trochlear groove prepared by this method is still quite different from the patient's normal trochlear groove, which may increase the incidence of patellofemoral joint complications such as patellar clicking and dislocation in patients after surgery.

[0008] Because existing patellofemoral joint prostheses have fewer design parameters and a large amount of data, and traditional manual measurement methods have large errors and low repeatability, previously designed pulley prostheses cannot match the patient's normal pulley anatomical structure well. Therefore, existing designs of patellofemoral joint prostheses can easily cause complications such as excessive filling of the patellofemoral joint and patellar clicking in patients after surgery. Summary of the invention

[0009] The technical problems to be solved by this application are: first, how to design a new patellofemoral joint prosthesis to overcome the problem that the existing patellofemoral joint prosthesis is prone to cause complications such as excessive filling of the patellofemoral joint and patellar snapping in patients after surgery; second, how to solve the problems of time-consuming, large errors, and low repeatability caused by traditional manual measurement. In order to solve the above problems, this application designs a patellofemoral joint prosthesis preparation method in order to prepare a patellofemoral joint prosthesis suitable for the characteristics of Chinese people, so that patients can better restore their physiological characteristics after completing patellofemoral joint prosthesis replacement.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] A method for preparing a patellofemoral joint prosthesis, comprising:

[0012] Use 2D femur segmentation algorithm to segment the femoral trochlea and obtain femoral trochlea data;

[0013] Establishing a three-dimensional segmentation model according to the femoral trochlea data, and acquiring trochlea surface data through the three-dimensional segmentation model;

[0014] The femoral trochlear groove curved surface is fitted according to the trochlear surface data, so as to prepare a patellofemoral joint prosthesis based on the femoral trochlear groove curved surface.

[0015] Preferably, a 2D femur segmentation algorithm is used to segment the femoral trochlea to obtain femoral trochlea data, including:

[0016] The slice samples of the femoral trochlea are used as input data sets and batch-input into the 2D femoral segmentation model;

[0017] The input data set is segmented using the 2D femur segmentation model to obtain the femoral trochlea data.

[0018] Preferably, the trochlear surface data includes measurement data of femoral trochlear width and depth, and the measurement method is:

[0019] Use the three-dimensional segmentation model to traverse the pixels of each slice sample;

[0020] Determine the highest point and the lowest point between the medial and lateral condyles of the femur of each slice sample;

[0021] The highest and lowest points of the entire femoral trochlear groove are determined to obtain measurement data of the width and depth of the femoral model.

[0022] Preferably, establishing a three-dimensional segmentation model according to the femoral trochlea data comprises: based on the femoral trochlea data, using 3D Slicer or ITK SNAP platform to develop a three-dimensional segmentation model.

[0023] Preferably, the input format of the slice sample is a (H, W, C) three-dimensional nii format; wherein H is the slice height, W is the slice width, and C is the number of slices.

[0024] Preferably, the 2D femur segmentation model includes an Encoder module, an ECA module and a Decoder module; correspondingly, the input data set is segmented using the 2D femur segmentation model to obtain the femoral trochlea data, including:

[0025] Processing the input data set through the Encoder module to obtain slice feature extraction data;

[0026] Performing intermediate data extraction on the slice feature extraction data through an ECA module to obtain intermediate feature data;

[0027] The intermediate feature data is predicted and segmented by a Decoder module to obtain labels, thereby obtaining the femoral trochlea data.

[0028] Preferably, the input data set is processed by an Encoder module to obtain slice feature extraction data, including: using the Encoder module to reduce the slice height H and slice width W of the input data set exponentially, and increase the number of slices C exponentially to obtain a low-level intermediate data set, and extract local detail information of the low-level intermediate data set to obtain slice feature extraction data.

[0029] Preferably, the intermediate feature data is predicted and segmented by a Decoder module to obtain the femoral trochlear groove data, including: using the Decoder module to multiply the slice height H and slice width W of the intermediate feature data, multiply the number of slices C, obtain the predicted segmentation label, and summarize the predicted labels to obtain the femoral trochlear groove data.

[0030] Preferably, the format of the femoral trochlea data is (b, H, W, 1), where b represents the number of slices in a batch and 1 represents the size of the channel.

[0031] Beneficial effects: The present application designs a method for preparing a patellofemoral joint prosthesis, which uses a 2D femoral segmentation algorithm to efficiently and accurately segment the femoral trochlea, then establishes a three-dimensional segmentation model based on the segmented femoral trochlea data, obtains the trochlea surface data, and finally fits a patellofemoral joint prosthesis that meets the patient's physiological characteristics based on the three-dimensional characteristics of the patient's femoral trochlea surface, thereby effectively improving the matching degree between the artificial joint prosthesis and the actual femoral trochlea anatomical morphology, maximally meeting actual needs, restoring the patient's normal posture, and bringing good news to patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 2D femur segmentation model designed for this application;

[0033] Figure 2 A flow chart of a method for preparing a patellofemoral joint prosthesis designed for this application;

[0034] Figure 3 A three-dimensional model of the femoral trochlear groove designed for this application;

[0035] Figure 4 Width measurement diagram designed for this application;

[0036] Figure 5 Visual fit diagram of the pulley surface designed for this application. DETAILED DESCRIPTION

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0038] Embodiment 1:

[0039] This embodiment mainly introduces a 2D femur segmentation model designed by the present application, and its specific design includes an Encoder module, an ECA module, and a Decoder module;

[0040] The slice samples of the patient's femoral trochlea are used as the input data set. The input data set is subjected to slice feature extraction through the Encoder module, and then important feature information is extracted through the ECA module, and then predicted segmentation labels are obtained through the Decoder module.

[0041] In some embodiments, slice samples of the patient's femoral trochlea are used as input data sets. The input data sets can be regarded as samples of patients. Assume that there are 30 data samples, i.e., 30 patients. Each patient sample is a three-dimensional nii format data. The data format can be recorded as a matrix of H*W*C, where H*W is the size of the slice, H is the slice height, W is the slice width, and C is the number of slices. Stacking all the slices of a patient becomes a nii format patient sample data.

[0042] For further information, please refer to Figure 2, the first part of the 2D femur segmentation model is the Encoder, that is, the left part of the network, which mainly realizes the feature extraction of the input slice, reduces the input size (H / W) by a factor of two, and increases the number of channels (C) by a factor of two..., this part will gradually reduce H and W by a factor of two, that is, become H / 2, H / 4 (the same as W)..., and C becomes 2C, 4C..., the Encoder uses a CNN convolutional neural network, which extracts low-level local detail information in the front, and as the number of CNN layers increases, abstract information and some high-level feature information are extracted. The second part of the 2D femur segmentation model is the ECA module, that is, an intermediate process module, which is an attention module, allowing the network to pay more attention to extracting useful and important feature information. The third part of the 2D femur segmentation model is the Decoder module, because it is ultimately necessary to obtain the predicted segmentation label of each pixel of the entire original feature map size, that is, each pixel on H*W must predict a label value, if this pixel is the segmented part, the predicted value is not 0; if it is the background part, the predicted label is 0. Therefore, it is necessary to use the CNN network step by step to restore the size of the feature map. This part H, W will increase exponentially, C will shrink exponentially, and finally a predicted label of the feature map size H*W will come out. Then take this predicted label and compare it with the marked true label. The comparison indicator is the DSC coefficient. The higher the DSC coefficient, the better the segmentation of the 2D femur segmentation model designed in this application.

[0043] Embodiment 2:

[0044] Based on a 2D femoral segmentation model designed in Example 1, this embodiment designs a method for preparing a patellofemoral joint prosthesis. Please refer to the flowchart Figure 1 , specifically including the following steps:

[0045] Step S1, using a 2D femur segmentation algorithm to segment the femoral trochlea to obtain femoral trochlea data;

[0046] Step S2, establishing a three-dimensional segmentation model according to the femoral trochlea data, and obtaining trochlea surface data through the three-dimensional segmentation model;

[0047] Step S3, fitting a femoral trochlear groove curved surface according to the trochlear surface data, so as to prepare a patellofemoral joint prosthesis based on the femoral trochlear groove curved surface.

[0048] Furthermore, the femoral trochlea is segmented using a 2D femoral segmentation algorithm, including:

[0049] Step S101, using the slice samples of the femoral trochlea as input data sets, and batch inputting a 2D femur segmentation model;

[0050] Step S102: Use the 2D femur segmentation model to segment the input data set to obtain the femoral trochlea data.

[0051] Furthermore, a slice sample of the patient's femoral trochlea is used as an input data set, and the input data set is subjected to slice feature extraction through the Encoder module, and then important feature information is extracted through the ECA module, and then predicted segmentation labels are obtained through the Decoder module.

[0052] The present invention is to enable the patient to better restore his physiological characteristics after completing patellofemoral joint prosthesis replacement, and to fit a patellofemoral joint prosthesis that meets the patient's physiological characteristics according to the patient's femoral trochlear three-dimensional characteristics, so there is a high requirement for the accurate measurement of the curve, angle and width of the femoral trochlear groove. Therefore, in the step S2, the trochlear surface data includes the width measurement data of the femoral model, and the measurement method is:

[0053] Step S201, using the three-dimensional segmentation model to traverse the pixels of each slice sample;

[0054] Step S202, determining the highest point position and the lowest point position between the medial and lateral condyles of the femur of each slice sample;

[0055] Step S203, determining the highest point position and the lowest point position of the entire femoral trochlear groove, thereby obtaining measurement data of the width and depth of the femoral model.

[0056] The trochlear surface data includes the width measurement data of the femoral model, and the measurement method thereof includes the following steps:

[0057] In some embodiments, Figure 4 The width measurement diagram shown is Figure 4 (a) To determine the highest and lowest points between the two strands of each slice, multiple slices are combined and the data of different highest and lowest points are summarized to obtain Figure 4 (b) Figure 4 (b) To determine the highest and lowest points of the entire femoral trochlea.

[0058] In some embodiments, the lowest point is at [97, 118, 95], and the highest point is at [105, 165, 88]; the horizontal distance of the highest point is: 32.89999943971634 mm; the vertical distance of the highest point is:

[0059] 5.599999904632568mm. The width of the femur model can be obtained by the above method.

[0060] The segmented results will be stacked slice by slice to generate nii prediction data, so a visual interactive program needs to be developed. Therefore, in step S2, the method for establishing a three-dimensional segmentation model is to develop a three-dimensional segmentation model based on the femoral trochlear groove data using 3DSlicer or ITK SNAP platform; the three-dimensional segmentation model architecture can be found in Figure 3 , Figure 3 (a) is the structure of the femoral trochlear groove with stacked slices. Figure 3 (b) is its front view, Figure 3 (c) is its side view, Figure 3 (d) is a back view thereof, Figure 3 (e) is its bottom view.

[0061] The specific method of data processing through the 3D segmentation model is: according to user needs, select points on the 3D segmentation model, perform pulley surface fitting, and generate a 3D printing stl format file to store the pulley surface data information. Figure 5 , Figure 5 (a) is the structure of the femoral trochlear groove. Figure 5 (b) is a schematic diagram of the left side of the pulley surface. Figure 5 (c) is a schematic diagram of the right side of the pulley surface.

[0062] The 2D femur segmentation algorithm designed in this application is based on the designed three-dimensional segmentation model. The comparison effect between the 2D femur segmentation algorithm and the existing segmentation model is shown in Table 1.

[0063] Table 1 Comparison of 2D femur segmentation model and existing segmentation models

[0064]

[0065] As can be seen from Table 1, the sensitivity and accuracy of the design of this application are relatively good.

[0066] In some embodiments, slice samples of the patient's femoral trochlear groove are used as input data sets. The data sets can be regarded as patient samples. Assume that there are 30 data samples, i.e., 30 patients. Each patient sample is a three-dimensional nii format data. The data format can be recorded as a H*W*C matrix, where H*W is the size of the slice, i.e., the length and width shown in the figure above, and C represents the number of slices. Stacking all the slices of a patient becomes a nii format patient sample data. The network model is a 2D network. The input to the network is in units of slices, and the size is a tensor of H*W*1, where 1 represents the channel size. The input to the network is in batches, so the final format entering the network is (b, H, W, 1), where b represents how many slices are in a batch.

[0067] The present application designs a method for preparing a patellofemoral joint prosthesis, which uses a 2D femoral segmentation algorithm to efficiently and accurately segment the femoral trochlea, then establishes a three-dimensional segmentation model based on the segmented femoral trochlea data, obtains trochlear surface data, and finally fits a patellofemoral joint prosthesis that meets the patient's physiological characteristics based on the three-dimensional characteristics of the patient's femoral trochlea groove, thereby effectively improving the matching degree between the artificial joint prosthesis and the real femoral trochlea, meeting actual needs to the greatest extent, restoring the patient's normal body shape, and bringing good news to the patient. Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for preparing a patellofemoral joint prosthesis, characterized in that: include: Use 2D femur segmentation algorithm to segment the femoral trochlea and obtain femoral trochlea data; Establishing a three-dimensional segmentation model according to the femoral trochlea data, and acquiring trochlea surface data through the three-dimensional segmentation model; Fitting a femoral trochlear groove curved surface according to the trochlear surface data, so as to prepare a patellofemoral joint prosthesis based on the femoral trochlear groove curved surface; Among them, the 2D femur segmentation algorithm is used to segment the femoral trochlea to obtain the femoral trochlea data, including: The slice samples of the femoral trochlea are used as input data sets and batch-input into the 2D femoral segmentation model; Segment the input data set using the 2D femur segmentation model to obtain the femoral trochlea data; The 2D femur segmentation model includes an Encoder module, an ECA module and a Decoder module, and the ECA module is an attention module; correspondingly, the input data set is segmented using the 2D femur segmentation model to obtain the femoral trochlea data, including: Processing the input data set through the Encoder module to obtain slice feature extraction data; Performing intermediate data extraction on the slice feature extraction data through an ECA module to obtain intermediate feature data; The intermediate feature data is predicted and segmented by a Decoder module to obtain the femoral trochlea data; The trochlear surface data includes measurement data of femoral trochlear width and depth, and the measurement method is: Use the three-dimensional segmentation model to traverse the pixels of each slice sample; Determine the highest point and the lowest point between the medial and lateral condyles of the femur of each slice sample; Determine the highest and lowest points of the entire femoral trochlear groove, thereby obtaining the measurement data of the width and depth of the femoral model; Establishing a three-dimensional segmentation model according to the femoral trochlea data, including: based on the femoral trochlea data, using 3DSlicer or ITK SNAP platform to develop a three-dimensional segmentation model; The input format of the slice sample is a three-dimensional nii format (H, W, C); wherein H is the slice height, W is the slice width, and C is the number of slices; Processing the input data set through the Encoder module to obtain slice feature extraction data, including: using the Encoder module to reduce the slice height H and slice width W of the input data set exponentially, and to increase the number of slices C exponentially to obtain a low-level intermediate data set, and extracting local detail information of the low-level intermediate data set to obtain slice feature extraction data; The intermediate feature data is predicted and segmented by a Decoder module to obtain the femoral trochlear groove data, including: using the Decoder module to multiply the slice height H and slice width W of the intermediate feature data, multiply the number of slices C, obtain the predicted segmentation label, and summarize the predicted label to obtain the femoral trochlear groove data; The format of the femoral trochlea data is (b, H, W, 1), where b represents the number of slices in a batch and 1 represents the size of the channel.

Images