Preoperative planning method and system for artificial-intelligence-based knee joint ligament reconstruction surgery, and device

Abstract

A preoperative planning method and system for artificial-intelligence-based knee joint ligament reconstruction surgery, and a device and a computer-readable storage medium. The method comprises: acquiring a first knee joint CT image (S101); performing image segmentation on the first knee joint CT image to obtain a second knee joint CT image (S102); performing key point identification on the second knee joint CT image to obtain a third knee joint CT image (S103); performing bone correction on the third knee joint CT image to obtain a fourth knee joint CT image (S104); performing position optimization on a key point in the fourth knee joint CT image to obtain a fifth knee joint CT image (S105); performing bone tunnel planning on the basis of the fifth knee joint CT image to obtain a bone tunnel planning result (S106); and performing three-dimensional reconstruction rendering on the basis of the bone tunnel planning result, so as to present a bone tunnel by means of the result (S107). The method can improve the accuracy of bone tunnel planning.

Description

Artificial intelligence preoperative planning method, system and equipment for knee ligament reconstruction Technical Field

[0001] The present application belongs to the field of deep learning intelligent recognition technology, and in particular relates to a preoperative planning method, system, device and computer-readable storage medium for artificial intelligence knee ligament reconstruction. Background Art

[0002] Currently, the knee ligament bone tunnel planning in related technologies mainly relies on experienced doctors to perform knee ligament bone tunnel planning, but the bone tunnel planning is inaccurate.

[0003] Therefore, how to improve the accuracy of bone tunnel planning is a technical problem that those skilled in the art urgently need to solve.

[0004] Summary of the Invention

[0005] The embodiments of the present application provide a preoperative planning method, system, device, and computer-readable storage medium for artificial intelligence knee ligament reconstruction surgery, which can improve the accuracy of bone tunnel planning.

[0006] In a first aspect, an embodiment of the present application provides a preoperative planning method for artificial intelligence knee ligament reconstruction, comprising:

[0007] Obtain a CT image of the first knee joint;

[0008] Performing image segmentation on the first knee joint CT image to obtain a second knee joint CT image; wherein the four segments of the femur, tibia, patella, and fibula segmented from the second knee joint CT image are displayed separately;

[0009] Perform key point recognition on the second knee joint CT image to obtain a third knee joint CT image;

[0010] Performing bone correction on the third knee joint CT image to obtain the fourth knee joint CT image;

[0011] Optimizing the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image;

[0012] Bone tunnel planning is performed based on the CT image of the fifth knee joint to obtain the bone tunnel planning results;

[0013] Based on the bone tunnel planning results, three-dimensional reconstruction rendering is performed to display the bone tunnel results.

[0014] Optionally, performing image segmentation on the first knee joint CT image to obtain a second knee joint CT image includes:

[0015] Inputting the first knee joint CT image into a preset knee joint image segmentation network model, and outputting a second knee joint CT image;

[0016] Among them, the knee joint image segmentation network model is based on a 2D neural network or a 3D neural network, and is obtained by model training using a training set, a validation set, and a test set.

[0017] Optionally, performing key point recognition on the second knee joint CT image to obtain a third knee joint CT image includes:

[0018] Inputting the second knee joint CT image into a preset key point recognition network model to obtain a third knee joint CT image;

[0019] Among them, the network structure of the key point recognition network model includes three branches, the first branch is configured to obtain the key point heat map; the second branch is configured to obtain the key point position offset; and the third branch is configured to obtain the key point border size.

[0020] Optionally, also include:

[0021] Determine the branch loss functions and weights corresponding to the three branches;

[0022] Determine the overall loss function based on the branch loss functions and weights corresponding to the three branches;

[0023] Calculate the loss value based on the overall loss function;

[0024] Based on the loss value, feedback is used to adjust the network parameters.

[0025] Optionally, bone correction is performed on the third knee joint CT image to obtain a fourth knee joint CT image, including:

[0026] Determine the rotation angle;

[0027] Based on the rotation angle, determine the rotation matrix;

[0028] The third knee joint CT image is multiplied by the rotation matrix to obtain a rotated fourth knee joint CT image.

[0029] Optionally, optimizing the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image includes:

[0030] Calculate the difference in length expansion and contraction between point A on the medial and posterior superior side of the lateral femoral condyle and point C on the anteromedial 1 / 3 of the tibial intercondylar eminence at different angles of the femur and tibia;

[0031] The positions of point A on the medial posterior superior side of the lateral femoral condyle and point C on the anteromedial 1 / 3 of the tibial intercondylar eminence, where the length expansion and contraction change difference is the smallest, are determined as the optimized positions.

[0032] Optionally, bone tunnel planning is performed based on the fifth knee joint CT image to obtain bone tunnel planning results, including:

[0033] Bone tunnel planning was performed based on the CT images of the fifth knee joint, and the planning results of three bone tunnels on the tibial side, ligament side, and femoral side were obtained.

[0034] In a second aspect, an embodiment of the present application provides an artificial intelligence preoperative planning system for knee ligament reconstruction, the system comprising:

[0035] an image acquisition module, configured to acquire a first knee joint CT image;

[0036] an image segmentation module configured to perform image segmentation on the first knee joint CT image to obtain a second knee joint CT image; wherein the four segments of the femur, tibia, patella, and fibula segmented from the second knee joint CT image are displayed separately;

[0037] a key point recognition module, configured to perform key point recognition on the second knee joint CT image to obtain a third knee joint CT image;

[0038] a bone correction module, configured to perform bone correction on the third knee joint CT image to obtain a fourth knee joint CT image;

[0039] a position optimization module configured to optimize the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image;

[0040] a bone tunnel planning module, configured to perform bone tunnel planning based on the fifth knee joint CT image and obtain a bone tunnel planning result;

[0041] The result display module is configured to perform three-dimensional reconstruction and rendering based on the bone tunnel planning result to display the bone tunnel as a result.

[0042] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory storing computer program instructions;

[0043] When the processor executes the computer program instructions, it implements the preoperative planning method for artificial intelligence knee ligament reconstruction as shown in the first aspect.

[0044] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, implement the preoperative planning method for artificial intelligence knee ligament reconstruction as shown in the first aspect.

[0045] The preoperative planning method, system, device and computer-readable storage medium for artificial intelligence knee ligament reconstruction surgery in the embodiments of the present application can improve the accuracy of bone tunnel planning.

[0046] The preoperative planning method for artificial intelligence knee ligament reconstruction surgery includes: obtaining a first knee joint CT image; performing image segmentation on the first knee joint CT image to obtain a second knee joint CT image; wherein the four parts of the femur, tibia, patella and fibula segmented from the second knee joint CT image are displayed separately; performing key point recognition on the second knee joint CT image to obtain a third knee joint CT image; performing bone correction on the third knee joint CT image to obtain a fourth knee joint CT image; performing position optimization on key points in the fourth knee joint CT image to obtain a fifth knee joint CT image; performing bone tunnel planning based on the fifth knee joint CT image to obtain a bone tunnel planning result; performing three-dimensional reconstruction rendering based on the bone tunnel planning result to display the bone tunnel as a result. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the specific implementation methods or the description of the prior art. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0048] FIG1 is a flow chart of a preoperative planning method for artificial intelligence knee ligament reconstruction provided by one embodiment of the present application;

[0049] FIG2 is a flowchart of a preoperative planning method for artificial intelligence knee ligament reconstruction provided by one embodiment of the present application;

[0050] FIG3 is a schematic diagram of a 3D segmentation network structure provided by an embodiment of the present application;

[0051] FIG4 is a schematic diagram of a key point recognition network structure provided by an embodiment of the present application;

[0052] FIG5 is a schematic diagram of the Hourglass network structure provided by an embodiment of the present application;

[0053] FIG6 is a schematic diagram of a network structure of a global context module provided by one embodiment of the present application;

[0054] FIG7 is a schematic diagram of a location point set of points A and C provided in one embodiment of the present application;

[0055] FIG8 is a schematic diagram showing the results provided by an embodiment of the present application;

[0056] FIG9 is a schematic diagram of the structure of a preoperative planning system for artificial intelligence knee ligament reconstruction provided by one embodiment of the present application;

[0057] FIG10 is a schematic structural diagram of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0058] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating the examples of the present application.

[0059] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, the elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or device comprising the elements.

[0060] To address the existing technical issues, the present invention provides a method, system, device, and computer-readable storage medium for preoperative planning of artificial intelligence knee ligament reconstruction surgery. The following first introduces the preoperative planning method for artificial intelligence knee ligament reconstruction surgery provided by the present invention.

[0061] FIG1 shows a flow chart of a preoperative planning method for artificial intelligence knee ligament reconstruction provided by one embodiment of the present application. As shown in FIG1 , the preoperative planning method for artificial intelligence knee ligament reconstruction includes:

[0062] S101, obtaining a first knee joint CT image;

[0063] S102, performing image segmentation on the first knee joint CT image to obtain a second knee joint CT image; wherein the four segments of the femur, tibia, patella, and fibula segmented from the second knee joint CT image are displayed separately;

[0064] S103. Key point identification is performed on the second knee joint CT image to obtain a third knee joint CT image. Specifically, the key points displayed in the third knee joint CT image may include: point A at the medial superior and posterior aspect of the lateral femoral condyle, point B at the lateral edge of the anterior femoral condyle, point C at the medial 1 / 3 of the tibial intercondylar eminence, point D at the medial edge of the tibial tuberosity, point E at the medial condyle, and point F at the lateral condyle.

[0065] S104, performing bone correction on the third knee joint CT image to obtain a fourth knee joint CT image;

[0066] S105, optimizing the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image;

[0067] Specifically, the positions of point A on the medial posterior superior side of the lateral femoral condyle and point C on the anteromedial 1 / 3 of the tibial intercondylar eminence in the fourth knee joint CT image may be optimized to obtain the fifth knee joint CT image.

[0068] S106, performing bone tunnel planning based on the fifth knee joint CT image to obtain a bone tunnel planning result;

[0069] S107 , performing three-dimensional reconstruction and rendering based on the bone tunnel planning result, and displaying the bone tunnel with the result.

[0070] Specifically, Figure 2 is a flow chart of a preoperative planning method for artificial intelligence knee ligament reconstruction provided by an embodiment of the present application, as shown in Figure 2.

[0071] In one embodiment, performing image segmentation on the first knee joint CT image to obtain the second knee joint CT image includes:

[0072] Inputting the first knee joint CT image into a preset knee joint image segmentation network model, and outputting a second knee joint CT image;

[0073] Among them, the knee joint image segmentation network model is based on a 2D neural network or a 3D neural network, and is obtained by model training using a training set, a validation set, and a test set.

[0074] Specifically, we obtained a dataset of medical knee joint images and manually annotated them. Finally, we extracted only the labels containing the various parts of the knee joint as our segmentation mask. We converted the CT data and mask data into nii.gz format, shuffled them, and then split them into training, validation, and test sets in a 6:2:2 ratio.

[0075] The knee joint image segmentation network can use a 2D neural network or a 3D neural network to segment CT data. Figure 3 shows the 3D segmentation network structure. The knee joint multi-task segmentation neural network uses three downsampling and three upsampling. Each convolution layer uses 1x1x1, 3x3x3, and 1x1x1 convolution kernels. A skip cascade structure is used between downsampling and upsampling at the same scale to reduce feature loss.

[0076] During model training, the batch_size is 64, the initial learning rate is set to 1e-4, and an additional learning rate decay strategy is used. Every 5000 iterations, the learning rate decays to the original 0.9. The optimizer uses the Adam optimizer and the loss function is DICE loss. The training set and validation set are validated once every 1000 iterations. The early stopping method is used to determine the stopping time of network training and obtain the final model.

[0077] In one embodiment, performing key point recognition on the second knee joint CT image to obtain a third knee joint CT image includes:

[0078] Inputting the second knee joint CT image into a preset key point recognition network model to obtain a third knee joint CT image;

[0079] Among them, the network structure of the key point recognition network model includes three branches, the first branch is configured to obtain the key point heat map; the second branch is configured to obtain the key point position offset; and the third branch is configured to obtain the key point border size.

[0080] Specifically, the segmented data results were obtained and a training dataset was created. The training dataset was shuffled and divided into a training set, a validation set, and a test set in a 6:2:2 ratio. The points that needed to be identified in the knee joint were the two condylar points (medial condylar point E and lateral condylar point F), point A on the medial and superior side of the lateral femoral condyle, point B on the lateral edge of the anterior femoral condyle, point C on the medial 1 / 3 of the tibial intercondylar eminence, and point D on the medial edge of the tibial tuberosity, for a total of six points.

[0081] The knee joint point recognition algorithm used in this paper mainly determines the target information through the center point coordinates and offset of the target, and regresses the target's posture information, direction information, size, etc., thereby converting the target detection problem into a key point estimation problem.

[0082] Model training uses standard supervised learning, and inference utilizes a single feedforward network. There are three main backbone networks for feature extraction: ResNet-18, DLA-34, and Hourglass. The Hourglass-based keypoint recognition network architecture used here is shown in Figure 4.

[0083] As shown in Figure 4, the entire knee joint recognition network consists of three branches. The input image is output by the hourglass network and then passes through three branches. The first branch generates a heatmap for each keypoint after passing through a 3x3 convolutional layer, a 1x1 convolutional layer, and maximum pooling. This is the keypoint heatmap branch. The second branch calculates the position offset of each keypoint through a 3x3 convolutional layer and a 1x1 convolutional layer. This is the keypoint position offset branch. The third branch calculates the bounding box size of each keypoint through a 3x3 convolutional layer and a 1x1 convolutional layer. This is the keypoint bounding box size branch.

[0084] The keypoint heatmap branch and keypoint bounding box size branch are configured to generate location maps for all object categories. The keypoint position offset branch represents the offset between the actual object position and the predicted position. The knee joint recognition network is configured to output feature maps by reducing the image resolution by a factor of 4 to accelerate model training. The point recognition process is as follows:

[0085] The image input to the network is represented as: I∈R W*H*3

[0086] The key point heat map predicted by the network output is:

[0087] Where W and H represent the width and height of the image, R is the output step size, here R=4 is set, and C represents the number of key point categories in the output heat map. Indicates that the corresponding position is the key point position of the detection, Indicates that the detected location is the background.

[0088] In the prediction stage, assuming (x1, x2, y1, y2) is the bounding box of the target, the center point p is calculated as: p = ((x1 + x2 / 2), (y1 + y2 / 2))

[0089] Other information of the target is obtained from the image information of the key points, and all the center points are obtained through The prediction is obtained, and then the target size is obtained through regression. The regressed position coordinate p* is:

[0090] in, are the offsets of the horizontal and vertical coordinate positions of the center point, represents the width and height of the detection target, δ represents the position weight, and Indicates the horizontal and vertical coordinate positions.

[0091] In one embodiment, it further includes:

[0092] Determine the branch loss functions and weights corresponding to the three branches;

[0093] Determine the overall loss function based on the branch loss functions and weights corresponding to the three branches;

[0094] Calculate the loss value based on the overall loss function;

[0095] Based on the loss value, the feedback adjusts the network parameters.

[0096] Specifically, the knee joint recognition network has three branch outputs, consisting of the key point heat map branch loss function (Lk), the key point position offset branch loss function (Lsize), and the key point bounding box size branch loss function (Loff):

[0097] The key point heat map branch loss function Lk is:

[0098] in, The hourglass network is used to predict the center point coordinates of the heat map. Here, α=2 and β=4 are set to represent hyperparameters, and N represents the number of key points.

[0099] The key point bounding box size branch loss function Loff is:

[0100] Among them, Loff is the L1 loss function, N represents the number of key points, f(x i ) and y i Represents the true coordinate value and predicted coordinate value of the i-th key point position.

[0101] The key point position offset branch loss function Lsize is expressed as:

[0102] Among them, Lsize is the L2 loss function, N represents the number of key points, f(x i ) and y i Represents the true coordinate value and predicted coordinate value of the i-th key point position.

[0103] Finally, the loss function is: Loss = λ0L k +λ size L size +λ off L off

[0104] Among them, λ0+λ size +λ off =1.

[0105] Knee joint key point heat map calculation process:

[0106] Assume that the confidence of the k-th key point at position p on the heat map Hk is HK(p), which is expressed as

[0107] Among them, (x k ,y k ) is the coordinate of the kth key point, k∈[0,6], and 6 key points need to be calculated here; σ is the standard deviation representing the energy concentration in the heat map. When σ is larger, the heat map covers a larger area, and when σ is smaller, the heat map covers a smaller area.

[0108] The heat maps of the six key points of the knee joint are combined to form the final heat map H∈R w*h*k , k represents the number of key points, w, h represent width and height. According to the predicted heat map Hk distribution, the final prediction result of the knee joint position is

[0109] in, It indicates that the model predicts the confidence of the coordinate point p on the heat map of the k-th key point. The coordinate of the point with the largest confidence value is the predicted position of the k-th key point.

[0110] Hourglass Network:

[0111] The attention mechanism is widely used in image processing, enabling networks to successfully learn information related to objects. Since the experimental dataset was large, with numerous object categories and varying sizes, a global context module was added to the Hourglass feature extraction network to better extract target features. This module can be configured at each layer of the feature extraction network to capture network dependencies, helping the network improve its detection capabilities for small objects.

[0112] As shown in Figure 5, the Hourglass network is quite deep and has a large number of parameters. Therefore, adding the attention module also requires considering the network's computational overhead. The global context module, on the other hand, uses a lightweight network, which adds negligible computational overhead and reduces the burden of network training.

[0113] The network structure of the global context module is shown in Figure 6:

[0114] Global context module calculation formula:

[0115] in, Expressed as global pooling weight, δ(x) = W v2 Relu(LN(W v1 (x))) represents feature transformation, z i =xi +δ(x) means merging the global context function into all position functions.

[0116] In one embodiment, performing bone correction on the third knee joint CT image to obtain a fourth knee joint CT image includes:

[0117] Determine the rotation angle;

[0118] Based on the rotation angle, determine the rotation matrix;

[0119] The third knee joint CT image is multiplied by the rotation matrix to obtain a rotated fourth knee joint CT image.

[0120] Specific rules for skeletal correction:

[0121] 1) The line connecting the condyles is parallel to the plane formed by the Y-axis and the Z-axis. Calculate the angle a between the line connecting the condyles and the Y-axis and the Z-axis, i.e., the rotation angle.

[0122] 2) Only the femur is adjusted, and the tibia moves with the femur;

[0123] 3) Calculate the rotation matrix based on the rotation angle. The 2D image rotation matrix is ​​a 3*3 matrix as follows: T 旋转前图像 *T m =T 旋转后图像

[0124] Among them, Tm represents the rotation matrix, a represents the rotation angle, T 旋转前图像 represents the femoral front rotation matrix, T 旋转后图像 represents the femoral anterior rotation matrix.

[0125] In one embodiment, optimizing the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image includes:

[0126] Calculate the difference in length expansion and contraction between point A on the medial and posterior superior side of the lateral femoral condyle and point C on the anteromedial 1 / 3 of the tibial intercondylar eminence at different angles of the femur and tibia;

[0127] The positions of point A on the medial posterior superior side of the lateral femoral condyle and point C on the anteromedial 1 / 3 of the tibial intercondylar eminence, where the length expansion and contraction change difference is the smallest, are determined as the optimized positions.

[0128] Specifically, key point position optimization:

[0129] Calculate the shortest AC length expansion and contraction position for the femur and tibia at 0°, 90°, and 120°. Points A and C are the initial points for point recognition.

[0130] Keeping the tibia stationary, rotate the femur 90° and 120° around the condylar line and calculate the distance between points A and C.

[0131] Femoral rotation matrix calculation method:

[0132] Calculate the rotation matrix based on the rotation angle. The femur data is three-dimensional data, and the three-dimensional image rotation matrix is ​​a 4*4 matrix, as follows: T 旋转前股骨 *T matrix =T 旋转后股骨

[0133] Among them, Tmatrix represents the rotation matrix, a here chooses 90° and 120°, T 旋转前股骨 represents the femoral front rotation matrix, T 旋转后股骨 represents the femoral anterior rotation matrix.

[0134] A, C point optimization method:

[0135] 1. Using the initial points A and C identified by the point recognition network as the center of the sphere, find a point set with a radius of 5 mm on the bone surface. Take one point from each of the A and C point sets to form N AC point combinations, as shown in Figure 5.

[0136] 2. Calculate the difference in AC length expansion and contraction at each AC point at 0°, 90°, and 120°.

[0137] 3. Repeat step 2 and calculate the difference in AC length expansion and contraction for each set of AC points at 0°, 90°, and 120°. Find the AC point pair with the minimum AC point expansion and contraction change.

[0138] 4. Output the optimized coordinates of points A and C.

[0139] The spatial distance d between two points AC is:

[0140] Among them, the coordinates of point A are (x1, y1, z1), and the coordinates of point C are (x2, y2, z2).

[0141] Among them, the location point sets of points A and C are shown in Figure 7.

[0142] In one embodiment, bone tunnel planning is performed based on a CT image of the fifth knee joint to obtain a bone tunnel planning result, including:

[0143] Bone tunnel planning was performed based on the CT images of the fifth knee joint, and the planning results of three bone tunnels on the tibial side, ligament side, and femoral side were obtained.

[0144] Specifically, bone tunnel planning:

[0145] The knee joint bone tunnel is divided into three sections: tibial side, ligament, and femoral side. The femoral side has a tunnel formed by points A and B, the tibial side has a tunnel formed by points C and D, and the ligament has a tunnel formed by points A and C.

[0146] A knee bone tunnel is formed based on the tibial side, ligament, and femoral side.

[0147] Results: The AC length expansion and contraction changes of the femur and tibia at 0°, 90°, and 120°.

[0148] The results are shown in Figure 8.

[0149] FIG9 is a schematic diagram of a preoperative planning system for artificial intelligence knee ligament reconstruction surgery provided by one embodiment of the present application. The preoperative planning system for artificial intelligence knee ligament reconstruction surgery includes:

[0150] An image acquisition module 901 is configured to acquire a first knee joint CT image;

[0151] The image segmentation module 902 is configured to perform image segmentation on the first knee joint CT image to obtain a second knee joint CT image; wherein the four segments of the femur, tibia, patella, and fibula segmented from the second knee joint CT image are displayed separately;

[0152] a key point recognition module 903 configured to perform key point recognition on the second knee joint CT image to obtain a third knee joint CT image;

[0153] a bone correction module 904 configured to perform bone correction on the third knee joint CT image to obtain a fourth knee joint CT image;

[0154] a position optimization module 905 configured to optimize the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image;

[0155] A bone tunnel planning module 906 is configured to perform bone tunnel planning based on the fifth knee joint CT image and obtain a bone tunnel planning result;

[0156] The result display module 907 is configured to perform three-dimensional reconstruction and rendering based on the bone tunnel planning result, and display the bone tunnel as a result.

[0157] FIG10 shows a schematic structural diagram of an electronic device provided in an embodiment of the present application.

[0158] The electronic device may include a processor 1001 and a memory 1002 storing computer program instructions.

[0159] Specifically, the processor 1001 may include a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0160] Memory 1002 may include a large capacity memory configured as data or instructions. By way of example and not limitation, memory 1002 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 1002 may include removable or non-removable (or fixed) media. Where appropriate, memory 1002 may be inside or outside the electronic device. In certain embodiments, memory 1002 may be a non-volatile solid-state memory.

[0161] In one embodiment, the memory 1002 may be a read-only memory (ROM). In one embodiment, the ROM may be a mask-programmable ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or a flash memory, or a combination of two or more of these.

[0162] The processor 1001 reads and executes the computer program instructions stored in the memory 1002 to implement any one of the artificial intelligence knee ligament reconstruction preoperative planning methods in the above embodiments.

[0163] In one example, the electronic device may further include a communication interface 1003 and a bus 1010. As shown in FIG10 , the processor 1001, the memory 1002, and the communication interface 1003 are connected via the bus 1010 and communicate with each other.

[0164] The communication interface 1003 is mainly configured to implement communication between various modules, systems, units and / or devices in the embodiments of the present application.

[0165] Bus 1010 comprises hardware, software or both, couples the parts of electronic equipment to each other.For example, and not limitation, bus can comprise accelerated graphics port (AGP) or other graphics bus, enhanced industry standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industry standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations.In suitable cases, bus 1010 can comprise one or more buses.Although the present application embodiment describes and shows specific bus, the application considers any suitable bus or interconnection.

[0166] In addition, in conjunction with the AI-powered preoperative planning method for knee ligament reconstruction surgery described in the aforementioned embodiments, embodiments of the present application may provide a computer-readable storage medium for implementation. The computer-readable storage medium stores computer program instructions; when executed by a processor, the computer program instructions implement any of the AI-powered preoperative planning methods for knee ligament reconstruction surgery described in the aforementioned embodiments.

[0167] It should be understood that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present application.

[0168] The functional modules shown in the above-mentioned block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), suitable firmware, a plug-in unit, a function card or the like. When implemented in software, the elements of the present application are programs or code segments configured to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memories, erasable ROMs (EROMs), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0169] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or systems. However, this application is not limited to the order of the above steps. In other words, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0170] The present application has been described above with reference to the flowcharts and / or block diagrams of the methods, systems and computer program products according to the embodiments of the present application. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing system to produce a machine so that these instructions executed via the processor of the computer or other programmable data processing system enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It is also understood that each box in the block diagram and / or the flowchart and the combination of the boxes in the block diagram and / or the flowchart can also be implemented by the dedicated hardware that performs the specified function or action, or can be implemented by the combination of dedicated hardware and computer instructions.

[0171] The above description is only a specific embodiment of the present application. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the scope of protection of the present application.

Claims

1. A preoperative planning method for artificial intelligence knee ligament reconstruction, wherein: include: Obtain the first knee joint CT image; Performing image segmentation on the first knee joint CT image to obtain a second knee joint CT image; wherein the four parts of the femur, tibia, patella and fibula segmented from the second knee joint CT image are displayed respectively; Performing key point recognition on the second knee joint CT image to obtain a third knee joint CT image; Performing bone correction on the third knee joint CT image to obtain a fourth knee joint CT image; Optimizing the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image; Bone tunnel planning is performed based on the CT image of the fifth knee joint to obtain the bone tunnel planning result; Three-dimensional reconstruction rendering is performed based on the bone tunnel planning results to show the bone tunnel results.

2. The preoperative planning method for artificial intelligence knee ligament reconstruction according to claim 1, wherein: Performing image segmentation on the first knee joint CT image to obtain a second knee joint CT image includes: Inputting the first knee joint CT image into a preset knee joint image segmentation network model, and outputting a second knee joint CT image; Among them, the knee joint image segmentation network model is based on a 2D neural network or a 3D neural network, and is obtained by model training using a training set, a validation set, and a test set.

3. The preoperative planning method for artificial intelligence knee ligament reconstruction according to claim 1, wherein: Key point recognition is performed on the second knee joint CT image to obtain a third knee joint CT image, including: Inputting the second knee joint CT image into a preset key point recognition network model to obtain a third knee joint CT image; Among them, the network structure of the key point recognition network model includes three branches, the first branch is configured to obtain the key point heat map; the second branch is configured to obtain the key point position offset; and the third branch is configured to obtain the key point border size.

4. The preoperative planning method for artificial intelligence knee ligament reconstruction according to claim 3, wherein: Also includes: Determine the branch loss functions and weights corresponding to each of the three branches; Based on the branch loss functions and weights corresponding to the three branches, the overall loss function is determined; Calculate the loss value based on the overall loss function; Based on the loss value, feedback is used to adjust the network parameters.

5. The preoperative planning method for artificial intelligence knee ligament reconstruction according to claim 1, wherein: Perform bone correction on the third knee joint CT image to obtain the fourth knee joint CT image, including: Determine the rotation angle; Based on the rotation angle, determine the rotation matrix; The third knee joint CT image is multiplied by the rotation matrix to obtain a rotated fourth knee joint CT image.

6. The preoperative planning method for artificial intelligence knee ligament reconstruction according to claim 1, wherein: The key points in the fourth knee joint CT image are optimized to obtain the fifth knee joint CT image, including: Calculate the difference in length expansion and contraction between point A on the medial posterior superior side of the lateral femoral condyle and point C on the anteromedial 1 / 3 of the tibial intercondylar eminence when the femur and tibia are at different angles; The positions of point A on the medial posterior superior side of the lateral femoral condyle and point C on the anteromedial 1 / 3 of the tibial intercondylar eminence where the length expansion and contraction change difference is the smallest are determined as the optimized positions.

7. The preoperative planning method for artificial intelligence knee ligament reconstruction according to claim 1, wherein: Bone tunnel planning is performed based on the CT image of the fifth knee joint, and the bone tunnel planning results are obtained, including: Bone tunnel planning was performed based on the CT images of the fifth knee joint, and the planning results of three sections of bone tunnels on the tibial side, ligament side, and femoral side were obtained.

8. An artificial intelligence preoperative planning system for knee ligament reconstruction, wherein: The system comprises: An image acquisition module, configured to acquire a first knee joint CT image; The image segmentation module is configured to perform image segmentation on the first knee joint CT image to obtain a second knee joint CT image; wherein the four parts of the femur, tibia, patella and fibula segmented from the second knee joint CT image are displayed respectively; A key point recognition module is configured to perform key point recognition on the second knee joint CT image to obtain a third knee joint CT image; a bone correction module, configured to perform bone correction on the third knee joint CT image to obtain a fourth knee joint CT image; a position optimization module, configured to optimize the positions of key points in the fourth knee joint CT image to obtain a fifth knee joint CT image; A bone tunnel planning module is configured to perform bone tunnel planning based on the fifth knee joint CT image to obtain a bone tunnel planning result; The result display module is configured to perform three-dimensional reconstruction and rendering based on the bone tunnel planning result to display the bone tunnel as a result.

9. An electronic device, wherein: The electronic device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the preoperative planning method for artificial intelligence knee ligament reconstruction as described in any one of claims 1-7.

10. A computer-readable storage medium, wherein: The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the preoperative planning method for artificial intelligence knee ligament reconstruction as described in any one of claims 1-7.

Images