Spinal operation planning method and device based on statistical morphological model

Abstract

The invention provides a spinal surgery planning method and device based on a statistical morphological model. The method comprises the following steps: acquiring CT image data of any object; inputting the CT image data into the improved segmentation network to obtain a segmented spine three-dimensional model; based on the spine three-dimensional model, determining a cortical bone screw plan of a vertebral body; and based on cortical bone screw planning of the vertebral body, determining spinal surgery planning information and guide plate planning information. According to the method, the spine three-dimensional model is constructed through the CT image data, and the cortical bone screw planning and the spine operation planning are performed through the spine three-dimensional model, so that the spine operation planning based on the cortical bone screw is realized.

Description

Technical Field

[0001] This application relates to the field of medical image recognition technology, and more specifically, to a method and apparatus for spinal surgery planning based on a statistical morphological model. Background Technology

[0002] Cortical bone (CBT) screws have extremely high value in the revision of spinal surgery. Although pedicle screws are the most widely used internal fixation devices for the spine, with the increase in revision cases and osteoporosis cases, more and more patients need to use CBT screws for replacement, revision, or to add spinal anchor points to maintain spinal stability.

[0003] However, current spinal surgery planning mainly focuses on pedicle screw placement, lacking spinal surgery planning based on cortical bone screws. Summary of the Invention

[0004] The problem addressed by this application is the current lack of spinal surgery planning based on cortical bone screws.

[0005] To address the aforementioned problems, the first aspect of this application provides a spinal surgery planning method based on a statistical morphological model, comprising:

[0006] Acquire CT image data of any object;

[0007] CT image data is input into an improved segmentation network to obtain a segmented 3D model of the spine.

[0008] Based on a three-dimensional model of the spine, the cortical bone screw planning of the vertebral body is determined;

[0009] Based on the cortical bone screw planning of the vertebral body, spinal surgery planning information and guide plate planning information are determined.

[0010] A second aspect of this application provides a spinal surgery planning device based on a statistical morphological model, comprising:

[0011] The image acquisition module is used to acquire CT image data of any object.

[0012] The image segmentation module is used to input CT image data into an improved segmentation network to obtain a segmented 3D model of the spine.

[0013] The screw planning module is used to determine the cortical bone screw planning of the vertebral body based on a three-dimensional model of the spine.

[0014] The surgical planning module is used for cortical bone screw planning based on the vertebral body, determining spinal surgical planning information and guide plate planning information.

[0015] A third aspect of this application provides an electronic device comprising: a memory and a processor;

[0016] The memory is used to store programs;

[0017] The processor, coupled to the memory, is used to execute the program for:

[0018] Acquire CT image data of any object;

[0019] CT image data is input into an improved segmentation network to obtain a segmented 3D model of the spine.

[0020] Based on a three-dimensional model of the spine, the cortical bone screw planning of the vertebral body is determined;

[0021] Based on the cortical bone screw planning of the vertebral body, spinal surgery planning information and guide plate planning information are determined.

[0022] The fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the above-described method for spinal surgery planning based on a statistical morphological model.

[0023] In this application, a three-dimensional model of the spine is constructed using CT image data, and the three-dimensional model of the spine is used for cortical bone screw planning and spinal surgery planning, thereby realizing spinal surgery planning based on cortical bone screws. Attached Figure Description

[0024] Figure 1 This is a flowchart of a spinal surgery planning method based on a statistical morphological model according to an embodiment of this application;

[0025] Figure 2 An architecture diagram of an improved segmentation network for a spinal surgery planning method based on a statistical morphological model, according to embodiments of this application;

[0026] Figure 3 This is an architecture diagram of the cross-layer fusion module of the spinal surgery planning method based on statistical morphological models according to an embodiment of this application;

[0027] Figure 4 This is a structural block diagram of a spinal surgery planning device based on a statistical morphological model according to an embodiment of this application;

[0028] Figure 5 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0029] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the following description, in conjunction with the accompanying drawings, will illustrate the specific features of this application.

[0030] The specific embodiments are described in detail below. Although exemplary embodiments of this application are shown in the accompanying drawings, it should be understood that this application can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.

[0031] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application pertains.

[0032] Cortical bone (CBT) screws have extremely high value in the revision of spinal surgery. Although pedicle screws are the most widely used internal fixation devices for the spine, with the increase in revision cases and osteoporosis cases, more and more patients need to use CBT screws for replacement, revision, or to add spinal anchor points to maintain spinal stability.

[0033] However, current spinal surgery planning mainly focuses on pedicle screw placement, lacking spinal surgery planning based on cortical bone screws.

[0034] To address the aforementioned issues, this application provides a novel spinal surgery planning scheme based on a statistical morphological model. This scheme can construct a three-dimensional model of the spine using CT image data, and then perform cortical screw planning and spinal surgery planning using the three-dimensional model, thereby achieving spinal surgery planning based on cortical screws.

[0035] This application provides a spinal surgery planning method based on a statistical morphological model. The specific scheme of this method is as follows: Figures 1-3 As shown, this method can be executed by a spinal surgery planning device based on a statistical morphological model, which can be integrated into electronic devices such as computers, servers, computer clusters, and data centers. Figure 1 The diagram shows a flowchart of a spinal surgery planning method based on a statistical morphological model according to an embodiment of this application; wherein the spinal surgery planning method based on the statistical morphological model includes:

[0036] S101, acquire CT image data of any object;

[0037] In this application, CT image data is acquired using a spiral CT scanner with a slice thickness of ≤1mm to ensure high resolution; the scanning range covers the target spinal region (such as T1-L5) and adjacent anatomical structures; and a standard bone reconstruction algorithm is used to reduce metal artifacts.

[0038] S102, Input CT image data into the improved segmentation network to obtain a segmented three-dimensional model of the spine;

[0039] S103, based on the three-dimensional model of the spine, determines the cortical bone screw planning of the vertebral body;

[0040] S104, based on the cortical bone screw planning of the vertebral body, determines the spinal surgery planning information and guide plate planning information.

[0041] In this application, a customized 3D printed guide plate is designed based on the segmented 3D model of the spine and screw trajectory planning.

[0042] In this application, a three-dimensional model of the spine is constructed using CT image data, and the three-dimensional model of the spine is used for cortical bone screw planning and spinal surgery planning, thereby realizing spinal surgery planning based on cortical bone screws.

[0043] In one implementation, it further includes:

[0044] The mechanical properties of cortical bone screw implantation were simulated using a finite element model.

[0045] Based on the simulation results, the cortical bone screw plan was adjusted;

[0046] Based on the adjusted cortical screw plan, the spinal surgery planning information and guide plate planning information are updated.

[0047] In this application, the finite element model is constructed as follows: Model construction: The segmented three-dimensional model of the spine is imported into the finite element analysis software, and material properties (such as the elastic modulus and Poisson's ratio of cortical bone, cancellous bone, and screws) are defined; Boundary conditions are applied: The stress conditions after screw implantation (such as compression, tension, and torsion) are simulated, and constraint conditions (such as fixing the bottom of the vertebral body to simulate a real surgical scenario) are set.

[0048] In this application, the mechanical performance analysis includes: stress distribution analysis based on a model, analyzing the stress distribution in the bone tissue surrounding the screw to identify potential stress concentration areas; and stability assessment, calculating the screw's pull-out force and overall stability. If excessive stress or insufficient stability is found in certain areas, the screw position or parameters are adjusted.

[0049] In this application, based on simulation results, the adjustments to the cortical bone screw planning include: screw trajectory: fine-tuning the angle and depth of the screw to avoid stress concentration or penetration of the anterior edge of the vertebral body; screw specifications: selecting a more suitable length and diameter based on simulation results; number of screws: increasing the number of screws if a single screw cannot provide sufficient stability.

[0050] In this application, spinal surgery planning information and guide plate planning information are regenerated based on the adjusted cortical screw planning to update the original spinal surgery planning information and guide plate planning information.

[0051] In this application, deep learning segmentation and finite element simulation are used to ensure the accuracy and safety of screw planning; surgical plans are customized according to the patient's specific anatomical structure and biomechanical characteristics; and planning information is continuously optimized through simulation and adjustment to reduce intraoperative risks.

[0052] In one implementation, determining the cortical bone screw planning of the vertebral body based on a three-dimensional model of the spine includes:

[0053] Obtain statistical shape models and cortical bone screw planning information;

[0054] Construct a registration matrix between the vertebral body and multiple statistical shape models with corresponding serial numbers;

[0055] Based on the constructed registration matrix, the statistical shape model that is closest to the vertebral body is selected;

[0056] Based on the registration matrix of the closest statistical shape model, the corresponding cortical bone screw planning information is mapped onto the vertebral body.

[0057] In this application, a statistical shape model is used to pre-plan the placement of cortical bone screws, and the pre-planned placement is mapped to the current three-dimensional model of the spine through registration. In this way, pre-planning of the placement only needs to be performed on the statistical shape model, without the need for a placement model. This solves the problem that a large number of cases need to be collected to realize a placement model before cortical bone screw placement can be performed.

[0058] In one implementation, the corresponding cortical screw planning information is mapped onto the vertebral body based on the registration matrix of the closest statistical shape model, including:

[0059] Calculate the registration matrix between each statistical shape model and the point cloud information of the cone;

[0060] Obtain quantitative evaluation indicators;

[0061] Based on quantitative evaluation indicators, the closest statistical shape model is determined;

[0062] Based on the registration matrix, the closest statistical shape model is mapped to the point cloud information of the cone.

[0063] In one implementation, acquiring the statistical shape model and cortical bone screw planning information includes:

[0064] Acquire multiple spinal CT images and segment the voxel data of the corresponding vertebral bodies;

[0065] Convert voxel data into point cloud data of the cone body;

[0066] A statistical shape model of the vertebra is established based on the point cloud data of the vertebra.

[0067] On each statistical shape model, set the cortical bone screw planning information.

[0068] In this application, the statistical shape model of the vertebral body, established based on point cloud data, includes:

[0069] Point cloud alignment (spatial normalization): Translate all cone point clouds to coincide with their centroids to eliminate positional differences; calculate the optimal rotation matrix using singular value decomposition (SVD) to initially align each point cloud with the first sample; calculate the average shape of all point clouds after the current alignment, and realign each sample to the average shape, repeating until the average shape changes.

[0070] <0.1mm.

[0071] Establish point correspondence: Perform template deformation on the registered point cloud: Use the first sample as a template and deform the remaining samples to the template point distribution using a non-rigid ICP algorithm, or use anatomical landmarks as guidance: Manually label key points (such as pedicle apex) and establish full surface point correspondence through thin plate spline interpolation (TPS).

[0072] Constructing the shape matrix: Flatten the 3D point cloud coordinates of each sample into a one-dimensional vector, calculate the average shape vector, and construct a centered shape matrix.

[0073] Principal Component Analysis: Calculate the average shape vector across all training samples; then calculate the covariance matrix and solve the characteristic equation; sort by eigenvalues ​​in descending order and retain the top K principal components (typically covering 95% of the cumulative variance).

[0074] Generate statistical shape models: Construct a formula for expressing statistical shape models, and derive the mean model representing the average cone morphology of the population and the variation pattern model representing the extreme shapes of each principal component.

[0075] In this application, the cortical bone screw planning information can be added manually after the statistical shape model is known.

[0076] In one implementation, combined with Figure 2 As shown, the step of inputting CT image data into an improved segmentation network to obtain a segmented 3D model of the spine includes:

[0077] The CT image data is encoded step by step to obtain four levels of encoded maps;

[0078] The encoding graphs of the second and third layers are fused across layers to obtain the third fused graph;

[0079] The encoding graphs of the third and fourth layers are fused across layers to obtain the fourth fused graph;

[0080] Adaptive adjustment is performed on the coding graph of the fourth layer to obtain the fourth adjusted graph;

[0081] The fourth adjustment diagram, the third fusion diagram, the fourth fusion diagram, and the encoded diagrams of the first and second layers are decoded to obtain a three-dimensional model of the spine.

[0082] In this application, such as Figure 2 As shown, residual blocks are used as the base blocks for encoding and decoding to prevent the loss of detailed features due to network deepening. Specifically, the residual block increases the feature channels through two consecutive 3×3 convolutions, and then a 1×1 convolution merges the information from the residual connections into the convolutional layer.

[0083] In this application, the coding maps of the second and third layers, and the coding maps of the third and fourth layers are fused only through cross-layer fusion, thereby avoiding the surge in data volume caused by the quadratic exponential increase in the transformer structure due to the increase in image size.

[0084] In this application, low-level features and high-level features are fused through cross-layer fusion, thereby enhancing the model's feature capture ability and improving the accuracy and robustness of segmentation.

[0085] In one implementation, combined with Figure 3 As shown, the cross-layer fusion of the coding graphs of the second and third layers to obtain the third fused graph includes:

[0086] Feature embedding is performed on the encoding maps of the second and third layers respectively to obtain the second embedding map and the third embedding map;

[0087] Multi-head attention processing is performed on the downsampled second and third embedding maps to obtain an attention map;

[0088] The attention map is fed forward to obtain the third fusion map.

[0089] In this application, such as Figure 3 As shown, the feature maps of the current layer and the previous layer are obtained. The feature maps are converted into low-dimensional feature vectors through a 2D convolutional layer and then flattened into a 1D sequence. Through multi-head attention, the input sequence is divided into multiple sub-sequences. Then, the attention of the sequences is calculated in different subspaces and finally concatenated together.

[0090] In this application, Cross-Layer Fusion (CFTrans) combines the features extracted from two layers through a dual-input approach and promotes cross-layer feature fusion through a transformer structure, greatly reducing information loss.

[0091] In this application, a multi-head attention method is used to enhance the parallelism and computational efficiency of the model.

[0092] In this application, two-dimensional convolutions are used instead of fully connected layers in the feedforward processing.

[0093] In one implementation, the adaptive adjustment (RPSE) of the coding graph of the fourth layer to obtain the fourth adjusted graph includes:

[0094] The encoded graph is divided into blocks to obtain independent blocks;

[0095] For each independent block, obtain the first and second neighboring blocks with different spacings;

[0096] A first feature block is generated based on the independent block and the first neighboring block;

[0097] A second feature block is generated based on the independent block and the second neighboring block.

[0098] The first and second feature blocks are compressed to obtain a compressed block.

[0099] Iterate through all independent blocks and generate a fourth adjusted graph based on the resulting compressed blocks.

[0100] In this application, the encoded image is divided into blocks, that is, the encoded image is divided into corresponding image blocks by using a checkerboard pattern; wherein, the image block can be at the pixel level (that is, each pixel is an image block) or other levels, and the specific division depends on the actual processing situation.

[0101] In this application, a sliding window or a fixed step size is used to divide the image into blocks of the same size.

[0102] Preferably, in this application, each image block consists of 100-1000 pixels, thereby enabling more feature calculations between local regions while ensuring generation accuracy and reducing computational load.

[0103] In this application, an image block is selected as an independent block. The image blocks above, below, to the left, and to the right of this independent block are the first neighboring blocks; the image blocks one grid away from the top, bottom, left, and right of this independent block are the second neighboring blocks. The spacing between the first and second neighboring blocks and the independent block is different.

[0104] In this application, neighborhood information is extracted for each independent block to capture local structure.

[0105] In this application, generating the first feature block is to generate a local feature representation using an independent block and its first neighboring block. Specifically, this can be done by processing the independent block and the first neighboring block with convolutional layers and attention layers to obtain the first feature block.

[0106] In this application, the specific structure and parameters of the convolutional layer and attention layer can be obtained from the training data or determined according to the actual situation.

[0107] It should be noted that in this application, there are four first neighboring blocks and multiple first feature blocks.

[0108] In this application, the independent block and the first neighboring block are processed by convolutional layers and attention layers to obtain the first feature block. The specific process is as follows: the independent block and four neighboring blocks are concatenated together to form a multi-channel input, and the convolutional layer is used to extract features from the concatenated block; an important feature is enhanced by using a self-attention mechanism or a channel attention mechanism, the attention weight is calculated, and the output of the convolutional layer is weighted to enhance the important feature; the output of the attention layer is split into multiple feature blocks, and each feature block corresponds to the processing result of the independent block and at least one neighboring block.

[0109] In this application, a second feature block is generated to generate a broader local feature representation using the independent block and its second neighboring block. The specific generation process is the same as that of the first feature block, except that the parameters of the convolutional layer and the attention layer are different.

[0110] In this application, the generated feature blocks are compressed into a more compact representation to reduce computational cost while retaining key information. Feature compression is performed using pooling operations (such as max pooling or average pooling) or fully connected layers.

[0111] In this way, multiple first and second feature blocks are compressed into a single compressed block, which corresponds to the size and position of the independent blocks and is used to replace them. All image blocks are replaced by the compressed block, resulting in the adjusted fourth image.

[0112] In this application, each image block of the coded image is traversed to obtain the corresponding compressed block.

[0113] In this application, for image blocks / independent blocks near the edge, their first and second neighboring blocks are incomplete. In this case, the incomplete blocks are completed by copying the first and second neighboring blocks in their relative positions. For example, if the first neighboring block above an independent block does not exist, the first neighboring block below it is copied and used as the block above it.

[0114] In this application, by completing the image blocks, the processing accuracy of adjacent image blocks is greatly improved.

[0115] In this application, an adaptive adjustment module is used to capture the similarity relationship between local regions, thereby enhancing the feature representation.

[0116] This application provides a spinal surgery planning device based on a statistical morphological model, used to execute the spinal surgery planning method based on a statistical morphological model described above. The following is a detailed description of the spinal surgery planning device based on a statistical morphological model.

[0117] like Figure 4 As shown, the spinal surgery planning device based on a statistical morphological model includes:

[0118] Image acquisition module 101 is used to acquire CT image data of any object;

[0119] Image segmentation module 102 is used to input CT image data into an improved segmentation network to obtain a segmented three-dimensional model of the spine.

[0120] Screw planning module 103 is used to determine the cortical bone screw planning of the vertebral body based on a three-dimensional model of the spine;

[0121] Surgical planning module 104 is used for cortical bone screw planning based on the vertebral body to determine spinal surgical planning information and guide plate planning information.

[0122] In one implementation, the surgical planning module 104 is further configured to:

[0123] The mechanical properties of the implanted cortical bone screw were simulated using a finite element model. Based on the simulation results, the cortical bone screw plan was adjusted. Based on the adjusted cortical bone screw plan, the spinal surgery planning information and guide plate planning information were updated.

[0124] In one embodiment, the screw planning module 103 is further configured to:

[0125] Obtain statistical shape models and cortical bone screw planning information; construct a registration matrix between the vertebral body and multiple statistical shape models with corresponding serial numbers; select the statistical shape model that is closest to the vertebral body based on the constructed registration matrix; map the corresponding cortical bone screw planning information onto the vertebral body based on the registration matrix of the closest statistical shape model.

[0126] In one embodiment, the screw planning module 103 is further configured to:

[0127] Multiple spinal CT images were acquired, and voxel data of the corresponding vertebral bodies were segmented. The voxel data were converted into point cloud data of the vertebral bodies. Based on the point cloud data of the vertebral bodies, a statistical shape model of the vertebral body was established. Cortical bone screw planning information was set on each statistical shape model.

[0128] In one embodiment, the image segmentation module 102 is further configured to:

[0129] The CT image data is encoded step by step to obtain four levels of encoded images; the encoded images of the second and third layers are fused across layers to obtain the third fused image; the encoded images of the third and fourth layers are fused across layers to obtain the fourth fused image; the encoded image of the fourth layer is adaptively adjusted to obtain the fourth adjusted image; the fourth adjusted image, the third fused image, the fourth fused image, and the encoded images of the first and second layers are decoded to obtain a three-dimensional model of the spine.

[0130] In one embodiment, the image segmentation module 102 is further configured to:

[0131] Feature embedding is performed on the encoding maps of the second and third layers respectively to obtain the second embedding map and the third embedding map; multi-head attention processing is performed on the downsampled second embedding map and the third embedding map to obtain the attention map; feedforward processing is performed on the attention map to obtain the third fusion map.

[0132] In one embodiment, the image segmentation module 102 is further configured to:

[0133] The encoded graph is divided into blocks to obtain independent blocks; for each independent block, first and second neighboring blocks with different spacings are obtained; a first feature block is generated based on the independent blocks and the first neighboring blocks; a second feature block is generated based on the independent blocks and the second neighboring blocks; the first and second feature blocks are compressed to obtain compressed blocks; all independent blocks are traversed, and an adjusted fourth adjustment graph is generated based on the obtained compressed blocks.

[0134] In one implementation, the surgical planning module 104 is further configured to:

[0135] The mechanical properties of the implanted cortical bone screw were simulated using a finite element model. Based on the simulation results, the cortical bone screw plan was adjusted. Based on the adjusted cortical bone screw plan, the spinal surgery planning information and guide plate planning information were updated.

[0136] In one embodiment, the screw planning module 103 is further configured to:

[0137] Obtain statistical shape models and cortical bone screw planning information; construct a registration matrix between the vertebral body and multiple statistical shape models with corresponding serial numbers; select the statistical shape model that is closest to the vertebral body based on the constructed registration matrix; map the corresponding cortical bone screw planning information onto the vertebral body based on the registration matrix of the closest statistical shape model.

[0138] In one embodiment, the screw planning module 103 is further configured to:

[0139] Multiple spinal CT images were acquired, and voxel data of the corresponding vertebral bodies were segmented. The voxel data were converted into point cloud data of the vertebral bodies. Based on the point cloud data of the vertebral bodies, a statistical shape model of the vertebral body was established. Cortical bone screw planning information was set on each statistical shape model.

[0140] In one embodiment, the image segmentation module 102 is further configured to:

[0141] The CT image data is encoded step by step to obtain four levels of encoded images; the encoded images of the second and third layers are fused across layers to obtain the third fused image; the encoded images of the third and fourth layers are fused across layers to obtain the fourth fused image; the encoded image of the fourth layer is adaptively adjusted to obtain the fourth adjusted image; the fourth adjusted image, the third fused image, the fourth fused image, and the encoded images of the first and second layers are decoded to obtain a three-dimensional model of the spine.

[0142] In one embodiment, the image segmentation module 102 is further configured to:

[0143] Feature embedding is performed on the encoding maps of the second and third layers respectively to obtain the second embedding map and the third embedding map; multi-head attention processing is performed on the downsampled second embedding map and the third embedding map to obtain the attention map; feedforward processing is performed on the attention map to obtain the third fusion map.

[0144] In one embodiment, the image segmentation module 102 is further configured to:

[0145] The encoded graph is divided into blocks to obtain independent blocks; for each independent block, first and second neighboring blocks with different spacings are obtained; a first feature block is generated based on the independent blocks and the first neighboring blocks; a second feature block is generated based on the independent blocks and the second neighboring blocks; the first and second feature blocks are compressed to obtain compressed blocks; all independent blocks are traversed, and an adjusted fourth adjustment graph is generated based on the obtained compressed blocks.

[0146] The spinal surgery planning device based on statistical morphological model provided in the above embodiments of this application and the spinal surgery planning method based on statistical morphological model provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by their stored applications.

[0147] The above describes the internal functions and structure of a spinal surgery planning device based on a statistical morphological model, such as... Figure 5 As shown, in practice, this spinal surgery planning device based on a statistical morphological model can be implemented as an electronic device, including: a memory 301 and a processor 303.

[0148] Memory 301 can be configured to store a program.

[0149] Additionally, memory 301 can also be configured to store various other data to support operation on the electronic device. Examples of such data include instructions for any application or method used to operate on the electronic device, contact data, phonebook data, messages, pictures, videos, etc. Memory 301 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0150] Processor 303, coupled to memory 301, is used to execute programs in memory 301 for:

[0151] Acquire CT image data of any object;

[0152] CT image data is input into an improved segmentation network to obtain a segmented 3D model of the spine.

[0153] Based on a three-dimensional model of the spine, the cortical bone screw planning of the vertebral body is determined;

[0154] Based on the cortical bone screw planning of the vertebral body, spinal surgery planning information and guide plate planning information are determined.

[0155] In one implementation, the processor 303 is further configured to:

[0156] The mechanical properties of the implanted cortical bone screw were simulated using a finite element model. Based on the simulation results, the cortical bone screw plan was adjusted. Based on the adjusted cortical bone screw plan, the spinal surgery planning information and guide plate planning information were updated.

[0157] In one implementation, the processor 303 is further configured to:

[0158] Obtain statistical shape models and cortical bone screw planning information; construct a registration matrix between the vertebral body and multiple statistical shape models with corresponding serial numbers; select the statistical shape model that is closest to the vertebral body based on the constructed registration matrix; map the corresponding cortical bone screw planning information onto the vertebral body based on the registration matrix of the closest statistical shape model.

[0159] In one implementation, the processor 303 is further configured to:

[0160] Multiple spinal CT images were acquired, and voxel data of the corresponding vertebral bodies were segmented. The voxel data were converted into point cloud data of the vertebral bodies. Based on the point cloud data of the vertebral bodies, a statistical shape model of the vertebral body was established. Cortical bone screw planning information was set on each statistical shape model.

[0161] In one implementation, the processor 303 is further configured to:

[0162] The CT image data is encoded step by step to obtain four levels of encoded images; the encoded images of the second and third layers are fused across layers to obtain the third fused image; the encoded images of the third and fourth layers are fused across layers to obtain the fourth fused image; the encoded image of the fourth layer is adaptively adjusted to obtain the fourth adjusted image; the fourth adjusted image, the third fused image, the fourth fused image, and the encoded images of the first and second layers are decoded to obtain a three-dimensional model of the spine.

[0163] In one implementation, the processor 303 is further configured to:

[0164] Feature embedding is performed on the encoding maps of the second and third layers respectively to obtain the second embedding map and the third embedding map; multi-head attention processing is performed on the downsampled second embedding map and the third embedding map to obtain the attention map; feedforward processing is performed on the attention map to obtain the third fusion map.

[0165] In one implementation, the processor 303 is further configured to:

[0166] The encoded graph is divided into blocks to obtain independent blocks; for each independent block, first and second neighboring blocks with different spacings are obtained; a first feature block is generated based on the independent blocks and the first neighboring blocks; a second feature block is generated based on the independent blocks and the second neighboring blocks; the first and second feature blocks are compressed to obtain compressed blocks; all independent blocks are traversed, and an adjusted fourth adjustment graph is generated based on the obtained compressed blocks.

[0167] In this application, Figure 5 The diagram only shows some components and does not mean that the electronic device includes only these components. Figure 5 The components shown.

[0168] The electronic device provided in this embodiment is based on the same inventive concept as the spinal surgery planning method based on statistical morphological models provided in this application embodiment, and has the same beneficial effects as the methods adopted, run or implemented by the application stored therein.

[0169] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0170] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0171] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0172] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0173] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory. Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0174] This application also provides a computer-readable storage medium corresponding to the spinal surgery planning method based on a statistical morphological model provided in the foregoing embodiments, wherein a computer program (i.e., a program product) is stored thereon, and the computer program, when run by a processor, executes the spinal surgery planning method based on a statistical morphological model provided in any of the foregoing embodiments.

[0175] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0176] As defined in this article, computer-readable media do not include transient computer-readable media, such as modulated data signals and carrier waves.

[0177] The computer-readable storage medium provided in the above embodiments of this application and the spinal surgery planning method based on statistical morphological models provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application stored therein.

[0178] It should be noted that numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known structures and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0179] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0180] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A spinal surgery planning method based on a statistical morphological model, characterized in that, include: Acquire CT image data of any object; CT image data is input into an improved segmentation network to obtain a segmented 3D model of the spine. Based on a three-dimensional model of the spine, the cortical bone screw planning of the vertebral body is determined; Based on the cortical bone screw planning of the vertebral body, spinal surgery planning information and guide plate planning information are determined.

2. The spinal surgery planning method based on a statistical morphological model according to claim 1, characterized in that, Also includes: The mechanical properties of cortical bone screw implantation were simulated using a finite element model. Based on the simulation results, the cortical bone screw plan was adjusted; Based on the adjusted cortical screw plan, the spinal surgery planning information and guide plate planning information were updated.

3. The spinal surgery planning method based on a statistical morphological model according to claim 1, characterized in that, The process of determining the cortical bone screw planning of the vertebral body based on a three-dimensional model of the spine includes: Obtain statistical shape models and cortical bone screw planning information; Construct a registration matrix between the vertebral body and multiple statistical shape models with corresponding serial numbers; Based on the constructed registration matrix, the statistical shape model that is closest to the vertebral body is selected; Based on the registration matrix of the closest statistical shape model, the corresponding cortical bone screw planning information is mapped onto the vertebral body.

4. The spinal surgery planning method based on a statistical morphological model according to claim 3, characterized in that, The acquisition of statistical shape model and cortical screw planning information includes: Acquire multiple spinal CT images and segment the voxel data of the corresponding vertebral bodies; Convert voxel data into point cloud data of the cone body; A statistical shape model of the vertebra is established based on the point cloud data of the vertebra. On each statistical shape model, set the cortical bone screw planning information.

5. The spinal surgery planning method based on a statistical morphological model according to claim 1, characterized in that, The step of inputting CT image data into an improved segmentation network to obtain a segmented 3D model of the spine includes: The CT image data is encoded step by step to obtain four levels of encoded maps; The encoding graphs of the second and third layers are fused across layers to obtain the third fused graph; The encoding graphs of the third and fourth layers are fused across layers to obtain the fourth fused graph; Adaptive adjustment is performed on the coding graph of the fourth layer to obtain the fourth adjusted graph; The fourth adjustment diagram, the third fusion diagram, the fourth fusion diagram, and the encoded diagrams of the first and second layers are decoded to obtain a three-dimensional model of the spine.

6. The spinal surgery planning method based on a statistical morphological model according to claim 5, characterized in that, The process of fusing the encoding maps of the second and third layers to obtain the third fused map includes: Feature embedding is performed on the encoding maps of the second and third layers respectively to obtain the second embedding map and the third embedding map; Multi-head attention processing is performed on the downsampled second and third embedding maps to obtain an attention map; The attention map is fed forward to obtain the third fusion map.

7. The spinal surgery planning method based on a statistical morphological model according to claim 5, characterized in that, The adaptive adjustment of the coding graph of the fourth layer to obtain the fourth adjusted graph includes: The encoded graph is divided into blocks to obtain independent blocks; For each independent block, obtain the first and second neighboring blocks with different spacings; A first feature block is generated based on the independent block and the first neighboring block; A second feature block is generated based on the independent block and the second neighboring block. The first and second feature blocks are compressed to obtain a compressed block. Iterate through all independent blocks and generate a fourth adjusted graph based on the resulting compressed blocks.

8. A spinal surgery planning device based on a statistical morphological model, characterized in that, include: The image acquisition module is used to acquire CT image data of any object. The image segmentation module is used to input CT image data into an improved segmentation network to obtain a segmented 3D model of the spine. The screw planning module is used to determine the cortical bone screw planning of the vertebral body based on a three-dimensional model of the spine. The surgical planning module is used for cortical bone screw planning based on the vertebral body, determining spinal surgical planning information and guide plate planning information.

9. An electronic device, characterized in that, include: Memory and processor; The memory is used to store programs; The processor, coupled to the memory, is used to execute the program for: Acquire CT image data of any object; CT image data is input into an improved segmentation network to obtain a segmented 3D model of the spine. Based on a three-dimensional model of the spine, the cortical bone screw planning of the vertebral body is determined; Based on the cortical bone screw planning of the vertebral body, spinal surgery planning information and guide plate planning information are determined.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by a processor to implement a spinal surgery planning method based on a statistical morphological model as described in any one of claims 1-7.

Images