Three-dimensional knee joint anatomical structure intelligent perception detection method, device and equipment

Abstract

This application provides a method, device, electronic device, and computer-readable storage medium for intelligently sensing and detecting the anatomical structure of the knee joint in three dimensions. The method comprises: acquiring knee CT data; projecting the knee CT data in the coronal plane to obtain a two-dimensional projection image; inputting the two-dimensional projection image into a preset knee joint key point recognition model to detect the positions of the two-dimensional knee joint key points; and regressing the positions of the three-dimensional knee joint key points based on the positions of the two-dimensional knee joint key points. According to embodiments of the present application, the efficiency of detecting three-dimensional knee joint key points can be improved.

Description

Technical Field

[0001] The present application belongs to the field of deep learning intelligent recognition technology, and in particular relates to a three-dimensional knee joint anatomical structure intelligent perception detection method, device, electronic device and computer-readable storage medium. Background Art

[0002] Currently, the detection of key points of the knee joint in three-dimensional space is done manually, resulting in low detection efficiency.

[0003] Therefore, how to improve the detection efficiency of knee joint key points in three-dimensional space is a technical problem that technical personnel in this field urgently need to solve. Summary of the Invention

[0004] The embodiments of the present application provide a method, device, electronic device and computer-readable storage medium for intelligent perception detection of the anatomical structure of the knee joint in three dimensions, which can improve the detection efficiency of key points of the knee joint in three dimensions.

[0005] In a first aspect, an embodiment of the present application provides a method for intelligently sensing and detecting the anatomical structure of a knee joint in three dimensions, comprising:

[0006] Acquire knee joint CT data;

[0007] Project the knee joint CT data in the coronal plane to obtain a two-dimensional projection image;

[0008] Input the two-dimensional projection image into a preset knee joint key point recognition model to detect the position of the knee joint key points in two-dimensional space;

[0009] Based on the positions of the key points of the knee joint in two-dimensional space, the positions of the key points of the knee joint in three-dimensional space are regressed.

[0010] Optionally, the knee joint CT data is projected in the coronal plane to obtain a two-dimensional projection image, including:

[0011] Calculate the sum of knee joint CT data in the target direction;

[0012] Normalize the sum;

[0013] Perform image format conversion on the normalized sum to obtain a two-dimensional projection image.

[0014] Optionally, the two-dimensional projection image is input into a preset knee joint key point recognition model to detect the positions of the knee joint key points in two-dimensional space, including:

[0015] The detection network in the knee joint key point recognition model is used to detect the preliminary position of the knee joint key points in two-dimensional space;

[0016] The correction network in the knee joint key point recognition model is used to correct the initial position of the knee joint key points in two-dimensional space.

[0017] Optionally, the detection network in the knee joint key point recognition model is used to detect the preliminary positions of the knee joint key points in two-dimensional space, including:

[0018] Directly detect the initial position of visible key points;

[0019] For invisible key points, their initial positions are detected by increasing the receptive field;

[0020] For undetected key points, their preliminary positions are detected through context.

[0021] Optionally, for undetected key points, their preliminary positions are detected through context, including:

[0022] After fusing multiple receptive field information, all feature maps of the same size are concatenated;

[0023] Detect its preliminary position based on the spliced ​​feature map.

[0024] Optionally, based on the positions of the knee joint key points in two-dimensional space, the positions of the knee joint key points in three-dimensional space are regressed, including:

[0025] Draw a horizontal line along the target direction in three-dimensional space from the two-dimensional knee joint key points;

[0026] Calculate the intersection of the horizontal line and the bone mask;

[0027] The outer point of the intersection is determined as the key point of the knee joint in three-dimensional space.

[0028] Optionally, during the model training process of the knee joint key point recognition model, set the training batch_size to 32;

[0029] Set the initial learning rate to 1e-4, and add a learning rate decay strategy. Every 5000 iterations, the learning rate decays to 0.9 of the previous learning rate.

[0030] Set the optimizer to Adam optimizer;

[0031] Set the loss function to DICE loss;

[0032] The training set and validation set were validated once every 1000 iterations. The early stopping method was used to determine the stopping time of network training and obtain the knee joint key point recognition model.

[0033] In a second aspect, an embodiment of the present application provides a three-dimensional intelligent sensing and detection device for the anatomical structure of a knee joint, the device comprising:

[0034] A data acquisition module, used to acquire knee joint CT data;

[0035] A projection module is used to project the knee joint CT data in the coronal plane to obtain a two-dimensional projection image;

[0036] A position detection module is used to input the two-dimensional projection image into a preset knee joint key point recognition model to detect the position of the knee joint key point in two-dimensional space;

[0037] The position regression module is used to regress the positions of the key points of the knee joint in three-dimensional space based on the positions of the key points of the knee joint in two-dimensional space.

[0038] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor and a memory storing computer program instructions;

[0039] When the processor executes the computer program instructions, the three-dimensional knee joint anatomical structure intelligent perception detection method as shown in the first aspect is implemented.

[0040] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the three-dimensional knee joint anatomical structure intelligent perception detection method as shown in the first aspect is implemented.

[0041] The three-dimensional knee joint anatomical structure intelligent perception detection method, device, electronic device and computer-readable storage medium of the embodiments of the present application can improve the detection efficiency of key points of the three-dimensional knee joint.

[0042] The intelligent perception detection method for the three-dimensional knee joint anatomical structure includes: acquiring knee joint CT data; projecting the knee joint CT data in the coronal plane direction to obtain a two-dimensional projection image; inputting the two-dimensional projection image into a preset knee joint key point recognition model to detect the positions of the two-dimensional knee joint key points; and regressing the positions of the three-dimensional knee joint key points based on the positions of the two-dimensional knee joint key points.

[0043] It can be seen that this method is to input the two-dimensional projection image into the preset knee joint key point recognition model, which can automatically detect the position of the knee joint key points in the two-dimensional space; based on the position of the knee joint key points in the two-dimensional space, the position of the knee joint key points in the three-dimensional space can be automatically regressed, thereby improving the detection efficiency of the knee joint key points in the three-dimensional space. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 This is a flow chart of a method for intelligently sensing and detecting the anatomical structure of a knee joint in three dimensions, provided by one embodiment of the present application;

[0046] Figure 2 This is a schematic diagram of the structure of a knee joint key point recognition model provided by an embodiment of the present application;

[0047] Figure 3 This is a schematic diagram of the structure of a correction network provided by an embodiment of the present application;

[0048] Figure 4 This is a schematic diagram of the structure of a three-dimensional knee joint anatomical structure intelligent sensing and detection device provided by one embodiment of the present application;

[0049] Figure 5 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0050] 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.

[0051] 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.

[0052] Currently, the detection of key points of the knee joint in three-dimensional space is done manually, resulting in low detection efficiency.

[0053] To solve the problems of the prior art, the present invention provides a method, apparatus, device, and computer-readable storage medium for intelligently sensing and detecting the anatomical structure of a knee joint in three dimensions. The following first introduces the intelligent sensing and detecting method for the anatomical structure of a knee joint in three dimensions provided by the present invention.

[0054] Figure 1 FIG. 1 is a flow chart showing a method for intelligently sensing and detecting the anatomical structure of a knee joint in three dimensions provided by an embodiment of the present application. Figure 1 As shown, the three-dimensional knee joint anatomical structure intelligent perception detection method includes:

[0055] S101, acquiring knee joint CT data;

[0056] S102, projecting the knee joint CT data in the coronal plane direction to obtain a two-dimensional projection image;

[0057] In one embodiment, the knee joint CT data is projected in the coronal plane to obtain a two-dimensional projection image, including:

[0058] Calculate the sum of knee joint CT data in the target direction;

[0059] Normalize the sum;

[0060] Perform image format conversion on the normalized sum to obtain a two-dimensional projection image.

[0061] Specifically, the extracted volume data is projected in the coronal plane direction, and the projection calculation formula is:

[0062]

[0063]

[0064] Q(i,k)=(255.0*(2*P o (i,k)-P o (i,k)*P o (i,k)))

[0065] Where P(i,k) represents the sum of the three-dimensional data in the y direction, w represents the size of the three-dimensional data in the y direction, i represents the x-direction value, j represents the y-direction value, k represents the z-direction value, max represents the maximum value, Po(i,k) represents the normalization processing of P(i,k), and Q(i,k) represents the conversion of Po(i,k) to an image format in the range of 0-255.

[0066] S103, inputting the two-dimensional projection image into a preset knee joint key point recognition model to detect the positions of the knee joint key points in two-dimensional space;

[0067] In one embodiment, the two-dimensional projection image is input into a preset knee joint key point recognition model to detect the positions of the knee joint key points in two-dimensional space, including:

[0068] The detection network in the knee joint key point recognition model is used to detect the preliminary position of the knee joint key points in two-dimensional space;

[0069] The correction network in the knee joint key point recognition model is used to correct the initial position of the knee joint key points in two-dimensional space.

[0070] In one embodiment, the detection network in the knee joint key point recognition model is used to detect the preliminary positions of the knee joint key points in two-dimensional space, including:

[0071] Directly detect the initial position of visible key points;

[0072] For invisible key points, their initial positions are detected by increasing the receptive field;

[0073] For undetected key points, their preliminary positions are detected through context.

[0074] In one embodiment, for undetected key points, detecting their preliminary positions through context includes:

[0075] After fusing multiple receptive field information, all feature maps of the same size are concatenated;

[0076] Detect its preliminary position based on the spliced ​​feature map.

[0077] Specifically, Figure 2This is a structural diagram of a knee joint key point recognition model provided by an embodiment of the present application. The knee joint key point recognition model consists of two parts: a detection network and a correction network. The detection network performs coarse extraction of key points, that is, uses the ResNet network architecture to regress the heat map, and the correction network finely processes key points that are difficult to identify. The detection network fuses information from different layers, and can better integrate features to locate key points. First, visible key points are directly predicted. For invisible key points, the key point position is obtained by increasing the receptive field. For points that have not been detected, the context is used for prediction, that is, multiple receptive field information is fused. Finally, all feature maps of the same size are spliced ​​to further regress some ambiguous key point positions.

[0078] The detection network adopts a feature pyramid structure similar to FPN, and A 1x1 convolution is added before the network, responsible for detecting all key points in the network, with a focus on those that are relatively easy to detect. The correction network, based on the feature pyramid generated by the detection network, links the pyramid features of all layers to locate difficult key points. The detection network has large prediction errors for key points that are occluded, invisible, or have complex backgrounds. The correction network specifically corrects these points. This structure is based on the idea of ​​skip connections. During this stage of training, a hard example mining strategy is used to increase the recognition of difficult cases and improve the model's generalization ability.

[0079] exist Figure 2 In the detection network, ResNet is used as the feature extraction network, and the network structure is designed as a U-shaped structure. Shallow features such as the first and second layers have high spatial information but low semantic information; deep convolutional layers such as the third and fourth layers have more semantic information due to convolution and pooling but low spatial resolution. The U-shaped structure is specially designed to maintain the spatial and semantic information of the feature layer. At the same time, during the upsampling process, Previously, a 1*1 convolutional layer was used, which was responsible for detecting all key point information.

[0080] The detection network roughly extracts key points, and the correction network finely processes key points that are difficult to identify. The detection network fuses information from different layers to better locate key points based on comprehensive features. Different loss functions are selected for each part of the network to learn their respective network features.

[0081] Loss1 uses the L2 loss function, Loss2 uses the cross entropy loss function, and different networks use different loss functions to learn their respective features.

[0082]

[0083] Among them, loss2(x,y) represents the L2 loss function, yi represents the true value, f(xi) represents the predicted value, and n represents the number of points.

[0084] CELoss=-[ylogy'+(1-y)log(1-y')]

[0085] Among them, CELoss represents the cross entropy loss function, y is the label value, and y' is the predicted value.

[0086] The final loss function is:

[0087] Loss=α·CELoss+(1-α)·loss2

[0088] Among them, α is the loss weight coefficient.

[0089] Figure 3 This is a schematic diagram of the structure of a correction network provided by one embodiment of the present application. To improve information transmission efficiency and ensure information integrity, the correction network transmits information between different layers, integrating upsampled information with each layer. The correction network receives feature information from all layers, rather than passing information between modules solely through the last upsampled feature, as is the case with traditional methods. Its smaller spatial scale allows for a good trade-off between efficiency and performance.

[0090] Before training the knee joint key point recognition model, a knee joint CT dataset was obtained. The CT data was projected in the coronal plane to generate projection images. Key points in the projection images were annotated and corresponding labels were obtained to create a training dataset. The training dataset was shuffled and then divided into a training set, a validation set, and a test set in a 6:2:2 ratio.

[0091] In one embodiment, during the model training process of the knee joint key point recognition model, the batch_size of the training is set to 32;

[0092] Set the initial learning rate to 1e-4, and add a learning rate decay strategy. Every 5000 iterations, the learning rate decays to 0.9 of the previous learning rate.

[0093] Set the optimizer to Adam optimizer;

[0094] Set the loss function to DICE loss;

[0095] The training set and validation set were validated once every 1000 iterations. The early stopping method was used to determine the stopping time of network training and obtain the knee joint key point recognition model.

[0096] S104. Based on the positions of the key points of the knee joint in two-dimensional space, regress the positions of the key points of the knee joint in three-dimensional space.

[0097] In one embodiment, regressing the positions of the knee joint key points in three-dimensional space based on the positions of the knee joint key points in two-dimensional space includes:

[0098] Draw a horizontal line along the target direction in three-dimensional space from the two-dimensional knee joint key points;

[0099] Calculate the intersection of the horizontal line and the bone mask;

[0100] The outer point of the intersection is determined as the key point of the knee joint in three-dimensional space.

[0101] Specifically, 2D to 3D key point regression: Since the key points are identified on the coronal plane of the 3D image, it is necessary to regress the 2D points to the 3D space. Draw a horizontal line along the Y-axis in the 3D space for the identified 2D points, calculate the intersection of the horizontal line and the bone mask, and select the outer point as the desired 3D space point.

[0102] Figure 4 : This is a schematic diagram of the structure of a three-dimensional knee joint anatomical structure intelligent sensing and detection device provided by one embodiment of the present application. The three-dimensional knee joint anatomical structure intelligent sensing and detection device includes:

[0103] Data acquisition module 401, used to acquire knee joint CT data;

[0104] The projection module 402 is used to project the knee joint CT data in the coronal plane direction to obtain a two-dimensional projection image;

[0105] The position detection module 403 is used to input the two-dimensional projection image into a preset knee joint key point recognition model to detect the position of the knee joint key point in two-dimensional space;

[0106] The position regression module 404 is used to regress the positions of the key points of the knee joint in three-dimensional space based on the positions of the key points of the knee joint in two-dimensional space.

[0107] Figure 5 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown.

[0108] The electronic device may include a processor 501 and a memory 502 storing computer program instructions.

[0109] Specifically, the processor 501 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.

[0110] Memory 502 may include a large capacity memory for data or instructions. By way of example and not limitation, memory 502 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 502 may include removable or non-removable (or fixed) media. Where appropriate, memory 502 may be internal or external to the electronic device. In certain embodiments, memory 502 may be a non-volatile solid-state memory.

[0111] In one embodiment, the memory 502 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.

[0112] The processor 501 reads and executes computer program instructions stored in the memory 502 to implement any one of the three-dimensional knee joint anatomical structure intelligent perception detection methods in the above embodiments.

[0113] In one example, the electronic device may further include a communication interface 503 and a bus 510. Figure 5 As shown, the processor 501, the memory 502, and the communication interface 503 are connected via a bus 510 and communicate with each other.

[0114] The communication interface 503 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0115] Bus 510 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 510 can comprise one or more buses.Although the present application embodiment describes and shows specific bus, the application considers any suitable bus or interconnection.

[0116] In addition, in conjunction with the three-dimensional knee joint anatomical structure intelligent sensing and detection methods in the above-mentioned 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 three-dimensional knee joint anatomical structure intelligent sensing and detection methods in the above-mentioned embodiments.

[0117] 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.

[0118] The functional modules shown in the above-described 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), appropriate 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 that are used 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 that can store or transmit 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.

[0119] 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 devices. 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.

[0120] Aspects of the present application have been described above with reference to the flowcharts and / or block diagrams of the methods, devices (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 device to produce a machine so that these instructions executed via the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. This 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.

[0121] 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 three-dimensional intelligent perception detection method for knee joint anatomical structure, characterized in that: include: Acquire knee joint CT data; Projecting the knee joint CT data in a coronal plane direction to obtain a two-dimensional projection image; Inputting the two-dimensional projection image into a preset knee joint key point recognition model to detect the positions of the knee joint key points in two-dimensional space; Based on the positions of the knee joint key points in the two-dimensional space, regressing the positions of the knee joint key points in the three-dimensional space; Projecting the knee joint CT data in the coronal plane direction to obtain a two-dimensional projection image includes: Calculating the sum of the knee joint CT data in a target direction; normalizing the sum; Performing image format conversion on the normalized sum to obtain the two-dimensional projection image; Inputting the two-dimensional projection image into a preset knee joint key point recognition model to detect positions of the knee joint key points in two-dimensional space includes: Utilizing the detection network in the knee joint key point recognition model, detecting preliminary positions of knee joint key points in two-dimensional space; Using the correction network in the knee joint key point recognition model, the preliminary position of the knee joint key point in the two-dimensional space is corrected; The method of using the detection network in the knee joint key point recognition model to detect the preliminary positions of the knee joint key points in two-dimensional space includes: Directly detect the initial position of visible key points; For invisible key points, their initial positions are detected by increasing the receptive field; For key points that have not been detected, their preliminary positions are detected through context; The method of detecting the initial position of undetected key points through context includes: After fusing multiple receptive field information, all feature maps of the same size are concatenated; Detect its preliminary position based on the spliced ​​feature map; The step of regressing the positions of the key points of the knee joint in three-dimensional space based on the positions of the key points of the knee joint in two-dimensional space includes: Draw a horizontal line along the target direction in the three-dimensional space for the two-dimensional knee joint key points; Calculate the intersection of the horizontal line and the skeleton mask; Determining the point on the outside of the intersection as the three-dimensional knee joint key point; During the model training process of the knee joint key point recognition model, the batch_size of the training is set to 32; Set the initial learning rate to 1e-4, and add a learning rate decay strategy. Every 5000 iterations, the learning rate decays to 0.9 of the previous learning rate. Set the optimizer to Adam optimizer; Set the loss function to DICEloss; The training set and the validation set were validated once every 1000 iterations, and the early stopping method was used to determine the stopping time of the network training, and the knee joint key point recognition model was obtained.

2. A three-dimensional intelligent sensing and detection device for knee joint anatomical structure, characterized in that: The device comprises: A data acquisition module, used to acquire knee joint CT data; A projection module, configured to project the knee joint CT data in a coronal plane direction to obtain a two-dimensional projection image; a position detection module, configured to input the two-dimensional projection image into a preset knee joint key point recognition model to detect the positions of the knee joint key points in two-dimensional space; A position regression module, configured to regress the positions of the key points of the knee joint in three-dimensional space based on the positions of the key points of the knee joint in two-dimensional space; Projecting the knee joint CT data in the coronal plane direction to obtain a two-dimensional projection image includes: Calculating the sum of the knee joint CT data in a target direction; normalizing the sum; Performing image format conversion on the normalized sum to obtain the two-dimensional projection image; Inputting the two-dimensional projection image into a preset knee joint key point recognition model to detect positions of the knee joint key points in two-dimensional space includes: Utilizing the detection network in the knee joint key point recognition model, detecting preliminary positions of knee joint key points in two-dimensional space; Using the correction network in the knee joint key point recognition model, the preliminary position of the knee joint key point in the two-dimensional space is corrected; The method of using the detection network in the knee joint key point recognition model to detect the preliminary positions of the knee joint key points in two-dimensional space includes: Directly detect the initial position of visible key points; For invisible key points, their initial positions are detected by increasing the receptive field; For key points that have not been detected, their preliminary positions are detected through context; The method of detecting the initial position of undetected key points through context includes: After fusing multiple receptive field information, all feature maps of the same size are concatenated; Detect its preliminary position based on the spliced ​​feature map; The step of regressing the positions of the key points of the knee joint in three-dimensional space based on the positions of the key points of the knee joint in two-dimensional space includes: Draw a horizontal line along the target direction in the three-dimensional space for the two-dimensional knee joint key points; Calculate the intersection of the horizontal line and the skeleton mask; Determining the point on the outside of the intersection as the three-dimensional knee joint key point; During the model training process of the knee joint key point recognition model, the batch_size of the training is set to 32; Set the initial learning rate to 1e-4, and add a learning rate decay strategy. Every 5000 iterations, the learning rate decays to 0.9 of the previous learning rate. Set the optimizer to Adam optimizer; Set the loss function to DICEloss; The training set and the validation set were validated once every 1000 iterations, and the early stopping method was used to determine the stopping time of the network training, and the knee joint key point recognition model was obtained.

3. An electronic device, characterized in that: The electronic device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the three-dimensional knee joint anatomical structure intelligent perception detection method according to claim 1 is implemented.

4. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the three-dimensional knee joint anatomical structure intelligent perception detection method according to claim 1.

Images