Method and system for centrum edge detection anomaly repair and centrum shape evaluation

Abstract

The invention discloses a centrum edge detection anomaly repair and centrum form evaluation method and system, electronic equipment and a computer readable storage medium. The centrum edge detection anomaly repair and centrum shape evaluation method comprises the following steps: acquiring spine CT data, and marking each section of centrum edge; performing data preprocessing on the spine CT data; based on the spine CT data after data preprocessing, detecting the edge of each section of vertebral body for the first time by using a vertebral body edge detection network; secondarily repairing the abnormal edge of the centrum by using a centrum edge repairing algorithm; and performing centrum form evaluation according to the spine centrum form. According to the centrum edge detection anomaly repair and centrum shape evaluation method and system, the electronic equipment and the computer readable storage medium, each centrum can be accurately and stably identified, fine detection and anomaly repair are performed on a single vertebra, and the centrum shape is accurately evaluated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of spinal vertebral body, and particularly relates to a method and system for vertebral body edge detection abnormality repair and vertebral body shape evaluation, an electronic device and a computer readable storage medium. BACKGROUND

[0002] With the rapid development of medical imaging technology, CT (computed tomography) imaging plays an increasingly important role in clinical diagnosis. CT can provide high-resolution, three-dimensional image information, providing sufficient information for doctors to diagnose.

[0003] For spinal vertebral body shape evaluation, vertebral body edge detection is crucial. For special cases, there are problems such as unclear boundaries, complex vertebral shape, patient variability, and image contrast changes, which affect vertebral body edge detection and accurate evaluation of vertebral body shape.

[0004] Traditional spinal edge detection algorithms have poor accuracy, especially three-dimensional spinal vertebral edge detection, which is not intelligent enough, and subsequent correction work brings a lot of workload to doctors.

[0005] Existing artificial intelligence spinal vertebral edge detection algorithms have poor generalization performance and are difficult to adapt to complex and variable spinal data, such as spinal compression fractures, spinal scoliosis, and disc herniation.

[0006] Therefore, the present application proposes a method for spinal vertebral edge detection abnormality repair and vertebral body shape evaluation to accurately detect the edge of each vertebral body using a vertebral edge detection network and a vertebral edge repair algorithm, restore the true structure of the vertebral body, and realize vertebral body shape evaluation.

[0007] The results show that the proposed method can accurately and robustly identify each vertebral body and perform fine detection on individual vertebrae, repair abnormalities, and accurately evaluate vertebral body shape. SUMMARY

[0008] The present application provides a method and system for vertebral body edge detection abnormality repair and vertebral body shape evaluation, an electronic device and a computer readable storage medium, which can:

[0009] Improve accuracy: solve the problem of poor three-dimensional spatial spinal vertebral edge detection accuracy.

[0010] Improve generalization ability: suitable for complex and variable spinal data, such as spinal compression fractures, spinal scoliosis, and disc herniation.

[0011] Doctors can intuitively and stereoscopically understand the shape of each vertebral body and visually observe the spatial shape of the spine to make accurate evaluations.

[0012] Efficiency improvement: through automatic identification technology, reduce manual intervention, realize one-key fast and accurate evaluation of vertebral shape, simple and efficient.

[0013] In a first aspect, the application provides a method for repairing abnormal vertebral edge detection and evaluating vertebral shape, comprising:

[0014] Collecting spine CT data and labeling the edge of each vertebral body;

[0015] Data preprocessing is performed on the spine CT data;

[0016] Based on the spine CT data after data preprocessing, the edge of each vertebral body is detected for the first time using a vertebral edge detection network;

[0017] The vertebral edge repair algorithm is used to repair the abnormal edge of the vertebral body for the second time;

[0018] According to the shape of the spine, the shape of the vertebral body is evaluated.

[0019] Optionally, the data preprocessing of the spine CT data comprises:

[0020] Adaptive histogram equalization method is adopted to enhance the contrast of the spine CT image by expanding the intensity value;

[0021] For the contrast-enhanced spine CT image, Gaussian filtering is applied to reduce background noise.

[0022] Optionally, the vertebral edge detection network introduces a multi-scale adaptive spatial attention gate MASAG module and combines a hybrid CNN-Transformer encoding and decoding structure, dynamically adjusts the receptive field and feature fusion in the skip connection, so as to capture local details and global context at the same time, and perform multi-scale feature fusion in the decoding stage, realize the vertebral edge detection of complex spine CT data.

[0023] Optionally, the vertebral edge detection network adopts a U-shaped structure, the left encoder adopts a 4-group MaxViT Block encoding structure, and each group of structure adopts two MaxViT Block combinations;

[0024] The right decoder adopts a hierarchical multi-scale feature fusion mechanism and a DAE-Former Block cross combination form, and the skip connection introduces a MASAG module to enhance the spine vertebral boundary extraction capability.

[0025] Optionally, the shallow layer of the decoder uses a hierarchical multi-scale feature fusion mechanism module to cleverly balance local and global features and channel interaction, overcoming the typical high-resolution image processing challenges related to self-attention;

[0026] The deeper decoder layer utilizes the DAE-Former Block to effectively preserve long-distance dependencies in the low-resolution image while combining spatial and channel attention mechanisms without destroying the two-dimensional structure of the image.

[0027] Optionally, the MASAG module has two inputs, X and G.

[0028] X represents the output of the same layer MaxViT Block module, and G represents the edge filling output result of the Patch Expanding. The two inputs are respectively extracted, fused, and then extracted again, and finally output.

[0029] Optionally, the vertebral body morphology is evaluated according to the morphology of the spinal vertebral body, including:

[0030] The category of each vertebral body is automatically evaluated according to the detected morphology of each vertebral body.

[0031] The category includes: lumbar disc lesion, lumbar bone lesion, lumbar inflammatory lesion, and lumbar traumatic lesion.

[0032] In a second aspect, the present application provides a system for vertebral body edge detection abnormality repair and vertebral body morphology evaluation, comprising:

[0033] A data acquisition module is configured to acquire spinal CT data and label the edge of each vertebral body.

[0034] A data preprocessing module is configured to preprocess the spinal CT data.

[0035] A vertebral body edge detection module is configured to detect the edge of each vertebral body based on the preprocessed spinal CT data using a vertebral body edge detection network.

[0036] An abnormal edge repair module is configured to repair the abnormal edge of the vertebral body using a vertebral body edge repair algorithm.

[0037] A vertebral body morphology evaluation module is configured to evaluate the morphology of the vertebral body according to the morphology of the spinal vertebral body.

[0038] In a third aspect, the present application provides an electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program is executed by the processor to implement the steps of the method for vertebral body edge detection abnormality repair and vertebral body morphology evaluation.

[0039] In a fourth aspect, the present application provides a computer readable storage medium, and a computer program is stored on the computer readable storage medium, and the computer program is executed by a processor to implement the steps of the method for repairing abnormality of vertebral body edge detection and evaluating vertebral body morphology.

[0040] The present application provides a method for repairing abnormality of vertebral body edge detection and evaluating vertebral body morphology, comprising:

[0041] Collecting CT data of a spine, and labeling the edge of each vertebral body;

[0042] Preprocessing the CT data of the spine;

[0043] Based on the CT data of the spine after preprocessing, the edge of each vertebral body is detected for the first time by using a vertebral body edge detection network;

[0044] The abnormal edge of the vertebral body is repaired for the second time by using a vertebral body edge repair algorithm;

[0045] The morphology of the vertebral body is evaluated according to the morphology of the vertebral body of the spine.

[0046] The method for repairing abnormality of vertebral body edge detection and evaluating vertebral body morphology can:

[0047] Improve accuracy: solve the problem of poor accuracy of three-dimensional space vertebral body edge detection.

[0048] Improve generalization ability: suitable for complex and variable spine data, such as spinal compression fracture, scoliosis and disc herniation.

[0049] Doctors can intuitively and stereoscopically understand the morphology of each vertebral body, visually observe the spatial morphology of the spine, and make accurate evaluation.

[0050] Improve efficiency: through automatic identification technology, reduce manual intervention, realize one-key fast and accurate evaluation of vertebral body morphology, and be simple and efficient. BRIEF DESCRIPTION OF DRAWINGS

[0051] The drawings described herein are used to provide further understanding of the present application, and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0052] Figure 1 A flowchart of a method for repairing abnormality of vertebral body edge detection and evaluating vertebral body morphology provided by the embodiments of the present application is shown;

[0053] Figure 2 A comparison result diagram of image preprocessing provided by the embodiments of the present application is shown;

[0054] Figure 3This is a schematic diagram of the vertebral edge detection network provided in the embodiments of this application;

[0055] Figure 4 A schematic diagram of the structure of the multi-scale adaptive spatial attention gate MASAG module provided in the embodiments of this application.

[0056] Figure 5 This is a schematic diagram of the vertebral body edge repair results provided in an embodiment of this application;

[0057] Figure 6 A schematic diagram of the system structure for vertebral edge detection abnormality repair and vertebral morphology assessment provided in an embodiment of this application;

[0058] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0060] As will be known to those skilled in the art, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0061] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application.

[0062] like Figure 1 As shown, this application provides a method for repairing abnormal vertebral body margins and evaluating vertebral body morphology, including:

[0063] S101. Collect spinal CT data and mark the edges of each vertebral body;

[0064] S102. Perform data preprocessing on spinal CT data;

[0065] In one embodiment, data preprocessing of spinal CT data includes:

[0066] An adaptive histogram equalization method is used to enhance the contrast of spinal CT images by expanding the intensity values;

[0067] For the contrast-enhanced spine CT image, Gaussian filtering is applied to reduce background noise.

[0068] Figure 2 The image preprocessing contrast result diagram provided by the embodiment of the present application is shown in Figure 2 (a) and (c) represent the original data CT image, (b) represents the image contrast enhancement result of the adaptive histogram equalization method of (a) image, and (d) represents the result of applying Gaussian filtering to reduce background noise of (c) image.

[0069] S103, based on the spine CT data after data preprocessing, the vertebral body edge detection network is used to detect the edge of each vertebral body for the first time;

[0070] In one embodiment, the vertebral body edge detection network introduces a multi-scale adaptive spatial attention gate MASAG module and combines a hybrid CNN-Transformer encoding and decoding structure, dynamically adjusts the receptive field and feature fusion in the skip connection, so as to capture local details and global context at the same time, and perform multi-scale feature fusion in the decoding stage, realize the vertebral body edge detection of complex spine CT data.

[0071] In one embodiment, the vertebral body edge detection network adopts a U-shaped structure, the left encoder adopts a 4-group MaxViTBlock encoding structure, and each group of structure adopts two MaxViT Block combinations.

[0072] The right decoder adopts a hierarchical multi-scale feature fusion mechanism and a DAE-Former Block cross combination form, and the skip connection introduces a MASAG module to enhance the spine vertebral body boundary extraction capability.

[0073] In one embodiment, the shallow layer of the decoder uses a hierarchical multi-scale feature fusion mechanism module to cleverly balance local and global features and channel interaction, overcoming the typical high-resolution image processing challenges related to self-attention;

[0074] The deeper decoder layer utilizes DAE-Former Block to effectively preserve long-range dependencies in low-resolution images, while combining spatial and channel attention mechanisms without destroying the two-dimensional structure of the image.

[0075] Figure 3 The structure diagram of the vertebral body edge detection network provided by the embodiment of the present application is shown in Figure 3As shown in the network structure, the vertebral body edge detection network adopts a U-shaped structure, the left encoder adopts a 4-group MaxViT Block encoding structure, and each group of structures adopts two MaxViT Block combinations. The right decoder adopts a hierarchical multi-scale feature fusion mechanism and a DAE-FormerBlock cross combination form, and the skip connection introduces a MASAG module to enhance the ability of extracting the boundary of the spinal vertebral body.

[0076] Through the innovative skip connection (MASAG module), the method ensures that spatially relevant features are selectively emphasized while minimizing background interference due to dynamically adjusting the spatial receptive field according to the input from the encoder and the decoder. In addition, MASAG introduces a spatial interaction stage, enabling bidirectional processing between local and global features, thereby providing detailed, context-based information representation for feature maps.

[0077] The shallow layers of the decoder use a hierarchical multi-scale feature fusion mechanism module to cleverly balance local and global features and channel interaction, overcoming typical high-resolution image processing challenges related to self-attention. The deeper decoder layers utilize DAE-Former Block to effectively preserve long-range dependencies in low-resolution images while incorporating spatial and channel attention mechanisms without disrupting the two-dimensional structure of the image. Overall, this approach aims to accurately and contextually aware detection and recognition of anatomical structures in complex medical images.

[0078] At the network skip connection position, a multi-scale adaptive spatial attention gate (MASAG) is inserted. MASAG dynamically adjusts the receptive field and adaptively fuses local and global features according to different levels of feature semantics, effectively alleviating the "semantic misalignment" problem in traditional skip connections.

[0079] Figure 4 The structure diagram of the multi-scale adaptive spatial attention gate MASAG module provided by the embodiments of the present application is shown.

[0080] As Figure 4 shown, the MASAG module has two inputs, X and G;

[0081] X represents the output of the same layer MaxViT Block module, and G represents the edge filling output result of Patch Expanding. The two inputs are respectively extracted, fused, and then extracted again, and finally outputted.

[0082] Such a design can efficiently extract target features, repeatedly filter useful features, and filter out invalid features.

[0083] The purpose of the MaxViT Block module is to address the difficulty in distinguishing between adjacent vertebrae. To address this issue, the MASAG Block module is introduced in the skip connection of the hybrid CNN-Transformer architecture to improve the accuracy of vertebra edge recognition.

[0084] The MASAG module consists of four stages: multi-scale fusion, spatial selection, spatial interaction and cross modulation, and recalibration, which processes the outputs of the encoder and decoder (denoted as X and G, respectively). Multi-scale fusion is used to aggregate X and G in a semantically similar manner, preparing the fused feature maps for subsequent processing. Spatial selection dynamically adjusts the receptive field to prioritize key features, while spatial interaction and cross modulation enrich the feature maps with local details and global context. In the recalibration stage, the dynamic selective attention map generated by the MASAG Block module is used to refine the initial input of the encoder, ensuring spatial accuracy in vertebra edge recognition.

[0085] This approach enables the MASAG module to effectively address various challenges in the field of medical imaging, ensuring accurate and reliable segmentation results.

[0086] MaxViT Block encoding structure:

[0087] The CNN part is used to capture local spatial details; the Transformer part models global contextual dependencies. Using a hybrid CNN-Transformer encoder, the encoder progressively downsamples the input medical image to extract multi-scale features.

[0088] Combining the local perception capabilities of CNN with the global dependency modeling capabilities of Transformer enables the network to capture fine boundaries and model overall anatomical structures in medical image segmentation.

[0089] DAE-Former Block decoding structure:

[0090] Features are progressively upsampled to restore spatial resolution, and multi-scale features processed by MASAG are fused at each stage. Through a hierarchical multi-scale fusion mechanism, boundary and detail information is enhanced.

[0091] Hierarchical multi-scale feature fusion mechanism (LKA Block):

[0092] Features are progressively upsampled to restore spatial resolution; multi-scale features processed by MASAG are fused at each stage. Through a hierarchical multi-scale fusion mechanism, boundary and detail information is enhanced.

[0093] Output layer (Segmentation Head):

[0094] The last layer of convolution maps the results of the spine vertebra edge detection. The segmentation head calculates as follows, f(x i ) represents the calculation results:

[0095]

[0096] Where xi is the original output probability of the i-th class, and C is the total number of classes. e is the base of the natural logarithm.

[0097] The gradient of the output layer is usually used in backpropagation to update the weights of the model. For class i, the gradient of f(x i ) can be expressed as:

[0098]

[0099] Where δ ij is an inner chain binary function that is 1 if i = j; otherwise 0, and f(x j ) is the probability of the j-th class output by f(x i ).

[0100] For the loss function of vertebra boundary detection, the cross-entropy loss function is as follows:

[0101]

[0102] Where L pBEC represents the cross-entropy loss function, y c and p c represent the true label and predicted probability map belonging to each section of the vertebra edge, respectively. For the labeled class i, the value of y c is 0 or 1, indicating whether the pixel belongs to the class; for the unlabeled class, the value of y c is -1, and c represents the number of vertebra classes.

[0103] The overall loss function is represented as:

[0104] L = L labeled + L unlabeled

[0105] = L(X w , Y l ) + L(X s , Y l s = L pBEC (X w , Y l ) + L(X s , Y l s )

[0106] where L labeled represents the vertebral body edge pixel point, L unlabeled represents the non-vertebral body edge pixel point, X w represents that the input image is randomly rotated and scaled to obtain a weakly enhanced image, X s represents that X w a strongly enhanced image Y is obtained by cross-set data augmentation. l represents the true vertebral body edge result, Y l s represents the pseudo label, i.e. the pixel value within 3 pixels of the true label.

[0107]

[0108] where P' represents the label set other than the pseudo label, the parameters α = 0.1, λ1 = 0.001, and λ2 = 0.01.

[0109] In order to solve the problem of recognition error between adjacent vertebral body boundaries, online clustering is used to generate K prototypes for each class c The prototype is updated at each iteration using a momentum coefficient u = 0.999, i.e.

[0110]

[0111] where, represents the prototype currently calculated, and the edge detection prediction of the prototype classifier is obtained by calculating the distance between each pixel and the prototype, i.e. p(c|i), and the formula is as follows:

[0112]

[0113] where i represents the normalized embedding of pixel i, and <i,i> represents the distance metric, represents the distance of the embedding of pixel i to its nearest prototype.

[0114] S104, repairing the abnormal vertebral body edge twice using the vertebral body edge repair algorithm.

[0115] The vertebral body edge repair algorithm is described as follows:

[0116] For each vertebral body edge detection result, the vertebral body edge repair algorithm is used to optimize the previously obtained vertebral body edge detection result. The spinal vertebral body MRI image Ip and each vertebral body edge detection result Yseg are repaired using the following vertebral body edge repair algorithm:

[0117]

[0118] Where ψ represents the level set function, H represents the three-dimensional step function, and G... σ Let g(x) represent the Gaussian convolution kernel, g(x) represent the edge detector, δ represent the regularization function, and Y represent the regularization function. seg This indicates the detection results for the edges of each vertebral body, where Ip represents the MRI image of the vertebral body of the spine. ∠ represents the calculation range of the boundary of each vertebral body, ▽ represents the gradient operator, Ω represents the differential equation, and λ represents the weight probability.

[0119] To calculate the energy function Minimization element Solve the following equations to obtain the optimal boundary state, and take the derivatives as follows:

[0120]

[0121] In the last term of Equation 1, the vertebral edge repair algorithm is applied to the vertebral edge detection result Yseg, serving as the initial vertebral edge detection result and ensuring high fidelity for the target detection result. The key role of this last term is to minimize Equation 1. The level set is very close to Y seg The edge of the vertebral body. For the first three terms of Formula 1, the contour is pushed to the target region using distance regularization and external energy. Formula 1. The result of the vertebral body edge repair is as follows. Figure 5 As shown.

[0122] S105. Assess vertebral morphology based on the morphology of the vertebral body.

[0123] In one embodiment, vertebral body morphology assessment based on spinal vertebral body morphology includes:

[0124] The system automatically assesses the category of each vertebra by detecting its morphology.

[0125] The categories include: lumbar disc herniation, lumbar bone disease, lumbar inflammatory disease, and lumbar traumatic disease.

[0126] Specifically, by detecting the morphology of each vertebra, the system automatically assesses which category the vertebra belongs to:

[0127] 1. Lumbar disc disease (such as lumbar disc herniation, lumbar disc bulging);

[0128] 2. Lumbar spine bone diseases (such as lumbar osteophyte formation, lumbar osteoporosis);

[0129] 3. Inflammatory lesions of the lumbar spine (such as lumbar tuberculosis, lumbar purulent inflammation);

[0130] 4. Traumatic lesions of the lumbar spine (such as lumbar fractures).

[0131] Figure 6A system structure schematic diagram for vertebral body edge detection abnormality repair and vertebral body shape evaluation provided by an embodiment of the present application is shown in FIG. 1. The system for vertebral body edge detection abnormality repair and vertebral body shape evaluation includes: Figure 6

[0132] A data acquisition module 601 is configured to acquire spinal CT data and label the edge of each vertebral body.

[0133] A data preprocessing module 602 is configured to perform data preprocessing on the spinal CT data.

[0134] A vertebral body edge detection module 603 is configured to detect the edge of each vertebral body based on the data-preprocessed spinal CT data using a vertebral body edge detection network.

[0135] An abnormal edge repair module 604 is configured to repair the abnormal edge of the vertebral body using a vertebral body edge repair algorithm.

[0136] A vertebral body shape evaluation module 605 is configured to perform vertebral body shape evaluation according to the shape of the spinal vertebral body.

[0137] Figure 7 A hardware structure schematic diagram of an electronic device for implementing various embodiments of the present application can include a processor 701 and a memory 702 having computer program instructions stored therein. Specifically, the processor 701 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the present application.

[0138] The memory 702 can include a mass storage for data or instructions. By way of example and not limitation, the memory 702 can include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 702 can include removable or non-removable (or fixed) media. Where appropriate, the memory 702 can be internal or external to the electronic device. In certain embodiments, the memory 702 can be a non-volatile solid-state memory.

[0139] ​In one embodiment, the memory 702 can be a Read Only Memory (ROM). In one embodiment, the ROM can be a mask programmed ROM, a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable PROM (EEPROM), an Electrically Alterable ROM (EAROM), or a flash memory, or a combination of two or more of these.

[0140] The processor 701 implements the method of any one of the above embodiments of the vertebral edge detection abnormality repair and the vertebral morphology evaluation by reading and executing the computer program instructions stored in the memory 702.

[0141] In one example, the electronic device can further include a communication interface 703 and a bus 710. As shown, the processor 701, the memory 702, and the communication interface 703 are connected through the bus 710 and complete communication with each other. Figure 7

[0142] The communication interface 703 is mainly used to realize the communication between the modules, devices, units and / or equipment in the embodiments of the present application.

[0143] The bus 710 includes hardware, software or both to couple the components of the electronic device to each other. By way of example and not limitation, the bus can include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association local (VLB) bus, or other suitable bus or combination of two or more of these. The bus 710 can include one or more buses, as appropriate. Although particular buses are described and shown in the embodiments of the present application, the present application contemplates any suitable bus or interconnect.

[0144] In addition, in combination with the spinal disc plane fitting method based on the discrete distance field model in the above embodiments, the embodiments of the present application can provide a computer readable storage medium to implement. The computer readable storage medium has computer program instructions stored thereon; the computer program instructions are executed by the processor to implement the method of any one of the above embodiments of the vertebral edge detection abnormality repair and the vertebral morphology evaluation.

[0145] ​It is to be understood that the application is not limited to particular configurations and process described herein and shown in the drawings. Detailed descriptions of known methods are omitted so as not to obscure the description of the present application. In the above embodiments, several specific steps are described and illustrated in order to provide a thorough disclosure of the application. However, the method process of the application can be performed in a number of different specific sequences and steps other than those described and illustrated. Various changes, modifications and additions can be made to the method process of the application by those skilled in the art once they learn of the basic inventive concepts disclosed herein.

[0146] The above description merely provides example embodiments of the application, and is not intended to limit the application. Based on the description above, a person skilled in the art can clearly understand the specific working process of the above-described system, modules and units. Therefore, the detailed description of the working process of the above-described system, modules and units is omitted here.

[0147] In addition, those skilled in the art will appreciate that embodiments of the application can be provided as a method, a system, or a computer program product. Therefore, the application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0148] The application is described with reference to the flowcharts and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus with a function specified in one or more flows and / or blocks.

[0149] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction apparatus that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus with a function specified in one or more flows and / or blocks.

[0150] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 Figure 1

[0151] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0152] The memory can include non-persistent memory and / or volatile memory, such as random access memory (RAM) about which the processor can execute instructions. The memory can also include non-volatile memory, such as read only memory (ROM), electrically programmable read only memory (EPROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), erasable programmable read only memory (EPROM), flash memory, or a combination of non-volatile memories in different types. The memory is an example of computer readable storage media.

[0153] Computer readable media includes permanent and non-permanent, removable and non-removable media implemented in any method or technology for storage of information such as 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 programmable read only memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital versatile disc (DVD), or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to computing devices. According to the definition herein, computer readable media does not include transitory media such as modulated data signals and carrier waves.

[0154] It should also be noted that the terms "comprising", "comprises", "including", "includes" or any other variation thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or apparatus that comprises a list of elements does not include only those elements recited, but can also include other elements not expressly listed or inherent to such process, method, article or apparatus. Without further limitation, an element preceded by "comprises a" does not, without more constraints, foreclose the existence of additional identical elements in the process, method, article or apparatus that includes the recited element.

[0155] ​​The above merely provides an example of the present application, but is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the scope of claims of the present application.

Claims

1. A method for detecting and repairing abnormalities at the vertebral body edges and for evaluating vertebral body morphology, characterized in that, The application relates to a spinal CT data processing method and device. The application comprises the following steps: Collecting spinal CT data and marking the edges of each vertebra; Data preprocessing is performed on the spinal CT data; Based on the data preprocessed spinal CT data, the edges of each vertebra are detected for the first time by using a vertebra edge detection network; The abnormal edges of the vertebra are repaired for the second time by using a vertebra edge repair algorithm; 2. The method of vertebral endplate detection of abnormal repair and vertebral morphology assessment of claim 1, wherein, The shape of the vertebra is evaluated according to the shape of the spinal vertebra. The data preprocessing of the spinal CT data comprises the following steps: An adaptive histogram equalization method is adopted to enhance the contrast of the spinal CT image by expanding the intensity value; 3. The method of vertebral endplate detection anomaly repair and vertebral morphology assessment of claim 2, wherein, For the contrast-enhanced spinal CT image, a Gaussian filter is applied to reduce background noise.

4. The method of vertebral endplate detection anomaly repair and vertebral morphology assessment of claim 3, wherein, The vertebra edge detection network introduces a multi-scale adaptive spatial attention gate (MASAG) module and combines a hybrid CNN-Transformer encoding-decoding structure, dynamically adjusts the receptive field and feature fusion in the jump connection, thereby simultaneously capturing local details and global context, and performing multi-scale feature fusion in the decoding stage to realize vertebra edge detection on complex spinal CT data. The vertebra edge detection network adopts a U-shaped structure, the left encoder adopts a 4-group MaxViT Block encoding structure, and each group of structures adopts two MaxViT Block combinations; 5. The method of vertebral endplate detection anomaly repair and vertebral morphology assessment of claim 4, wherein, The right decoder adopts a hierarchical multi-scale feature fusion mechanism and a DAE-Former Block cross combination form, and the jump connection introduces a MASAG module to enhance the spinal vertebra boundary extraction capability. The shallow layer of the decoder uses a hierarchical multi-scale feature fusion mechanism module to cleverly balance local and global features and channel interaction, overcoming the typical high-resolution image processing challenges related to self-attention; 6. The method of vertebral endplate detection anomaly repair and vertebral morphology assessment of claim 5, wherein, The deeper decoder layer uses a DAE-Former Block to effectively retain long-distance dependencies in low-resolution images, while combining spatial and channel attention mechanisms without destroying the two-dimensional structure of the image. The MASAG module has two inputs, X and G; 7. The method of vertebral endplate detection anomaly repair and vertebral morphology assessment of claim 6, wherein, X represents the output of the MaxViT Block module at the same layer, and G represents the output of the Patch Expanding edge filling, the two inputs are respectively extracted, and the strategy of respectively extracting features, then fusing, then respectively extracting features, and then fusing and outputting is adopted. The vertebra shape evaluation according to the shape of the spinal vertebra comprises the following steps: The category to which the vertebra belongs is automatically evaluated through the detected shape of each vertebra; 8. A system for detecting and repairing abnormalities in vertebral body margins and assessing vertebral body morphology, comprising: The category includes: lumbar disc lesions, lumbar bone lesions, lumbar inflammatory lesions and lumbar traumatic lesions. The application comprises the following steps: The data acquisition module is used for collecting spinal CT data and marking the edges of each vertebra; The data preprocessing module is used for data preprocessing of the spinal CT data; The vertebra edge detection module is used for detecting the edges of each vertebra for the first time based on the data preprocessed spinal CT data by using a vertebra edge detection network; The abnormal edge repair module is used for repairing the abnormal edges of the vertebra for the second time by using a vertebra edge repair algorithm; 9. An electronic device, comprising: The vertebra shape evaluation module is used for evaluating the shape of the vertebra according to the shape of the spinal vertebra. The application comprises the following steps: A memory, a processor, and a computer program stored on the memory and executable on the processor, which, when executed by the processor, implement the steps of the method for vertebral edge detection anomaly repair and vertebral morphology assessment according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer readable storage medium, which, when executed by a processor, implements the steps of the method for vertebral edge detection anomaly repair and vertebral morphology assessment according to any one of claims 1 to 7.

Images