A spinal surgery navigation positioning system design method and related equipment

Abstract

The present application provides a spinal surgery navigation and positioning system design method and related equipment, the method comprising: acquiring a spinal medical image; generating a spinal three-dimensional model based on the spinal medical image; determining a spinal screw placement channel based on the spinal three-dimensional model; generating fitting area data of a spinal surgery navigator based on the spinal three-dimensional model and the screw placement channel; generating the spinal surgery navigator based on the fitting area data. In the present application, a three-dimensional model of the spine is generated to complete the design of the fitting area data and the generation of the spinal surgery navigator, so that the generated spinal surgery navigator is fully adapted to the individual, avoiding lateral sliding of the spinal surgery navigator, and improving the implantation accuracy of pedicle screws and the surgical effect.

Description

Technical Field

[0001] The present application relates to the field of medical device technology, and in particular to a spinal surgery navigation and positioning system design method and related equipment. Background Art

[0002] Spinal pedicle screw implant surgery is a treatment option for fractures, osteoporosis, and spinal vertebral degenerative lesions that appear in clinical practice at this stage. During spinal pedicle screw implant surgery, the implantation of pedicle screws requires the assistance of a surgical navigator.

[0003] In the prior art, the traditional spinal surgery navigator design method is to design different models, and select the corresponding model of spinal surgery navigator for surgery according to the specific conditions of the patient. However, due to the differences in the spines of different individuals, the spinal surgery navigator of adjacent models has poor fit with the bone surface, which makes it easy to slide sideways, and cannot improve the implantation accuracy of the pedicle screws, reducing the surgical effect.

[0004] Therefore, a new design method for a spinal surgery navigator is urgently needed to achieve a complete fit between the spinal surgery navigator and the individual. Summary of the invention

[0005] The problem solved by the present application is that the spinal surgery navigator designed by the existing design method cannot fully fit the individual.

[0006] In order to solve the above problems, the first aspect of the present application provides a spinal surgery navigator design method, which includes:

[0007] Acquire medical images of the spine;

[0008] Based on the spinal medical image, generating a three-dimensional model of the spinal column;

[0009] Determining a screw placement channel of the spine based on the three-dimensional model of the spine;

[0010] Generating fitting area data of a spinal surgery navigator according to the three-dimensional spine model and the nail placement channel;

[0011] The spinal surgery navigator is generated according to the fitting region data.

[0012] The second aspect of the present application provides a spinal surgery navigator, which is designed by the aforementioned design method, and comprises: a fitting portion, a bending portion and at least two nail placement holes;

[0013] The bottom surface of the fitting portion is a fitting curved surface, and the fitting curved surface fits with the fitting area of ​​the spine;

[0014] The nail placement hole is arranged on the fitting portion and is located on the opposite side to the fitting curved surface; the nail placement hole passes through the fitting portion;

[0015] The curved portion is used to connect the two fitting portions and avoid the spinous process of the spine.

[0016] In the present application, a three-dimensional model of the spine is generated to complete the design of the fitting area data and the generation of a spinal surgery navigator, so that the generated spinal surgery navigator is fully adapted to the individual, avoiding lateral sliding of the spinal surgery navigator and improving the implantation accuracy of the pedicle screws and the surgical effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart of a method for designing a spinal surgery navigation and positioning system according to an embodiment of the present application;

[0018] Figure 2 An architectural diagram of a three-dimensional model generated by a spinal surgery navigation and positioning system design method according to an embodiment of the present application;

[0019] Figure 3 It is an architectural diagram of the self-attention structure of the spinal surgery navigation and positioning system design method according to an embodiment of the present application;

[0020] Figure 4 The architecture diagram of the self-attention block of the spinal surgery navigation and positioning system design method according to the embodiment of the present application;

[0021] Figure 5 A schematic diagram of a three-dimensional spine model generated according to an embodiment of the present application;

[0022] Figure 6 A schematic diagram of a nail placement channel determination process in a spinal surgery navigation and positioning system design method according to an embodiment of the present application;

[0023] Figure 7 is a stereoscopic diagram of a spinal surgery navigator according to an embodiment of the present application;

[0024] Figure 8 is a forward schematic diagram of a spinal surgery navigator combined with a spine according to an embodiment of the present application;

[0025] Fig. 9 It is a lateral schematic diagram of the combination of a spinal surgery navigator and a spine according to an embodiment of the present application;

[0026] Fig.10 An architectural diagram of a spinal surgery navigation and positioning system design device according to an embodiment of the present application;

[0027] Fig.11 is an architectural diagram of an electronic device according to an embodiment of the present application.

[0028] Figure numerals: 1-fitting portion; 2-bending portion; 3-nail hole. DETAILED DESCRIPTION

[0029] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application are described in detail below in conjunction with the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0030] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by technicians in the field to which this application belongs.

[0031] In the present application, it should be emphasized that the spinal surgery navigator is the spinal surgery navigation and positioning system, which is a name for different uses of the same object; if it is used in a specific design process, it can be called a spinal surgery navigation and positioning system, and if it is used for a specific object, it can be called a spinal surgery navigator.

[0032] The present application embodiment provides the above-mentioned spinal surgery navigation and positioning system design method, the specific scheme of the method is as follows: Figure 1-Figure 6 As shown, the method can be performed by a design device for a spinal surgery navigation and positioning system, and the design device for the spinal surgery navigation and positioning system can be integrated into electronic devices such as computers, servers, computers, server clusters, and data centers. Figure 1 As shown, the spinal surgery navigation and positioning system design method includes:

[0033] S100, acquiring a spinal medical image;

[0034] In the present application, the spinal medical image may be a medical CT image (Computed Tomography, CT) containing the spine, or may be other divisible medical images such as an MRI medical image.

[0035] S200, generating a three-dimensional model of the spine based on the spinal medical image;

[0036] In the present application, the three-dimensional model of the spine can be a point cloud model or a three-dimensional real-scene model.

[0037] In the present application, the segmented spinal medical image can be generated into a three-dimensional hip joint model through a deep convolution model, or the segmented spinal medical image can be generated into a three-dimensional point cloud model through conventional software such as numpy and mimics.

[0038] S300, determining a screw placement channel of the spine based on the three-dimensional model of the spine;

[0039] S400, generating fitting area data of a spinal surgery navigator according to the spinal three-dimensional model and the nail placement channel;

[0040] In the present application, the nail placement channel is known, and the specific position of the nail placement hole on the spinal surface can be determined. The area surrounding the nail placement hole and extending toward the middle is set, which can be used as the fitting area data.

[0041] S500: Generate the spinal surgery navigator according to the fitting area data.

[0042] In the present application, once the fitting area data and the nail placement holes are determined, the entire spinal surgery navigator can be generated by simply adding a curved portion connecting the two fitting areas based on the pre-selected logic.

[0043] In the present application, a three-dimensional model of the spine is generated to complete the design of the fitting area data and the generation of a spinal surgery navigator, so that the generated spinal surgery navigator is fully adapted to the individual, avoiding lateral sliding of the spinal surgery navigator and improving the implantation accuracy of the pedicle screws and the surgical effect.

[0044] In one embodiment, in combination Figure 6 As shown, the step S300, based on the three-dimensional model of the spine, determines the screw placement channel of the spine, including:

[0045] The spinal vertebrae where the screws are placed are selected, and the three-dimensional image of the spinal vertebrae is cut layer by layer along the coronal plane;

[0046] A pedicle isthmus segmentation image is selected from a plurality of coronal segmentation images after segmentation;

[0047] The screw placement channel of the spine is determined according to the three-dimensional image of the spinal vertebral body and the pedicle isthmus section diagram.

[0048] After obtaining the three-dimensional model of the spine, the spinal vertebrae for screw placement can be selected based on the number or characteristics of the spinal vertebrae. Figure 5 As shown, the dark vertebrae are the selected spinal vertebrae.

[0049] The coronal section refers to the sectioning along the front-to-back direction of the human body in the three-dimensional model, and the human body is sequentially sectioned from front to back into a plurality of planes, and each plane is a coronal sectioning diagram.

[0050] In one embodiment, the segmentation is performed using the distance between two adjacent pixels in the front-to-back direction of the three-dimensional image of the spinal vertebra as the thickness, that is, segmentation is performed according to the minimum resolution unit of the three-dimensional image of the spinal vertebra. The minimum resolution unit is the distance between two adjacent pixels. Segmentation by the minimum resolution unit can achieve more detailed segmentation, thereby improving the precision and accuracy of recognition.

[0051] In the present application, the spinal vertebral body has a pedicle, and the thinnest part of the pedicle is the isthmus. In the present application, it is assumed that the axis of the screw placement / screw placement channel passes through the midpoint of the isthmus of the pedicle.

[0052] In one embodiment, the coronal section image is a binary image. In the present application, by setting a binary image, the computing resources required in the process of determining the nail placement channel are greatly reduced.

[0053] In the present application, the nail placement channel includes the starting point and end point of the nail placement, the nail placement angle, the nail diameter and the nail length;

[0054] In the present application, the nail placement angle can be determined by the angle between the nail and the transverse plane and the sagittal plane specified in the surgical operation specifications.

[0055] In the present application, the starting point of screw placement can be determined by the pedicle isthmus section diagram.

[0056] In the present application, the nail diameter and nail length of the nail placement can be determined based on the selected nail model. When the starting point, nail length, and nail placement angle are known, the end point of the nail placement can be determined.

[0057] In the present application, the method for determining the starting point of nail placement may be:

[0058] Determine two connected domains in the pedicle isthmus segmentation graph; calculate the symmetry center of the minimum circumscribed rectangle of each connected domain; and use the symmetry center of each minimum circumscribed rectangle as the starting point for screw placement.

[0059] In the present application, the minimum circumscribed rectangle of each connected domain can be directly determined based on the data of the connected domain, that is, a figure composed of straight lines / line segments passing through the top, bottom, leftmost and rightmost edge points of the edge of the connected domain.

[0060] In this application, a pedicle isthmus segmentation diagram is selected from a plurality of coronal segmentation diagrams after segmentation, including:

[0061] Identify the connected domain in the coronal segmentation map; determine the pedicle segmentation map according to the connected domain in the coronal segmentation map, wherein the pedicle segmentation map is one of a plurality of coronal segmentation maps having two connected domains that are arranged continuously and the areas of the two connected domains in the pedicle segmentation map are the smallest;

[0062] Combination Figure 6 As shown in the figure, it can be seen that the number of connected domains in different coronal segmentation maps is different, and the number of connected domains corresponds to the three-dimensional image of the spinal vertebra. In the actual execution process, the number of connected domains is converted from two connected domains to one connected domain, then from one connected domain to two connected domains, and finally to three connected domains. The coronal segmentation map encircled by the box in the figure is the one with the smallest area of ​​the two connected domains. The image of the two connected domains distributed on the left and right is the coronal segmentation map of the pedicle.

[0063] The smallest area of ​​the connected domain means that the area of ​​each of the two connected domains is the smallest, or the area of ​​one of the connected domains is the smallest and the area of ​​the other connected domain is the second smallest, or the total area of ​​the two connected domains is the smallest.

[0064] In one embodiment, the fitting area of ​​the spinal surgery navigation and positioning system is extended from the screw placement channel toward the vertebral lamina.

[0065] Combined with follow-up Figure 8 , Fig. 9 As shown, it can be seen that the fitting area of ​​the spinal surgery navigation and positioning system extends from the screw placement channel to the vertebral lamina. This direction is the concave area on the side of the entire vertebral body. Fitting this part can form a local high point / local low point of the entire spinal surgery navigation and positioning system, thereby forming a fixed effect after the spinal surgery navigation and positioning system is engaged.

[0066] In addition, there is less adhesion material in the direction of the vertebral lamina, which can reduce the adhesion area that needs to be cleaned during surgery.

[0067] It should be noted that this portion is also the opening portion in spinal surgery, and a fitting area is formed by extending to this portion, thereby avoiding the consequence of a secondary opening or an enlarged opening to place the surgical navigator.

[0068] In one embodiment, in combination Figure 2 As shown, the S200 generates a three-dimensional spine model based on the spine medical image, including:

[0069] Inputting the spinal medical image into the downsampling structure to obtain a plurality of downsampled feature maps after downsampling;

[0070] Input multiple downsampled feature maps into the improved self-attention structure to obtain the transformed feature map;

[0071] Inputting the conversion feature map into an upsampling structure multiple times to obtain a segmented spinal medical image;

[0072] Generate a segmented three-dimensional model of the spine according to the segmented spinal medical image;

[0073] Among them, the downsampling structure, the upsampling structure, and the improved self-attention structure are trained model structures.

[0074] In this application, the spinal medical image is first segmented using a trained deep convolutional model, and then a spinal three-dimensional model is generated based on the segmented spinal medical image; since the spinal medical image is a segmented image, the generated spinal three-dimensional model is also a segmented model.

[0075] In the present application, the training process of the deep convolution model is the training process of the downsampling structure, the upsampling structure, and the improved self-attention structure. The training can be completed according to the training method in the prior art, and the specific training process will not be repeated in this application.

[0076] In one embodiment, in combination Figure 2 As shown, the spinal medical image is input into the downsampling structure to obtain multiple downsampling feature maps after downsampling, including:

[0077] Inputting the spinal medical image into a patch embedding layer to obtain an embedding feature map;

[0078] Inputting the embedded feature map into a first improved self-attention block to obtain a first downsampled feature map;

[0079] The first downsampled feature map is input into the patch merging layer and the second improved self-attention block in sequence to obtain the second downsampled feature map;

[0080] The second downsampled feature map is input into the patch merging layer and the third improved self-attention block in sequence to obtain the third downsampled feature map;

[0081] The third downsampled feature map is sequentially input into the patch merging layer and the fourth improved self-attention block to obtain the fourth downsampled feature map.

[0082] Combination Figure 2As shown in the figure, the downsampling structure includes a patch embedding layer, an Improved TransformerBlock, and a patch merging layer; among them, there are multiple Improved Transformer Blocks and patch merging layers. After the input image passes through the patch embedding layer, it passes through the Improved Transformer Block, the patch merging layer, the Improved Transformer Block, the patch merging layer, the Improved Transformer Block, the patch merging layer, and the Improved Transformer Block in sequence; among them, the output of each Improved Transformer Block is not only used as the input of the next patch merging layer, but also as the input of the entire Improved Transformer structure (four inputs). Among them, the Improved Transformer Block is the improved self-attention block.

[0083] Among them, the patch merging layer, also known as the convolution-free downsampling technology, is to group the patches in the feature map and splice them by depth, and sample the input data n times.

[0084] In this application, patch merging groups each nxn adjacent patches and concatenates them deeply. This effectively downsamples the input by a factor of n, converting the input from a shape of HxWxC to (H / n)x(W / n)x(2nC), where H, W, and C represent height, width, and channel depth, respectively.

[0085] In this application, the patch embedding layer converts the input image into a series of token sequences; specifically, the image is first divided into patches of fixed size and position, and then they are converted into a token sequence through a linear embedding layer. In fact, this layer is a mechanism for capturing the dependencies between different positions in the sequence. Path embedding introduces position information for different positions in the input sequence by introducing a unique embedding vector for each position, so that the model can distinguish the importance and correlation of different positions.

[0086] In this application, due to the change in image size, the patch merging layer can also be regarded as a layer with a downsampling function.

[0087] It should be noted that in this application, the downsampled feature map is not used as the output of the patch merging layer, but as the output of the improved self-attention block. It is only because of random naming and has no other meaning and will not impose any restrictions.

[0088] In one embodiment, in combination Figure 2As shown, the conversion feature map is input into the upsampling structure multiple times to obtain a segmented spinal medical image, including:

[0089] Inputting the converted feature map into a patch expansion layer to obtain a first expanded feature map;

[0090] Inputting the first expanded feature map and the converted feature map into the first improved self-attention block and the patch expansion layer to obtain a second expanded feature map;

[0091] Inputting the second extended feature map and the converted feature map into the second improved self-attention block and the patch expansion layer to obtain a third extended feature map;

[0092] Inputting the third extended feature map and the converted feature map into the third improved self-attention block and the patch expansion layer to obtain a fourth extended feature map;

[0093] The fourth expanded feature map is input into the linear projection layer to obtain the segmented spine medical image.

[0094] Combination Figure 2 As shown, it can be seen that the Improved Transformer / improved self-attention structure has four outputs, which are respectively used as inputs of three Improved Transformer Blocks / improved self-attention blocks and the first patch expansion layer. The three Improved Transformer Blocks / improved self-attention blocks also have the output of the previous patch expansion layer as input, thereby realizing the entire upsampling.

[0095] It should be noted that the Improved Transformer / improved self-attention structure has four outputs, and all four outputs are the same output.

[0096] In this application, the patch expansion layer refers to dividing the input image into multiple small blocks (patches), and then expanding each small block to form a larger image area. The specific process can be: first, the input image is divided into multiple small blocks (Patches) of equal size, each small block is usually composed of continuous pixels, and can be divided overlappingly or non-overlappingly; each small block is expanded to include nearby pixels, and the expansion method is usually to add a circle of pixels or borders around each small block, so that the size of each small block becomes larger, thereby forming an expanded small block; the expanded small blocks are combined into a larger image area to be input to the improved self-attention block for processing.

[0097] In this application, the purpose of the patch expansion layer is to increase the contextual information of each small patch while retaining the local information of the image, thereby improving the ability of the self-attention block to understand the global structure of the image.

[0098] In this application, due to the change in image size, the patch expansion layer can also be regarded as a layer with upsampling function.

[0099] In this application, combined Figure 2 As shown in Figure 1, linear projection is the process of mapping input data into another space through linear transformation.

[0100] In one embodiment, in combination Figure 3 As shown, the multiple downsampled feature maps are input into the improved self-attention structure to obtain a conversion feature map, including:

[0101] Multiple downsampled feature maps are concatenated and input into the FC layer to obtain a fully connected feature map;

[0102] Perform depth-wise separable convolution on the fully connected feature map to obtain a convolution feature map;

[0103] The fully connected feature map and the convolutional feature map are added and sequentially input into the LN layer, the GELU layer and the FC layer to obtain the conversion feature map.

[0104] It should be noted that in this application, the improved self-attention structure and the improved self-attention block are similar but different structures. The specific architecture of the two structures is described in this application, and they cannot be regarded as the same structure.

[0105] In the present application, the processing process of depthwise separable convolution can be: Depthwise Convolution: First, apply a convolution kernel to each channel of the input feature map for convolution operation. This step only focuses on the feature information within the channel and does not involve information interaction between channels. Depthwise convolution can effectively extract the spatial information of the feature map. Pointwise Convolution: Then, a pointwise convolution is applied to the output of the depthwise convolution, that is, a 1x1 convolution kernel is used to perform convolution operations on all channels. This step aims to integrate and mix the information between channels to obtain the final output feature map.

[0106] In this application, by decomposing the convolution operation into two steps: depthwise convolution and pointwise convolution, depthwise separable convolution can greatly reduce the number of parameters and computational complexity in the model. Compared with traditional convolution, depthwise separable convolution has higher computational efficiency and smaller model size while maintaining model performance.

[0107] In one embodiment, in combination Figure 4 As shown, the processing process of the improved self-attention block includes:

[0108] When there are several input feature maps, the input feature maps are concatenated and input into the FC layer to obtain a connected feature map;

[0109] Perform depth-separable convolution on the connected feature map to obtain a deep feature map;

[0110] After adding the depth feature map and the connection feature map, input it into the LN layer to obtain the first standard feature map;

[0111] After adding the first standard feature map and the connection feature map, the map is input into the LN layer to obtain the second standard feature map;

[0112] After adding the second standard feature map to the connection feature map, the map is input into the LN layer to obtain the output feature map.

[0113] Combination Figure 4 As shown, in this application, when there is only one input feature map, it is directly input into the FC layer for processing; when there are multiple input feature maps, they are spliced ​​and then input into the FC layer for processing. It should be noted here that the number of input feature maps (one or more) in this application refers to the source of the input feature maps (which can also be considered as the arrow mark in the attached figure). If there are multiple sources, the number of feature maps is considered to be multiple, rather than the actual number of feature maps or the number of channels (the feature map itself has the number of channels, and the number of channels may also be understood as the number).

[0114] In the present application, multiple residual processing is performed to associate the overall features of the connected feature map with the features after multiple upsampling, thereby increasing the correlation and consistency between local information and overall information.

[0115] In this application, the existing Transformer Block / Transformer is not suitable for processing high-resolution images due to the problem of computational complexity; existing medical images are generally high-resolution images, and local information needs to be processed during image segmentation. These are the defects of the existing Transformer Block / Transformer.

[0116] In this application, an improved Transformer Block / Transformer is used to simplify the image using patch merging layers and patch expansion layers, thereby reducing the computational complexity; through multiple residual processing methods, and an improved self-attention structure with multiple inputs from different downsampling degrees and multiple outputs to different upsampling degrees, different feature distributions and consistencies of local information and overall information are constructed, thereby solving the problem that the original structure cannot process high-resolution images and is not good at processing local information.

[0117] In this application, multiple inputs from different downsampling levels are spliced ​​and processed with improved self-attention to construct the Great Wall dependency and the correlation of local information.

[0118] The present application provides a spinal surgery navigator. The specific solution of the spinal surgery navigator is as follows: Figure 7-Figure 9 The spinal surgery navigator is designed by the spinal surgery navigator design method described above, and comprises: a fitting portion 1, a bending portion 2 and at least two nail placement holes 3;

[0119] The bottom surface of the fitting portion is a fitting curved surface, and the fitting curved surface fits with the fitting area of ​​the spine;

[0120] The nail placement hole is arranged on the fitting portion and is located on the opposite side to the fitting curved surface; the nail placement hole passes through the fitting portion;

[0121] The curved portion is used to connect the two fitting portions and avoid the spinous process of the spine.

[0122] In the present application, after the spinal surgery navigator is designed, the entity of the spinal surgery navigator can be manufactured by 3D printing.

[0123] In this application, if Figure 7 As shown, the fitting part 1 is a curved element with a certain thickness, and the side facing the spine is a fitting surface, which completely fits the fitting area of ​​the spine; the other side can be a smooth regular surface or any other curved surface structure.

[0124] In the present application, the fitting part 1 and the curved part 2 are non-contiguous elements, and the curved part 2 is fixed to the middle part of the fitting part 1, so that when the fitting part 1 and the vertebral lamina are completely fitted, a certain gap is achieved between the curved part 2 and the spinous process and other parts of the spine. In this way, on the one hand, the gap can facilitate the assembly of the spinal surgery navigator; on the other hand, the gap is convenient for adjusting the thickness of the curved part.

[0125] The curved portion 2 is an inverted U-shaped structure, and the U-shaped opening is narrowed, so as to be combined with the fitting portion 1. It should be noted that since the 3D printed material is an elastic material, when the spinal surgery navigator is assembled to the spine, in order to completely fit the fitting portion 1 to the spine, it is necessary to press downward, and the pressing will cause the curved portion 2 to deform, and the deformation will cause the direction of the nail placement hole 3 to deform to a certain extent.

[0126] In addition, it should be noted that Figure 8 , Fig. 9As shown, the fitting part of the fitting portion 1 is at a relatively low position near the spine, thereby achieving a more fixed setting in terms of mechanics. However, due to the shape changes of different spines in the human body and the differences between different individuals, the position near some spines may not be the lowest position. In addition, the fitting area in this application is determined by a three-dimensional model. Although a complete fit can be achieved on the three-dimensional model, due to the inherent defects of medical images, there is a defect that the similarity between the actual spine and the three-dimensional model is reduced.

[0127] In the present application, by setting the fitting area at a relatively low position near the spine, the spinal surgery navigator can be fixed even when there are defects.

[0128] In the present application, if the U-shaped opening is not narrowed, when the middle part of the curved portion 2 is pressed, the pressing force will be transmitted to the U-shaped opening along with the elements of the curved portion; at this time, since the fitting portion 1 is placed laterally, after receiving the force transmitted in a straight line by the curved portion 2, it will have a tendency to slide laterally (slide outward) along the vertebral arch plate. At this time, if the actual fit between the fitting portion 1 and the spine is low, actual lateral sliding may occur, resulting in deformation of the entire spinal surgical navigator and deflection of the screw holes.

[0129] In the present application, if the U-shaped opening is narrowed, the pressing force will be changed from being transmitted in a straight line from the spinous process to the lamina to being transmitted after bending inward (following the bending angle of the curved portion), so that the tendency of lateral sliding along the lamina (sliding outward) will be greatly reduced, and even if the actual fit between the fitting portion 1 and the spine is low, lateral sliding will not occur. Thus, the spinal surgery navigator is prevented from deforming in a way that affects the screw holes when pressed (at this time, only the curved portion may deform, and the greater the deformation, the more the pressing force generated deviates in the direction perpendicular to the lamina, and the more stable it is).

[0130] Preferably, the U-shaped opening is narrowed, and the entire spinal surgery navigator can be stabilized by pressing the curved portion when the nail is placed.

[0131] In one embodiment, the thickness of the bending portion is 3 mm-12 mm; the thickness of the fitting portion is 2 mm-8 mm.

[0132] In this way, by setting a preset thickness, the space occupied by the spinal surgery navigator can be reduced while avoiding excessive deformation during the use of the spinal surgery navigator.

[0133] In one embodiment, the thickness of the middle portion of the curved portion is greater than the thickness of the two sides of the curved portion, so as to strengthen the middle portion of the curved portion and avoid excessive deformation of the middle portion of the curved portion.

[0134] In one embodiment, the thickness of the middle part of the side of the fitting part away from the spine is greater than the thickness of the edge of the side. In this way, the middle part of the fitting part is strengthened, thereby preventing the nail placement hole and the curved part from being deformed relative to each other, which affects the normal nail placement of the nail placement hole.

[0135] In the present application, the nail placement hole is hollow inside, and the hollow part forms a hole that penetrates the fitting part for placing nails.

[0136] In the present application, the diameter of the nail placement hole corresponds to the model of the nail to be placed; the outer diameter of the nail placement hole is determined according to a preset thickness.

[0137] An embodiment of the present application provides a design device for a spinal surgery navigation and positioning system, which is used to execute the spinal surgery navigation and positioning system design method described in the above content of the present application. The design device for the spinal surgery navigation and positioning system is described in detail below.

[0138] like Fig.10 As shown, the design device of the spinal surgery navigation and positioning system includes:

[0139] An image acquisition module 101 is used to acquire spinal medical images;

[0140] A three-dimensional generation module 102, which is used to generate a three-dimensional model of the spine based on the spinal medical image;

[0141] A screw placement determination module 103, which is used to determine a screw placement channel of the spine based on the three-dimensional model of the spine;

[0142] A fitting generation module 104, which is used to generate fitting area data of a spinal surgery navigation and positioning system according to the spinal three-dimensional model and the screw placement channel;

[0143] The navigator generating module 105 is used to generate the spinal surgery navigation positioning system according to the fitting area data.

[0144] In one implementation, the nail placement determination module 103 is further configured to:

[0145] A spinal vertebra for screw placement is selected, and the three-dimensional image of the spinal vertebra is segmented layer by layer along the coronal plane; a pedicle isthmus segmentation map is selected from the multiple coronal segmentation maps after segmentation; and a spinal screw placement channel is determined based on the three-dimensional image of the spinal vertebra and the pedicle isthmus segmentation map.

[0146] In one embodiment, the fitting area of ​​the spinal surgery navigation and positioning system is extended from the screw placement channel toward the vertebral lamina.

[0147] In one embodiment, the three-dimensional generation module 102 is further used for:

[0148] Input a spinal medical image into a downsampling structure to obtain multiple downsampling feature maps after downsampling; input the multiple downsampling feature maps into an improved self-attention structure to obtain a conversion feature map; input the conversion feature map multiple times into an upsampling structure to obtain a segmented spinal medical image; generate a segmented spinal three-dimensional model based on the segmented spinal medical image; wherein the downsampling structure, the upsampling structure, and the improved self-attention structure are trained model structures.

[0149] In one embodiment, the three-dimensional generation module 102 is further used for:

[0150] The spinal medical image is input into a patch embedding layer to obtain an embedded feature map; the embedded feature map is input into a first improved self-attention block to obtain a first down-sampled feature map; the first down-sampled feature map is sequentially input into a patch merging layer and a second improved self-attention block to obtain a second down-sampled feature map; the second down-sampled feature map is sequentially input into a patch merging layer and a third improved self-attention block to obtain a third down-sampled feature map; the third down-sampled feature map is sequentially input into a patch merging layer and a fourth improved self-attention block to obtain a fourth down-sampled feature map.

[0151] In one embodiment, the three-dimensional generation module 102 is further used for:

[0152] The conversion feature map is input into the patch expansion layer to obtain a first extended feature map; the first extended feature map and the conversion feature map are input into the first improved self-attention block and the patch expansion layer to obtain a second extended feature map; the second extended feature map and the conversion feature map are input into the second improved self-attention block and the patch expansion layer to obtain a third extended feature map; the third extended feature map and the conversion feature map are input into the third improved self-attention block and the patch expansion layer to obtain a fourth extended feature map; the fourth extended feature map is input into the linear projection layer to obtain a segmented spinal medical image.

[0153] In one embodiment, the three-dimensional generation module 102 is further used for:

[0154] Multiple downsampled feature maps are concatenated and input into the FC layer to obtain a fully connected feature map; a depthwise separable convolution is performed on the fully connected feature map to obtain a convolutional feature map; the fully connected feature map and the convolutional feature map are added and input into the LN layer, the GELU layer and the FC layer in sequence to obtain the conversion feature map.

[0155] In one embodiment, the three-dimensional generation module 102 is further used for:

[0156] When there are several input feature maps, the input feature maps are concatenated and input into the FC layer to obtain a connection feature map; the connection feature map is subjected to depthwise separable convolution to obtain a depth feature map; the depth feature map is added to the connection feature map and then input into the LN layer to obtain a first standard feature map; the first standard feature map is added to the connection feature map and then input into the LN layer to obtain a second standard feature map; the second standard feature map is added to the connection feature map and then input into the LN layer to obtain an output feature map.

[0157] The design device of the spinal surgery navigation and positioning system provided in the above-mentioned embodiment of the present application corresponds to the design method of the spinal surgery navigation and positioning system provided in the embodiment of the present application. Therefore, the specific content in the device corresponds to the design method of the spinal surgery navigation and positioning system. The specific content can be referred to the records in the design method of the spinal surgery navigation and positioning system, and will not be repeated in this application.

[0158] The design device of the spinal surgery navigation and positioning system provided in the above-mentioned embodiment of the present application and the design method of the spinal surgery navigation and positioning system provided in the embodiment of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0159] The above describes the internal functions and structures of the design device of the spinal surgery navigation and positioning system. Fig.11 As shown, in practice, the design device of the spinal surgery navigation and positioning system can be realized as an electronic device, including: a memory 301 and a processor 303.

[0160] The memory 301 may be configured to store programs.

[0161] In addition, the memory 301 may also be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device, contact data, phone book data, messages, pictures, videos, etc.

[0162] The 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 memory, flash memory, magnetic disk or optical disk.

[0163] The processor 303 is coupled to the memory 301 and is configured to execute a program in the memory 301 to:

[0164] Acquire medical images of the spine;

[0165] Based on the spinal medical image, generating a three-dimensional model of the spinal column;

[0166] Determining a screw placement channel of the spine based on the three-dimensional model of the spine;

[0167] Generating fitting area data of a spinal surgery navigation and positioning system according to the spinal three-dimensional model and the screw placement channel;

[0168] The spinal surgery navigation and positioning system is generated according to the fitting area data.

[0169] In one implementation, the processor 303 is further configured to:

[0170] A spinal vertebra for screw placement is selected, and the three-dimensional image of the spinal vertebra is segmented layer by layer along the coronal plane; a pedicle isthmus segmentation map is selected from the multiple coronal segmentation maps after segmentation; and a spinal screw placement channel is determined based on the three-dimensional image of the spinal vertebra and the pedicle isthmus segmentation map.

[0171] In one embodiment, the fitting area of ​​the spinal surgery navigation and positioning system is extended from the screw placement channel toward the vertebral lamina.

[0172] In one implementation, the processor 303 is further configured to:

[0173] Input a spinal medical image into a downsampling structure to obtain multiple downsampling feature maps after downsampling; input the multiple downsampling feature maps into an improved self-attention structure to obtain a conversion feature map; input the conversion feature map multiple times into an upsampling structure to obtain a segmented spinal medical image; generate a segmented spinal three-dimensional model based on the segmented spinal medical image; wherein the downsampling structure, the upsampling structure, and the improved self-attention structure are trained model structures.

[0174] In one implementation, the processor 303 is further configured to:

[0175] The spinal medical image is input into a patch embedding layer to obtain an embedded feature map; the embedded feature map is input into a first improved self-attention block to obtain a first down-sampled feature map; the first down-sampled feature map is sequentially input into a patch merging layer and a second improved self-attention block to obtain a second down-sampled feature map; the second down-sampled feature map is sequentially input into a patch merging layer and a third improved self-attention block to obtain a third down-sampled feature map; the third down-sampled feature map is sequentially input into a patch merging layer and a fourth improved self-attention block to obtain a fourth down-sampled feature map.

[0176] In one implementation, the processor 303 is further configured to:

[0177] The conversion feature map is input into the patch expansion layer to obtain a first extended feature map; the first extended feature map and the conversion feature map are input into the first improved self-attention block and the patch expansion layer to obtain a second extended feature map; the second extended feature map and the conversion feature map are input into the second improved self-attention block and the patch expansion layer to obtain a third extended feature map; the third extended feature map and the conversion feature map are input into the third improved self-attention block and the patch expansion layer to obtain a fourth extended feature map; the fourth extended feature map is input into the linear projection layer to obtain a segmented spinal medical image.

[0178] In one implementation, the processor 303 is further configured to:

[0179] Multiple downsampled feature maps are concatenated and input into the FC layer to obtain a fully connected feature map; a depthwise separable convolution is performed on the fully connected feature map to obtain a convolutional feature map; the fully connected feature map and the convolutional feature map are added and input into the LN layer, the GELU layer and the FC layer in sequence to obtain the conversion feature map.

[0180] In one implementation, the processor 303 is further configured to:

[0181] When there are several input feature maps, the input feature maps are concatenated and input into the FC layer to obtain a connection feature map; the connection feature map is subjected to depthwise separable convolution to obtain a depth feature map; the depth feature map is added to the connection feature map and then input into the LN layer to obtain a first standard feature map; the first standard feature map is added to the connection feature map and then input into the LN layer to obtain a second standard feature map; the second standard feature map is added to the connection feature map and then input into the LN layer to obtain an output feature map.

[0182] In this application, Fig.11 Only some components are shown schematically, which does not mean that the electronic device only includes Fig.11 Components shown.

[0183] The electronic device provided in this embodiment is based on the same inventive concept as the spinal surgery navigation and positioning system design method provided in the embodiment of the present application, and has the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0184] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0185] The present application is described 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 process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0186] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0187] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0188] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0189] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0190] The present application also provides a computer-readable storage medium corresponding to the spinal surgery navigation and positioning system design method provided in the aforementioned embodiment, on which a computer program (i.e., program product) is stored. When the computer program is run by a processor, it will execute the spinal surgery navigation and positioning system design method provided in any of the aforementioned embodiments.

[0191] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. 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 technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0192] The computer-readable storage medium provided in the above-mentioned embodiments of the present application and the spinal surgery navigation and positioning system design method provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0193] It should be noted that in the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known structures and technologies are not shown in detail so as not to obscure the understanding of this description.

[0194] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0195] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A spinal surgery navigation and positioning system design method, characterized in that: include: Acquire medical images of the spine; Based on the spinal medical image, generating a three-dimensional model of the spinal column; Determining a screw placement channel of the spine based on the three-dimensional model of the spine; Generating fitting area data of a spinal surgery navigator according to the three-dimensional spine model and the nail placement channel; generating the spinal surgery navigator according to the fitting area data; The step of generating a three-dimensional model of the spine based on the spinal medical image comprises: Inputting the spinal medical image into the downsampling structure to obtain a plurality of downsampled feature maps after downsampling; Input multiple downsampled feature maps into the improved self-attention structure to obtain the transformed feature map; Inputting the conversion feature map into an upsampling structure multiple times to obtain a segmented spinal medical image; Generate a segmented three-dimensional model of the spine according to the segmented spinal medical image; Wherein, the downsampling structure, the upsampling structure, and the improved self-attention structure are trained model structures; The step of inputting multiple downsampled feature maps into the improved self-attention structure to obtain a conversion feature map includes: Multiple downsampled feature maps are concatenated and input into the FC layer to obtain a fully connected feature map; Perform depth-wise separable convolution on the fully connected feature map to obtain a convolution feature map; The fully connected feature map and the convolutional feature map are added and sequentially input into the LN layer, the GELU layer and the FC layer to obtain the conversion feature map; The spinal medical image is input into the downsampling structure to obtain a plurality of downsampling feature maps after downsampling, including: Inputting the spinal medical image into a patch embedding layer to obtain an embedding feature map; Inputting the embedded feature map into a first improved self-attention block to obtain a first downsampled feature map; The first downsampled feature map is input into the patch merging layer and the second improved self-attention block in sequence to obtain the second downsampled feature map; The second downsampled feature map is input into the patch merging layer and the third improved self-attention block in sequence to obtain the third downsampled feature map; The third downsampled feature map is input into the patch merging layer and the fourth improved self-attention block in sequence to obtain the fourth downsampled feature map; The processing of the improved self-attention block includes: When there are several input feature maps, the input feature maps are concatenated and input into the FC layer to obtain a connected feature map; Perform depth-separable convolution on the connected feature map to obtain a deep feature map; After adding the depth feature map and the connection feature map, input it into the LN layer to obtain the first standard feature map; After adding the first standard feature map and the connection feature map, the map is input into the LN layer to obtain the second standard feature map; After adding the second standard feature map to the connection feature map, the map is input into the LN layer to obtain the output feature map; The processing process of the depthwise separable convolution is as follows: first, a convolution kernel is applied to each channel of the input feature map for convolution operation; then, a point-by-point convolution is applied to the output of the depthwise convolution, that is, a convolution operation is performed on all channels using a 1x1 convolution kernel.

2. The design method according to claim 1, characterized in that: The step of determining the screw placement channel of the spine based on the three-dimensional model of the spine includes: The spinal vertebrae where the screws are placed are selected, and the three-dimensional image of the spinal vertebrae is cut layer by layer along the coronal plane; A pedicle isthmus segmentation image is selected from a plurality of coronal segmentation images after segmentation; The screw placement channel of the spine is determined according to the three-dimensional image of the spinal vertebral body and the pedicle isthmus section diagram.

3. The design method according to claim 1, characterized in that: The fitting area of ​​the spinal surgery navigator is obtained by extending from the screw placement channel toward the vertebral lamina.

4. The design method according to any one of claims 1 to 3, characterized in that: The step of inputting the conversion feature map into an upsampling structure multiple times to obtain a segmented spinal medical image comprises: Inputting the converted feature map into a patch expansion layer to obtain a first expanded feature map; Inputting the first expanded feature map and the converted feature map into the first improved self-attention block and the patch expansion layer to obtain a second expanded feature map; Inputting the second extended feature map and the converted feature map into the second improved self-attention block and the patch expansion layer to obtain a third extended feature map; Inputting the third extended feature map and the converted feature map into the third improved self-attention block and the patch expansion layer to obtain a fourth extended feature map; The fourth expanded feature map is input into the linear projection layer to obtain the segmented spine medical image.

Images