Lamina decompression surgery path planning method and device

Abstract

The application discloses a laminectomy decompression operation path planning method, which comprises the following steps: inputting a CT image of a local spine to be operated into a spatial key point positioning model to position a plurality of spatial key points on the local spine; establishing a spatial coordinate system for the local spine according to the positions of the plurality of spatial key points; determining a plurality of cutting planes required for laminectomy according to the positions of the plurality of spatial key points and the spatial coordinate system; and combining the plurality of cutting planes to form a planning path of laminectomy decompression operation. According to the method, the spatial key points on the local spine are determined by the spatial key point positioning model, then the spatial coordinate system is established according to the spatial key points, the cutting planes are determined according to the positions of the spatial key points and the spatial coordinate system, and finally the operation planning path is formed by combining the cutting planes, so that the laminectomy decompression operation path planning error can be effectively reduced, the operation precision and robustness can be improved, and the injury can be avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical treatment, in particular to a lamina decompression surgery path planning method and device. BACKGROUND

[0002] Lumbar spinal stenosis (LSS) is a degenerative disease. With age, intervertebral discs, ligamentum flavum and facet joints undergo degenerative changes, leading to narrowing of the spinal canal and compression of neural and vascular structures, which in turn can cause back and lower extremity pain, mobility impairment and other disabilities. LSS is the most common reason for patients over 65 years of age to undergo spinal surgery, and laminectomy is a conservative and standard treatment for lumbar spinal stenosis.

[0003] Laminectomy is a delicate surgery because there are some important neural and vascular tissues around the lamina. Once there is tissue damage such as dural sac injury, complications such as nerve root injury and cauda equina syndrome can occur. In order to avoid these injuries as much as possible, surgeons generally proceed from two aspects: 1. Ensure that each cutting item does not penetrate the lamina to avoid damage by bone cutting tools; 2. Reasonably plan the decompression range of laminectomy to avoid the possibility of repeated resection due to too small initial decompression range. The first point requires the surgeon to maintain precise hand muscle control at all times, which is difficult in such a long and difficult surgery; the second point requires the surgeon to be familiar with the lamina anatomy and be able to reasonably plan the decompression range of laminectomy through anatomical markers, which often requires decades of experience, which is difficult to achieve in remote areas.

[0004] The advent of surgical robots provides a new way to solve this problem. However, the surgical robot system itself cannot decide where to cut the lamina, and the surgeon still needs to manually move the end of the bone cutting tool to the target point. Therefore, it is necessary to pre-plan the cutting plane of laminectomy.

[0005] In the prior art, although some semi-automatic and fully automatic laminectomy path planning strategies have appeared, the semi-automatic laminectomy path planning strategy still needs some professional guidance from the surgeon; the fully automatic laminectomy path planning strategy (ALPP) uses artificial intelligence to automatically extract the center point of the lamina from CT data and calculate a reasonable laminectomy trajectory. However, the shape of the lamina is an irregular structure, and the center point is not an accurate anatomical position, so in the output of the center point of the lamina, errors are inevitable, and the robustness is not high. SUMMARY

[0006] The present application aims to at least solve one of the technical problems existing in the prior art.

[0007] Therefore, this invention proposes a method and device for planning surgical pathways for laminectomy, which can effectively reduce the error in planning surgical pathways for laminectomy and improve the accuracy and robustness of the surgery.

[0008] According to a first aspect of this application, a method for planning surgical pathways for laminectomy is provided, comprising the following steps:

[0009] Input the CT image of the spinal region to be operated on into the spatial key point localization model to locate multiple spatial key points on the spinal region.

[0010] A spatial coordinate system for the local area of ​​the spine is established based on the positions of multiple spatial key points obtained from the positioning.

[0011] Multiple cutting planes required for laminectomy are determined based on the positions of multiple spatial key points and the spatial coordinate system.

[0012] The planned path for laminectomy is formed by combining multiple cutting planes.

[0013] In the above method, the spatial key point localization model adopts the SPU-Net network model.

[0014] In the above method, the step of inputting the CT image of the spinal region to be operated on into the spatial key point localization model to locate multiple spatial key points on the spinal region includes:

[0015] Extract feature information from the CT images;

[0016] The CT image after feature information extraction is gradually reduced in size and the number of channels is increased;

[0017] The CT image, after being reduced in size and expanded in the number of channels, is restored and its image features are integrated to obtain the corresponding feature image;

[0018] The feature images of different sizes are unified.

[0019] In the above method, the step of inputting the CT image of the spinal region to be operated on into the spatial key point localization model to locate multiple spatial key points on the spinal region further includes:

[0020] The feature images after being standardized in size are stitched together;

[0021] Obtain thermal images corresponding to each of the aforementioned spatial key points.

[0022] In the above method, the step of inputting the CT image of the spinal region to be operated on into the spatial key point localization model to locate multiple spatial key points on the spinal region further includes:

[0023] Each of the spatial key points is located based on the thermal image corresponding to each of the spatial key points.

[0024] The positioning formula is as follows:

[0025]

[0026] In the formula: P represents a spatial key point; i is an integer greater than or equal to 1; This represents the heatmap corresponding to the i-th channel; argmax represents the function that finds the maximum value in the tensor.

[0027] The locations of the various spatial key points are, respectively, the center point of the anterior edge of the vertebral body A, the center point of the posterior edge of the vertebral body B, the medial edge point of the left pedicle C, the lower edge point D, the lower edge point of the right pedicle E, the medial edge point F, and the center point of the posterior side of the inferior endplate G.

[0028] In the above method, the step of establishing a spatial coordinate system for the local area of ​​the spine based on the positions of the multiple spatial key points obtained from the positioning includes:

[0029] The spatial coordinate system is established by fitting a coordinate system based on the positions of multiple key spatial points.

[0030] by Given a normal vector, establish a virtual plane passing through point B, and project points C, D, E, and F onto the virtual plane, with projection points C', D', E', and F' respectively.

[0031] Calculate the midpoint H of points C' and D', and the midpoint I of points E' and F' respectively;

[0032] With point B as the origin, In the Z-axis direction, Establish the spatial coordinate system along the Y-axis, with the intersections of the Y-axis with C'D' and E'F' being H' and I', respectively.

[0033] In the above method, the step of determining the multiple cutting planes required for laminectomy based on the positions of the multiple spatial key points and the spatial coordinate system includes:

[0034] Based on the positions of multiple spatial key points, three planes parallel to the Z-axis are selected as the cutting planes.

[0035] According to a second aspect of this application, a laminectomy surgical pathway planning device is provided, comprising:

[0036] The spatial key point localization module is used to input CT images of the local area of ​​the spine to be operated on into the spatial key point localization model in order to locate multiple spatial key points on the local area of ​​the spine.

[0037] A spatial coordinate system establishment module is used to establish a spatial coordinate system for the local area of ​​the spine based on the positions of multiple spatial key points obtained from the positioning.

[0038] The cutting plane determination module is used to determine multiple cutting planes required for laminectomy based on the positions of multiple spatial key points and the spatial coordinate system.

[0039] The planning path formation module is used to form a planning path for laminectomy based on the combination of multiple cutting planes.

[0040] According to a third aspect of this application, a terminal is provided, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the above-described laminectomy surgical path planning method when running the computer program.

[0041] According to a fourth aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored computer program, wherein, when the computer program is run by a processor, it controls the terminal where the storage medium is located to execute the above-described laminectomy surgical path planning method.

[0042] According to the technical solution provided in this application, it has at least the following beneficial effects: by determining multiple spatial key points on the local area of ​​the spine through a spatial key point positioning model, establishing a corresponding spatial coordinate system based on the spatial key points, further determining the cutting plane based on the position of the spatial key points and the spatial coordinate system, and finally forming the surgical planning path by combining the cutting planes, it can effectively reduce the planning error of the laminectomy surgical path, which is conducive to improving the accuracy and robustness of the surgery and avoiding accidental injury.

[0043] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description

[0044] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0045] Figure 1 This is a flowchart of the surgical path planning method for laminectomy decompression according to an embodiment of the present invention;

[0046] Figure 2 This is a structural diagram of the SPU-Net according to an embodiment of the present invention;

[0047] Figure 3 This is a workflow diagram of 3D Patch Merging according to an embodiment of the present invention;

[0048] Figure 4 This is a workflow diagram of 3D Patch Expanding according to an embodiment of the present invention;

[0049] Figure 5 This is a distribution diagram of multiple spatial key points on a local part of the spine according to an embodiment of the present invention;

[0050] Figure 6 This refers to the spatial coordinate system corresponding to the local area of ​​the spine in this embodiment of the invention.

[0051] Figure 7 This is a diagram showing the generation of the local spinal cutting path according to an embodiment of the present invention;

[0052] Figure 8 This is a diagram showing the effect of cutting along the local cutting path of the spine according to an embodiment of the present invention.

[0053] Figure 9 This is a structural block diagram of the laminectomy decompression surgical path planning device according to an embodiment of the present invention.

[0054] Figure label:

[0055] Key point positioning module 10; spatial coordinate system establishment module 20; cutting plane determination module 30; planned path formation module 40. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0057] It should be noted that although functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects, not to describe a specific order or sequence. "Several" means one or more; "multiple" means two or more; "greater than," "less than," and "exceeding" are understood to exclude the stated number; "above," "below," and "within" are understood to include the stated number.

[0058] This application aims to establish an SPU-Net network structure and replace pooling operations that may lose information and upsampling operations that lack learning capabilities by adding corresponding functional modules; it also aims to achieve multi-scale feature fusion and, based on this, to achieve accurate localization of key points in the three-dimensional space of the spine; furthermore, it proposes a spatial coordinate system fitting method based on key points and realizes autonomous planning of the laminectomy decompression surgical path based on the spatial relationship between the coordinate system and key points; this can effectively reduce the planning error of the laminectomy decompression surgical path and improve accuracy and robustness.

[0059] like Figure 1 As shown, a first aspect of this application provides a method for planning surgical pathways for laminectomy, the method comprising the following steps:

[0060] Step S100: Input the local CT image of the spine to be operated on into the spatial key point localization model to locate multiple spatial key points on the local spine.

[0061] In this step, to achieve accurate localization of multiple spatial key points on the spine, the Spatial Pyramid Upsampling Network (SPU-Net) model is preferred. This model integrates feature information from multiple scales and has good accuracy and robustness.

[0062] The structure diagram of SPU-Net is as follows: Figure 2 As shown, SPU-Net contains an Encoder and a Decoder. The Encoder is a process that progressively reduces the size of the feature image from top to bottom and gradually expands its number of channels, mainly used to extract the feature information of the image. The Decoder, on the other hand, is a process that progressively restores (expands) the size of the feature image from bottom to top and reduces its number of feature channels, mainly used to restore the size of the feature image reduced by the Encoder and integrate image features to obtain the corresponding feature image.

[0063] Specifically, the Encoder includes a Basic Layer module and a 3D Patch Merging module; the Decoder includes a Basic Layer module, a 3D Patch Expanding module, an SPU (Spatial Pyramid Upsampling module), and an OutLayer (output module); the SPU includes a CRB module and a 3D Patch Expanding module; the OutLayer includes a 3DConv module, a BN module, and a Softmax module. The SPU is used to unify the size of multi-scale feature images to achieve feature fusion; the OutLayer is used to acquire a thermal image H with the same number of keypoints N in the target space.

[0064] It should be noted that 3D Conv represents three-dimensional convolution; BN (Batch Normalizer) represents batch normalization; ReLU represents the ReLU activation function; CBR represents the combination and serial operation of 3D Conv, BN, and ReLU; Rearrange represents the tensor rearrangement operation module; 3D Patch Expanding represents the combination and serial operation of CBR, Rearrange, and BN; SPU represents the combination and serial operation of CBR and 3D Patch Expanding; Softmax represents the Softmax activation function module; OutLayer represents the combination and serial operation of 3D Conv, BN, and Softmax; Concat represents tensor concatenation; Basic Layer represents the combination and serial operation of three CBRs; 3D Patch Merging represents first sampling the tensor to obtain eight smaller tensors, then concatenating them, and then processing them through CBR.

[0065] In this step, 3D Patch Merging and 3D Patch Expanding are used to replace pooling operations that may lose information and upsampling operations that do not have learning capabilities. This compensates for the lack of learning capabilities while ensuring the integrity of information, making the localization of spatial key points more accurate.

[0066] Specifically, 3D Patch Merging is primarily used to implement downsampling (image downsampling) to replace pooling operations. For example... Figure 3 As shown, in 3D Patch Merging, assuming the original feature image has a size (shape), 8 feature images of size are obtained through interval sampling. The feature image is then concatenated along the feature channel dimension to obtain a size of [size missing]. The feature image is then processed using a 3D convolution module and a batch processing normalization module to obtain a size of [size missing]. The feature image is halved. Therefore, the 3D Patch Merging operation in SPU-Net reduces the size of the input feature image by half without affecting the number of feature channels. It should be noted that B represents Batch Size, indicating the number of feature images; C represents the number of feature channels in the feature image; Z represents the thickness of the feature image (size in the Z direction); Y represents the height of the feature image (size in the Y direction); and X represents the width of the feature image (size in the X direction).

[0067] 3D Patch Expanding is primarily used to implement upsampling (enlarging images) to replace upsampling operations that lack learning capabilities. For example... Figure 4 As shown, unlike 3D Patch Merging, 3D Patch Expanding in SPU-Net allows setting the magnification factor. In 3D Patch Expanding, assuming the original feature image size is (B, C, Z, Y, X), a 3D convolution operation expands the number of feature channels. If the feature image size needs to be expanded to S times the input feature image size, the number of feature channels needs to be expanded to S times the original size. 3 Therefore, by expanding the number of feature channels, a feature image of size (B, S × 3 × C, Z, Y, X) can be obtained. Rearranging this image yields a feature image of size (B, C, S × Z, S × Y, S × X). Thus, the 3DPatch Expanding operation in SPU-Net expands the input feature image size by a factor of S without affecting the number of feature channels.

[0068] In the Encoder and Decoder, the Basic Layer module is typically used to extract feature information. The Basic Layer module includes three CBR modules.

[0069] In the Encoder, 3D Patch Merging is used to downsample (reduce) the feature image, thereby expanding the receptive field of the convolutional kernel. It's important to note that the receptive field refers to the area in the input image that a pixel on the feature map output by each layer of the convolutional neural network maps back to. Simply put, a point on the feature map, relative to the size of the original image, represents the area of ​​the input image that the convolutional neural network's features can see. In the Decoder, 3D Patch Expanding is used to upsample (enlarge) the feature image. Furthermore, skip connections are used between the Encoder and Decoder to fuse shallow and deep features.

[0070] Since the various parts of the decoder contain feature information at different scales, the size of different feature images is unified based on the SPU. In the SPU, a CBR module is first used to compress and unify the number of feature channels in each feature image to reduce subsequent computation. After unifying the number of channels in each feature image, 3D PatchExpanding is used to unify the size. Based on this, the feature images of unified size are stitched together along the feature dimension and input into the OutLayer for processing, resulting in a thermal image H with the same number of spatial keypoints N as the target. Finally, by finding the point with the highest brightness in each channel, the corresponding spatial keypoint can be accurately located. For example, for spatial keypoint P... i For i = {1, ..., N}, we only need to The search can be performed within the heatmap corresponding to the i-th channel. The location method is as follows:

[0071]

[0072] Here, argmax represents the function that finds the maximum value in the tensor.

[0073] Step S110: Establish a spatial coordinate system for the local area of ​​the spine based on the positions of multiple spatial key points obtained from the positioning.

[0074] In this step, after locating multiple spatial key points on the selected spine, a spatial coordinate system corresponding to that spine is generated through coordinate system fitting. Taking seven detected spatial key points as an example... Figure 5 As shown, these seven spatial key points are: A, the center point of the anterior edge of the vertebral body; B, the center point of the posterior edge of the vertebral body; C, the medial edge point of the left pedicle; D, the lower edge point; E, the lower edge point of the right pedicle; F, the medial edge point; and G, the center point of the posterior side of the inferior endplate. Figure 6 As shown, firstly Let B be the normal vector. Establish a virtual plane passing through point B, and project points C, D, E, and F onto this plane. The projection points of points C, D, E, and F on this plane are C', D', E', and F', respectively. After projection, calculate the midpoint H of points C' and D', and the midpoint I of points E' and F'. With point B as the origin, In the Z-axis direction, Using the Y-axis as the reference, determine the X-axis direction according to the Cartesian coordinate system and the right-hand rule, establish a personalized spatial coordinate system, and name the intersections of the Y-axis with C'D' and E'F' as H' and I'.

[0075] Step S120: Determine the multiple cutting planes required for laminectomy based on the positions of multiple spatial key points obtained from the positioning and the established spatial coordinate system.

[0076] In this step, after establishing the spatial coordinate system corresponding to the spine, three cutting planes parallel to the Z-axis are selected based on clinical experience to generate the following: Figure 7 The spinal cutting path is shown. Point J is the midpoint between points D' and E'. The distance from point M to the origin B along the Y-axis is 75% of the distance from point C' to the origin B along the Y-axis. The distance from point N to the origin B along the Y-axis is 75% of the distance from point F' to the origin B along the Y-axis. The distance from point K to point J along the X-axis is 40% of the distance from point J to point G along the X-axis. Cutting plane 1 is set perpendicular to the Y-axis, located in the positive direction of the Y-axis, and passes through point M. At this time, cutting plane 1 is perpendicular to the positive direction of the Y-axis. The direction is taken as the direction of its normal vector; set cutting plane 2 to be perpendicular to the Y-axis and located in the negative direction of the Y-axis and passing through point N. At this time, the negative direction of the Y-axis is taken as the direction of the normal vector of cutting plane 2; set cutting plane 3 to be perpendicular to the X-axis and located in the negative direction of the X-axis and passing through point K. At this time, the negative direction of the X-axis is taken as the direction of the normal vector of cutting plane 3; the positions of cutting plane 1, cutting plane 2 and cutting plane 3 can be determined according to their respective normal vectors and the points they pass through.

[0077] Step S130: The planned path for laminectomy is formed based on the determined combination of multiple cutting planes.

[0078] In this step, the planned path for laminectomy is formed by combining cutting plane 1, cutting plane 2, and cutting plane 3 as determined in step S120. The effect of laminectomy following this planned path is as follows: Figure 8 As shown.

[0079] The above-mentioned technical solution employs a path planning method for laminectomy, which replaces pooling operations that may lose information and upsampling operations that lack learning capabilities with 3D Patch Merging and 3DPatch Expanding, respectively. Multi-scale feature fusion is achieved based on the SPU module, enabling precise localization of key points in the three-dimensional space of the spine with high accuracy. A corresponding spatial coordinate system is established based on the precisely located key points through coordinate fitting. The three cutting planes required for laminectomy are determined based on the key points and the established spatial coordinate system, and the combination of these three cutting planes forms the planned path for laminectomy.

[0080] like Figure 9 As shown, a second aspect of this application provides a laminectomy surgical path planning device, which includes a spatial key point positioning module 10, a spatial coordinate system establishment module 20, a cutting plane determination module 30, and a planned path formation module 40.

[0081] Among them, the spatial key point localization module 10 is used to input the CT image of the local area of ​​the spine to be operated on into the spatial key point localization model in order to locate multiple spatial key points on the local area of ​​the spine.

[0082] Spatial coordinate system establishment modulus 20 is used to establish a spatial coordinate system for the local area of ​​the spine based on the positions of multiple spatial key points obtained from the positioning.

[0083] The cutting plane determination module 30 is used to determine multiple cutting planes required for laminectomy based on the positions of multiple spatial key points and the spatial coordinate system.

[0084] The planning path forming module 40 is used to form a planning path for laminectomy based on the combination of multiple cutting planes.

[0085] The aforementioned laminectomy surgical pathway planning device can be implemented using integrated circuits or chips, and no specific limitations are imposed in this application.

[0086] The method and device for planning the surgical path of laminectomy in the above technical solution determine multiple spatial key points on the local area of ​​the spine through a spatial key point positioning model, establish a corresponding spatial coordinate system based on the spatial key points, and further determine the cutting plane based on the position of the spatial key points and the spatial coordinate system. Finally, the surgical path is formed by combining the cutting planes. This method can effectively reduce the error in planning the surgical path of laminectomy, improve the accuracy and robustness of the operation, and avoid accidental injury.

[0087] A third aspect of this application also provides a terminal, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the above-described laminectomy surgical path planning method when running the computer program.

[0088] Specifically, the processor can be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0089] Specifically, the processor connects to the memory via a bus, which may include pathways for transmitting information. The bus can be a PCI bus or an EISA bus, among others. Buses can be categorized as address buses, data buses, control buses, etc.

[0090] The memory may be ROM or other types of static storage devices that can store static information and instructions, RAM or other types of dynamic storage devices that can store information and instructions, or EEPROM, CD-ROM or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.

[0091] Optionally, the memory stores the code of the computer program that executes the scheme of this application, and the execution is controlled by the processor. The processor executes the application code stored in the memory to realize the operation of the above-mentioned laminectomy surgical path planning device.

[0092] A fourth aspect of this application also provides a computer-readable storage medium, which includes a stored computer program, wherein the computer program, when run by a processor, controls the terminal where the storage medium is located to execute the above-described laminectomy surgical path planning method.

[0093] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0094] The above is a detailed description of the preferred embodiments of this application. However, this application is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for planning surgical pathways for laminectomy, characterized in that, Includes the following steps: Step S100: Input the CT image of the local area of ​​the spine to be operated on into the spatial key point localization model to locate multiple spatial key points on the local area of ​​the spine. Step S110: Establish a spatial coordinate system for the local area of ​​the spine based on the positions of the multiple spatial key points obtained from the positioning; Step S120: Determine multiple cutting planes required for laminectomy based on the positions of multiple spatial key points and the spatial coordinate system; Step S130: Forming a planned path for laminectomy surgery based on the combination of multiple cutting planes; wherein step S100 includes: Feature information is extracted from the CT image; the size of the CT image after feature information extraction is gradually reduced and the number of channels is expanded. The CT image, after being reduced in size and expanded in the number of channels, is restored and its image features are integrated to obtain the corresponding feature image; The feature images of different sizes are unified; the unified feature images are stitched together; and thermal images corresponding to each of the spatial key points are obtained. Each of the spatial key points is located based on the thermal image corresponding to each of the spatial key points. The positioning formula is as follows: In the formula: P represents a spatial key point; i is an integer greater than or equal to 1; This represents the heatmap corresponding to the i-th channel; argmax represents the function that finds the maximum value in the tensor. The locations of the aforementioned key spatial points are, respectively, the center point of the anterior edge of the vertebral body A, the center point of the posterior edge of the vertebral body B, the medial edge point of the left pedicle C, the lower edge point D, the lower edge point of the right pedicle E, the medial edge point F, and the center point of the posterior side of the inferior endplate G.

2. The surgical path planning method according to claim 1, characterized in that, The spatial key point localization model adopts the SPU-Net network model.

3. The surgical path planning method according to claim 1, characterized in that, The step of establishing a spatial coordinate system for the local area of ​​the spine based on the positions of the multiple spatial key points obtained from the positioning includes: The spatial coordinate system is established by fitting a coordinate system based on the positions of multiple key spatial points. by Given a normal vector, establish a virtual plane passing through point B, and project points C, D, E, and F onto the virtual plane, with projection points C', D', E', and F' respectively. Calculate the midpoint H of points C' and D', and the midpoint I of points E' and F' respectively; With point B as the origin, In the Z-axis direction, Establish the spatial coordinate system along the Y-axis, with the intersections of the Y-axis with C'D' and E'F' being H' and I', respectively.

4. The surgical path planning method according to claim 3, characterized in that, The step of determining the multiple cutting planes required for laminectomy based on the positions of the multiple spatial key points and the spatial coordinate system includes: Based on the positions of multiple spatial key points, three planes parallel to the Z-axis are selected as the cutting planes.

5. A surgical path planning device for laminectomy, characterized in that, include: The spatial key point localization module is used to input CT images of the local area of ​​the spine to be operated on into the spatial key point localization model in order to locate multiple spatial key points on the local area of ​​the spine. A spatial coordinate system establishment module is used to establish a spatial coordinate system for the local area of ​​the spine based on the positions of multiple spatial key points obtained from the positioning. The cutting plane determination module is used to determine multiple cutting planes required for laminectomy based on the positions of multiple spatial key points and the spatial coordinate system. A path planning module is used to form a planned path for laminectomy based on the combination of multiple cutting planes; The spatial key point localization module includes: Feature information is extracted from the CT image; the size of the CT image after feature information extraction is gradually reduced and the number of channels is expanded. The CT image, after being reduced in size and expanded in the number of channels, is restored and its image features are integrated to obtain the corresponding feature image; The feature images of different sizes are unified; the unified feature images are stitched together; and thermal images corresponding to each of the spatial key points are obtained. Each of the spatial key points is located based on the thermal image corresponding to each of the spatial key points. The positioning formula is as follows: In the formula: P represents a spatial key point; i is an integer greater than or equal to 1; This represents the heatmap corresponding to the i-th channel; argmax represents the function that finds the maximum value in the tensor. The locations of the aforementioned key spatial points are, respectively, the center point of the anterior edge of the vertebral body A, the center point of the posterior edge of the vertebral body B, the medial edge point of the left pedicle C, the lower edge point D, the lower edge point of the right pedicle E, the medial edge point F, and the center point of the posterior side of the inferior endplate G.

6. A terminal comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor runs the computer program, it executes the laminectomy surgical path planning method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein when the computer program is run by a processor, it controls the terminal where the storage medium is located to execute the laminectomy surgical path planning method according to any one of claims 1 to 4.

Images