A neurosurgical imaging diagnostic method and system based on image processing

Abstract

This invention relates to the field of medical image processing technology, specifically to a neurosurgical image diagnosis method and system based on image processing. The method includes the following steps: acquiring gray matter edge nodes in a triaxial section and constructing symmetrical path units; collecting edge direction vectors and generating direction trajectory maps; extracting continuous slices to construct rotation trajectory sequences to identify abnormal regions; filling gaps and merging path voxels; and dividing spatial levels to generate an image structure map. In this invention, by acquiring gray matter edge nodes in a triaxial section and selecting node paths with symmetrical features based on spatial projection trends, accurate identification of continuous structural regions can be achieved. Extracting direction vectors from continuous slices and identifying directional abrupt change regions in the trajectory allows for timely capture of structural jumps and breaks, precise annotation of abnormal regions, and the realization of higher-dimensional expression and finer-grained identification of neural structures, effectively supporting spatial modeling and visual analysis of complex neuropathies.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a neurosurgical image diagnosis method and system based on image processing. Background Technology

[0002] The field of neurosurgical medical image processing technology encompasses multiple stages, including medical image acquisition, image preprocessing, image reconstruction, and image analysis. Its core lies in the precise analysis of imaging data of the brain and nervous system structures to aid diagnosis. This technology covers image acquisition techniques such as magnetic resonance imaging (MRI), computed tomography (CT), and functional imaging. It enhances image quality and structural clarity through image enhancement, image segmentation, and feature extraction, and uses methods such as quantitative analysis and stereoscopic visualization to locate and determine lesions in the nervous system. This overall technological field is based on neuroanatomical information and combines multimodal imaging data processing and analysis to form a complete diagnostic support system, widely applied in the preoperative and postoperative evaluation of various neurological diseases such as cerebral hemorrhage, brain tumors, and cerebral infarction.

[0003] Among them, the neurosurgical imaging diagnostic method of image processing refers to the method of extracting and analyzing information on specific lesions or regions of the brain by means of multimodal registration, image segmentation and anatomical structure reconstruction for the diagnostic images required for neurosurgical diseases. The technical matters involved in this patent subject include spatial registration of magnetic resonance imaging data and functional magnetic resonance imaging data to achieve alignment of anatomical information and functional information, segmentation of gray matter, white matter and cerebrospinal fluid regions in brain images to clarify structural boundaries, reproduction of the spatial distribution of specific neural structures through three-dimensional reconstruction, and identification and labeling of target regions by combining statistical template-based localization technology. All of these rely on the fine analysis of medical image data and prior knowledge of neural structures, aiming to complete the process of identifying and labeling the neurological lesion region on the image.

[0004] Current technologies for identifying neural structures rely on spatial registration of multimodal images and segmentation of gray matter, white matter, and cerebrospinal fluid. However, they fail to fully utilize the spatial symmetry and directional coherence between structures, relying solely on anatomical templates and explicit segmentation for structural identification, ignoring the directional evolution of neural tissue within slice sequences. This static structural assessment suffers from insensitivity to identifying structural breakpoints in practice. Consequently, when faced with brain tissue damage, deformation, or blurred lesion margins, the ability to reconstruct structures is limited, resulting in reduced localization accuracy and impacting the reference value for preoperative assessment and intraoperative navigation. In scenarios where brain tumors compress and deform local fiber bundles, traditional template-matching-based identification methods are prone to misidentifying normal structures as lesion areas or missing abnormal path information, limiting diagnostic sensitivity and robustness. Existing methods lack directional guidance for restoring the coherence of three-dimensional structures, making it difficult to reconstruct the actual structural orientation of breakpoint areas, further restricting in-depth analysis of the complex relationships between lesions and surrounding tissues. Summary of the Invention

[0005] To address the shortcomings of existing technologies that fail to fully utilize the spatial symmetry and directional continuity between structures, relying solely on anatomical templates and explicit segmentation for structural identification while ignoring the directional evolution of neural tissue within slice sequences, this static structural assessment suffers from insensitivity to identifying structural fracture zones. This results in limited structural reconstruction capabilities and reduced localization accuracy when faced with brain tissue damage, deformation, or blurred lesion edges, impacting the reference value for preoperative assessment and intraoperative navigation. In scenarios where brain tumors compress and deform local fiber bundles, traditional template-matching-based identification methods are prone to misidentifying normal structures as lesion areas or missing abnormal path information, limiting diagnostic sensitivity and robustness. Existing methods lack directional guidance for restoring the continuity of three-dimensional structures, making it difficult to reconstruct the actual structural orientation of fractured areas, further hindering the in-depth analysis of the complex relationship between lesions and surrounding tissues. This invention provides a neurosurgical imaging diagnostic method and system based on image processing. The technical solution is as follows:

[0006] On the one hand, an image processing-based neurosurgical imaging diagnostic method is provided, which includes:

[0007] S1: Obtain gray matter edge nodes in the coronal, sagittal, and horizontal three-axis image sections, construct node pair sets according to the axis, compare the spatial projection trends of multiple node pairs in the three-axis directions, and if the directional extensions are mutually symmetrical, they are identified as candidate regions for structural continuous paths, and a sequence of symmetrical path units is generated.

[0008] S2: Call the gray matter edge points of the calibrated region in the symmetric path unit sequence, collect the neighborhood direction vectors, and compare them with the standard anatomical direction group item by item. If the directions tend to be consistent, the points are retained. Identify edge path segments with consistent directions and generate a neural boundary direction trajectory map.

[0009] S3: Based on the brain lobe edge region within the neural boundary direction trajectory map, extract the nerve fiber direction vectors from the continuous slices, construct a rotation trajectory sequence according to the slice order, analyze the direction change trend between adjacent vectors, identify trajectory segments with abrupt direction changes, and generate the spatial distribution of structural fractures.

[0010] S4: Call the directional path segments that have not broken in the spatial distribution of the structural fracture, combine them with the neighborhood direction vector to construct three-dimensional path voxels, fill the gap segments by means of staggered path directions, and generate a set of continuous structural fillers.

[0011] As a further embodiment of the present invention, the symmetrical path unit sequence includes structural extension direction identifiers, spatial projection correspondences, and node pair set numbers; the neural boundary direction trajectory map includes edge direction annotation sequences, distribution of direction consistency segments, and a table of eliminated point indexes; the spatial distribution of structural fractures includes jump trajectory identifier areas, tension abnormality block numbers, and fracture trend extension markers; and the continuous structural filler set includes path projection connection groups, staggered filler path segments, and voxel-level connected units.

[0012] As a further aspect of the present invention, the step of obtaining the symmetric path unit sequence specifically includes:

[0013] S101: Obtain gray matter edge nodes in the coronal, sagittal and horizontal three-axis image sections, identify pixels whose signal intensity change rate at the edge of the section exceeds the gray matter boundary change value, connect multiple adjacent pixels to construct a continuous boundary path, extract it as an edge node sequence, and generate a three-axis gray matter edge node set;

[0014] S102: Call the three-axis gray matter edge node set, pair the nodes in pairs according to the positional relationship of the multi-axis coordinate system, and remove combinations whose horizontal or vertical coordinate spacing exceeds the average node spacing multiple, retain the pairing combinations that meet the node distance requirements, and obtain the displacement distribution of the paired node pairs.

[0015] S103: Based on the spatial coordinates of the node pair in the three-axis direction in the displacement distribution of the paired node pair, calculate the projection change direction and path expansion degree of the node pair on the three axes, and compare the angle difference between the projection vectors in the differentiated direction with the angle spacing of the three-axis symmetry direction to screen out the path region with symmetrical extension direction and establish a symmetric path unit sequence.

[0016] As a further aspect of the present invention, the step of obtaining the neural boundary direction trajectory map specifically includes:

[0017] S201: Based on the edge nodes of the calibrated region in the symmetric path unit sequence, call the spatial position of the edge nodes in the grayscale image, extract the grayscale edge gradient change direction in the pixel neighborhood, and generate the edge node neighborhood direction vector group.

[0018] S202: Call the neighborhood direction vector group of the edge node, calculate the direction angle difference between the direction vector and the three main directions of the standard anatomical direction group, and sort the angle difference according to the main direction to obtain the direction-consistent edge node set;

[0019] S203: Based on the spatial arrangement of the nodes in the directional consistent edge node cluster, connect the continuous nodes and remove the path segments whose lateral or longitudinal distance changes exceed the average edge segment length, combine them into directionally coherent edge path segments, and obtain the displacement map of continuous edge directional path segments.

[0020] S204: Call the continuous edge direction path segment displacement map, locate and map the direction distribution vector of the path segment according to the spatial coordinates of the anatomical structure, and convert the mapping result into a direction vector trajectory to establish a nerve boundary direction trajectory map.

[0021] As a further aspect of the present invention, the step of obtaining the spatial distribution of structural fractures specifically includes:

[0022] S301: Based on the brain lobe edge region marked in the neural boundary direction trajectory map, extract the direction vectors of nerve fibers located in the region in the continuous slices, and arrange the direction vectors in sequence according to the slice number to obtain the nerve fiber rotation trajectory sequence.

[0023] S302: Call the nerve fiber rotation trajectory sequence, calculate the angle change value between adjacent direction vectors, and compare the average value and the maximum value of the direction change in three consecutive slices to filter the trajectory segments whose direction change amplitude exceeds twice the average change range, and obtain the distribution of direction change trajectory segments.

[0024] S303: Based on the number of consecutive jumps between adjacent abrupt change segments in the distribution of the directional abrupt change trajectory segments, the jump segments are compared segment by segment in spatial position to identify the path regions in the trajectory where consecutive jumps occur, and the regions are located and marked in three-dimensional space according to image slice coordinates to obtain the spatial distribution of structural fractures.

[0025] The spatial distribution value of the continuous jump path region is calculated using the following formula:

[0026]

[0027] Where SA represents the spatial distribution value of the continuous jump path region, P i P represents the spatial coordinates of the i-th jump path. i+1 |P represents the spatial coordinates of the (i+1)th jump path. i+1 -P i | represents the absolute distance between the i-th and i+1-th jump paths, A i B represents the jump amplitude of the i-th jump path. i The jump stability coefficient represents the jump path of the i-th segment. C is the ratio of the jump amplitude to the stability coefficient. i represents the weighting factor of the i-th jump path, and N represents the total number of jump paths.

[0028] As a further aspect of the present invention, the step of obtaining the continuous structure infill set specifically includes:

[0029] S401: Call the directional path segments that have not broken in the spatial distribution of the structural fracture, collect the directional vector values ​​of adjacent voxels within the boundary range, determine the extension direction of the path segments, connect the three-dimensional spatial paths according to the voxel points with the same direction, and generate a three-dimensional directional path voxel set.

[0030] S402: Based on the gaps in space of the path segments in the three-dimensional directional path voxel set, extract the directional information of the boundary voxels of the segments, and fill them by staggering the vertical, front-back, and left-right directions. Calculate the connectivity probability of the path units, and merge the local path units according to the splicing results to generate a continuous structure filling set.

[0031] As a further aspect of the present invention, the formula for calculating the connectivity probability of the path unit is:

[0032]

[0033] Among them, P ab V represents the connectivity probability between voxels a and b. a V represents the volume of voxel a in space. b d represents the volume of voxel b in space. a Represents the directional information of voxel a, d b This represents the directional information of voxel b.

[0034] As a further aspect of the present invention, the method further includes step S5:

[0035] S5: Based on the path direction and connection relationship in the continuous structure filler set, divide the spatial hierarchy of the structural blocks, identify the structural connection hierarchy and spatial interlacing order, count the connection order and interlacing position, and generate a neural image spatial hierarchy structure map.

[0036] The neuroimaging spatial hierarchy diagram includes a spatial hierarchy index matrix, connection structure affiliation clusters, and staggered orientation mapping relationships.

[0037] As a further aspect of the present invention, the step of obtaining the neural image spatial hierarchy structure map specifically includes:

[0038] S501: Based on the extension direction and connection relationship of the path segments in the continuous structure infill set, extract the direction vector of the path unit in three-dimensional coordinates, and determine the belonging relationship based on the position between the connection points, divide the structure block set with primary and secondary connection relationship, and generate the structure block spatial connection level value.

[0039] S502: Call the spatial connection hierarchy value of the structural block, identify the stacking order and staggered position between structural blocks according to the spatial arrangement order between the starting path unit and the subordinate path unit in the connection relationship, and calculate the degree of staggered position according to the overlap rate of the structural blocks in the coordinate axis direction to obtain the distribution value of the structural connection staggered order.

[0040] S503: Based on the spatial position and hierarchical division results of the structural blocks in the staggered distribution value of the structural connection, the structural blocks are sequentially located in the spatial three-axis coordinate system according to the connection order, and the differentiated hierarchical markers are assigned to the corresponding position areas of the structural blocks to obtain the spatial hierarchical structure map of the neural image.

[0041] On the other hand, the image processing-based neurosurgical imaging diagnostic system is used to perform the above-described image processing-based neurosurgical imaging diagnostic method, and the system includes:

[0042] The section node extraction module obtains gray matter edge nodes in coronal, sagittal, and horizontal triaxial image sections, detects the brightness change gradient of gray matter region edges within the slice, collects the location of gray value changes, identifies whether edge feature trends are present, and generates a symmetrical path unit sequence.

[0043] The path symmetry construction module, based on the symmetric path unit sequence, pairs nodes in the three-axis direction according to their spatial location, calculates the difference value of the directional projection trend of the node pair in the differential axis, filters nodes that meet the directional symmetry relationship, and generates a neural boundary directional trajectory map.

[0044] The edge direction recognition module calls the neural boundary direction trajectory map, collects the edge direction vectors within the neighborhood, and compares them with the direction vectors of typical brain regions in the anatomical reference direction group in turn to determine whether the direction consistency meets the continuity standard and generates the spatial distribution of structural breaks.

[0045] The continuous structural filling module determines whether there is a significant change in direction between adjacent vectors based on the spatial distribution of the structural fracture and marks the fracture area in the trajectory accordingly. After excluding the fracture segment, it calls the directional path segment that has not undergone a change in direction, detects the angle difference between the path endpoint direction and the path direction of the surrounding area, and generates a set of continuous structural filling bodies.

[0046] The spatial structure partitioning module calls the direction information of each path in the continuous structure filling set, determines the directional relationship between the main axis direction and the three-dimensional coordinate axis, sets the hierarchical label according to the typical orientation of the brain region structure, detects the degree of spatial position overlap and connection order between paths, and generates a neural image spatial hierarchy structure map.

[0047] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0048] By acquiring gray-matter edge nodes in triaxial sections and selecting node paths with symmetrical features based on spatial projection trends, accurate identification of continuous structural regions can be achieved. A sequence of directionally continuous path units can be constructed, and edge points are screened for directional consistency using standard anatomical directions, generating trajectory diagrams with neural pathway logic, effectively improving the clarity of directional representation of edge structures. Extracting direction vectors from continuous slices and identifying directional abrupt change regions in the trajectory allows for timely capture of structural jumps and breaks, enabling precise labeling of anomalous areas. Based on paths without breaks, gaps are filled using neighborhood vectors, and continuous path voxels are integrated to construct a spatially coherent set of infill volumes, making the structural reconstruction of broken areas more complete and accurate. By hierarchically dividing and spatially intersecting based on path direction and connection relationships, a clear structural hierarchy map is established. Combined with the directional symmetry of node pairs, consistency judgment of edge paths, analysis of the changing trend of directional trajectories, and three-dimensional reconstruction of fracture space, the spatial continuity, anatomical consistency, and accuracy of lesion localization of neural imaging structures can be improved without relying on explicit segmentation. This enables higher-dimensional expression and finer-grained recognition of neural structures, effectively supporting spatial modeling and visual analysis of complex neuropathies. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0050] Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0051] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0052] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0053] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0054] Please see Figure 1 This invention provides a neurosurgical imaging diagnostic method based on image processing. The processing flow of this method may include the following steps:

[0055] S1: Obtain gray matter edge nodes in the coronal, sagittal, and horizontal three-axis image sections, construct node pair sets according to the axis, compare the spatial projection trends of multiple node pairs in the three-axis directions, and if the directional extensions are mutually symmetrical, they are identified as candidate regions for structural continuous paths, and a sequence of symmetrical path units is generated.

[0056] S2: Call the gray matter edge points of the calibrated region in the symmetric path unit sequence, collect the neighborhood direction vectors, and compare them with the standard anatomical direction group item by item. If the directions tend to be consistent, the points are retained. Identify edge path segments with consistent directions and generate a neural boundary direction trajectory map.

[0057] S3: Based on the cerebral lobe edge region within the neural boundary direction trajectory map, extract the direction vectors of nerve fibers in continuous slices, construct a rotation trajectory sequence according to the slice order, analyze the direction change trend between adjacent vectors, identify trajectory segments with abrupt direction changes, and if the trajectory jumps continuously, the corresponding area is marked as a structural abnormality area, generating a spatial distribution of structural fractures.

[0058] S4: Call the directional path segments that have not broken in the spatial distribution of structural fractures, combine them with the neighborhood direction vectors to construct three-dimensional path voxels, fill the gap segments by the path direction intersection, and merge the path units according to the connectivity trend to generate a continuous structural filling set.

[0059] S5: Based on the path direction and connection relationship in the continuous structure infill set, divide the spatial hierarchy of the structural blocks, identify the structural connection hierarchy and spatial interlacing order, count the connection order and interlacing position, and generate a neural image spatial hierarchy structure map.

[0060] The symmetrical path unit sequence includes structural extension direction identifiers, spatial projection correspondences, and node pair set numbers; the neural boundary direction trajectory map includes edge direction annotation sequences, distribution of direction consistency segments, and a culling point index table; the spatial distribution of structural fractures includes jump trajectory identifier areas, tension abnormality block numbers, and fracture trend extension markers; the continuous structural filler set includes path projection connection groups, staggered filler path segments, and voxel-level connected units; and the neural image spatial hierarchy structure map includes a spatial hierarchy index matrix, connected structure affiliation clusters, and staggered direction mapping relationships.

[0061] The specific steps for obtaining the symmetric path unit sequence are as follows:

[0062] S101: Obtain gray matter edge nodes in the coronal, sagittal and horizontal three-axis image sections, identify pixels whose signal intensity change rate at the edge of the section exceeds the gray matter boundary change value, connect multiple adjacent pixels to construct a continuous boundary path, extract it as an edge node sequence, and generate a three-axis gray matter edge node set;

[0063] The triaxial MRI images are imported into image processing software, such as 3DSlicer or ITK-SNAP. The required slice images are extracted by axis using the batch slice browsing function. Coronal, sagittal, and horizontal slices are labeled and numbered to ensure the independence and integrity of subsequent processing. Then, an image enhancement module is applied to adjust the image contrast, making the boundary between gray matter and remaining brain tissue clearer. A trained image recognition model is then invoked to perform edge recognition on the gray matter region. The model can be a U-Net-based segmentation network that generates a clear boundary mask at the interface between gray matter and white matter or cerebrospinal fluid. The image intensity value of each pixel on the edge mask is read, and the intensity changes of neighboring pixels are traversed. By comparing the intensity differences between neighbors, it is determined whether there is a significant boundary change. Pixels that meet the rate of change threshold are filtered out. Multiple adjacent pixels that meet the criteria are classified as the same continuous boundary path. A contour tracking algorithm is used to connect these points one by one to form closed or semi-open paths, which are then numbered. All valid path nodes in the slices are organized, archived, and labeled with the direction of the slice, forming a triaxial gray matter edge node set.

[0064] S102: Call the three-axis gray mass edge node set, pair the nodes in pairs according to the positional relationship of the multi-axis coordinate system, and remove combinations whose horizontal or vertical coordinate spacing exceeds the multiple of the average node spacing, retain the pairing combinations that meet the node distance requirements, and obtain the displacement distribution of the paired node pairs.

[0065] The position coordinates of each axial node are standardized to ensure that the node coordinates obtained from different sections can be compared uniformly in three-dimensional space. The pairing process is traversed in pairs, and the combination method is limited to different axes, such as setting coronal and sagittal planes, sagittal and horizontal planes, to avoid repeated pairings from the same section. The lateral and longitudinal coordinate distances of each combination are extracted and compared with the average spacing calculated for the entire node set. A multiplier threshold is set as the basis for rejection to exclude long-distance node pairings and avoid interference from unstructured connections to the analysis. When analyzing the insula region, if the node distribution is dense, a smaller multiplier coefficient can be set to exclude long-distance mismatches caused by brain split. Node combinations that meet the spacing requirements will be included in the effective pairing result set, and the three-dimensional offset information of each pairing combination is recorded. The entire pairing process can be completed in batches by script, which greatly improves the pairing efficiency and serves as the basis for the next stage of directional judgment and symmetry screening to obtain the displacement distribution of paired node pairs.

[0066] S103: Based on the spatial coordinates of the node pairs in the three-axis direction in the displacement distribution of the paired node pairs, calculate the direction of projection change of the node pairs in the three axes and the degree of path expansion, and compare the angle difference between the projection vectors in the differentiated direction with the angle spacing in the three-axis symmetric direction to screen out the path regions with symmetrical direction extension and establish a symmetric path unit sequence.

[0067] The coordinate changes of each pair of paired nodes in three directions in space need to be decomposed to determine the expansion direction between paths. In practice, the coordinate difference of each pair of nodes will be extracted along the X, Y, and Z axes to form projected paths in three directions. Subsequently, it is necessary to compare whether the expansion directions of adjacent paired node pairs are consistent. The determination of the path expansion degree depends on the series relationship of multiple paired nodes. By analyzing the offset direction of continuous paths on each axis, the directional change sequence is extracted, and the change amplitude of the path direction is calculated segment by segment. Path sequences with small directional differences will be aggregated into structurally symmetrical path regions. The angle difference between the aggregated path direction and the standard three-axis symmetrical direction is compared. If the difference is within the set threshold range, it is determined to be a symmetrical extended path region. This process can be executed in three-dimensional coordinate space using Python scripts, and the path segments that meet the requirements can be marked in the graphical tool to establish a complete symmetrical path unit sequence. During the entire process, it is necessary to pay close attention to whether there are abrupt changes in node direction. If the path turns back or deviates, it is necessary to further verify its spatial continuity and structural rationality to ensure that the selected path segments have similar structural features and spatial distribution to establish a symmetrical path unit sequence.

[0068] The specific steps for obtaining the neural boundary direction trajectory map are as follows:

[0069] S201: Based on the edge nodes of the calibrated region in the symmetric path unit sequence, call the spatial position of the edge nodes in the gray image, extract the gray edge gradient change direction in the pixel neighborhood, and generate the edge node neighborhood direction vector group.

[0070] The node spatial coordinates are restored to the original grayscale image, and the pixel neighborhood data is extracted using these coordinates as the center. The neighborhood range is defined as a 3×3 or 5×5 pixel area to cover sufficient boundary information. Each neighborhood pixel set is scanned, and the intensity value changes between the central node and surrounding pixels are judged. Grayscale values ​​are read and a difference distribution map is constructed according to the adjacent directions. The main gradient change direction of each edge node is extracted through the directional change trend. During the gradient direction extraction process, the specific angle is not directly calculated. Instead, unit direction vectors are constructed by judging the distribution of the change trend in the horizontal, vertical, or diagonal directions. These vectors are saved in the form of groups. The direction vector group of each edge node contains multiple main directional changes observed from the neighborhood, which has the ability to reflect the local morphology of the edge. At the edge turning point, the direction vector will show a clear bending trend, while in the straight line segment, the direction vector tends to be consistent. In this way, a large number of neighborhood direction vector groups with nodes as the core can be formed, providing basic information for subsequent direction consistency calculation, and generating the edge node neighborhood direction vector group.

[0071] S202: Call the neighborhood direction vector group of the edge node, calculate the direction angle difference between the direction vector and the three main directions of the standard anatomical direction group, and sort the angle difference according to the main direction to obtain the direction-consistent edge node set;

[0072] The main direction vector group in the standard anatomical direction reference model is invoked, including three main directions extending along the sagittal, coronal, and horizontal planes. These are compared with the local direction vectors of the edge nodes. During execution, the neighborhood direction vectors of each edge node are traversed, and the angular direction difference is calculated with the anatomical direction vector one by one. The degree of deviation is judged by comparing the directional relationship. In the operation, the specific angle value is not used as the sole criterion. Instead, the results of the angular difference in the three directions are sorted according to the principle of directional consistency to determine the main direction closest to each node and classify it accordingly. After sorting, a direction priority sequence can be established to determine whether the node has a main directional tendency. The node with the smallest directional difference is assigned to the directional consistency edge node set. The nodes in this set tend to be uniformly distributed in the anatomical direction and can be used to represent the directional characteristics of a certain structure. When analyzing the edge of the frontal cortex, if multiple nodes tend to the coronal main direction, they can be judged to belong to the same anatomical edge structure. This process is verified by graphical annotation. The distribution characteristics of nodes under different directional consistency are displayed through visualization, which improves the accuracy of node selection and the clarity of classification, and obtains the directional consistency edge node set.

[0073] S203: Based on the spatial arrangement of the nodes in the directional consistent edge node cluster, connect the continuous nodes and remove the path segments whose lateral or longitudinal distance changes exceed the average edge segment length, combine them into directionally coherent edge path segments, and obtain the displacement map of continuous edge directional path segments.

[0074] The nodes need to be located in three-dimensional space and their continuity analyzed according to the node arrangement order. The nodes are ordered according to their spatial coordinates and connected based on the spatial distance between adjacent nodes. The connection operation follows two principles: consistent edge segment direction and controlled distance change. For each pair of consecutive nodes, it is determined whether the change values ​​of their horizontal and vertical coordinates exceed the preset average edge segment length. If the change values ​​deviate significantly from the overall average, they are judged as discontinuous paths and are removed. In neural images, the edge segment length can be estimated by the average value of all node connection paths, which serves as a structural hierarchy identifier. The remaining node sequences that meet the conditions will form new edge path segments according to the principle of directional continuity. The path segments are recorded and classified into a continuous edge direction path segment displacement map. This map reflects the continuous distribution in the local direction, which is convenient for subsequent structural mapping and spatial trajectory calculation. This process can be combined with image display tools to distinguish and display the retained path segments with different colors or line styles, and attach displacement data for each segment, laying a stable data foundation for subsequent spatial projection and trajectory generation, and obtaining a continuous edge direction path segment displacement map.

[0075] S204: Call the continuous edge direction path segment displacement map, locate and map the direction distribution vector of the path segment according to the spatial coordinates of the anatomical structure, and convert the mapping result into a direction vector trajectory to establish a nerve boundary direction trajectory map;

[0076] Spatial mapping of the path segment direction vectors is required, that is, mapping the direction of each segment in the path to the coordinate system of the overall anatomical structure. The mapping process requires setting the reference point of the anatomical coordinate system and the standard of the three directional axes, which is usually corrected based on the MNI or Talairach spatial model. The three-dimensional position of each path segment is standardized to this coordinate system. Then, the principal direction vector of each path segment is extracted in sequence, and according to its relative position in space, it is mapped to a vector point in the coordinate system. All vectors are combined to form a set of three-dimensional spatial direction trajectories. By connecting adjacent vector points, a set of continuous direction vector trajectory line segments is constructed, and finally, a neural boundary direction trajectory map is formed. During the operation, three-dimensional rendering tools can be used to visualize the trajectory line segments and mark different types of orientation patterns according to the direction change rules, such as radial, fan-shaped or concentrated distribution, so that the overall trajectory has a clear geometric structure description. This map, as a visual representation of the path direction, has the ability to describe the continuity and spatial arrangement information of the neural edge. It can be used as the basis for further analysis, such as edge structure stability, edge morphology classification, etc., to establish the neural boundary direction trajectory map.

[0077] The specific steps for obtaining the spatial distribution of structural fractures are as follows:

[0078] S301: Based on the brain lobe edge region marked in the neural boundary direction trajectory map, extract the direction vectors of nerve fibers located in the region in the continuous slices, arrange the direction vectors in sequence according to the slice number, and obtain the nerve fiber rotation trajectory sequence.

[0079] In the image processing platform, a target analysis region is set and all corresponding image slices are loaded. Each image is read according to the slice number order. The direction vector information of nerve fibers falling within the edge region of each slice is extracted. The acquisition of direction vectors depends on the pre-calculated direction path map. Each pixel or path segment in the direction path map has its dominant extension direction marked. The reading operation requires setting the region mask condition in the program script to extract only the path vectors within the mask. After extracting each slice in sequence, the vectors obtained in each image are sequentially appended to form a new trajectory sequence, which constitutes the dynamic evolution path of nerve fiber direction. This trajectory sequence has a stable structure and can record the natural rotation trend of direction as the slice layer progresses, which is used to reveal the directional continuity of local structures. Throughout the process, the data format must be consistent. The direction vectors must be standardized to unit length and accompanied by the corresponding slice number to facilitate the processing and retrieval of their angle change amplitude in the subsequent analysis stage, so as to obtain the nerve fiber rotation trajectory sequence.

[0080] S302: Call the nerve fiber rotation trajectory sequence, calculate the angle change value between adjacent direction vectors, and compare the average value with the maximum value of the direction change in three consecutive slices. Filter out trajectory segments whose direction change amplitude exceeds twice the average change range and obtain the distribution of direction change trajectory segments.

[0081] The program needs to analyze the directional changes of continuous vectors in adjacent slices. It reads the vector groups in the trajectory sequence one by one and calculates the angle difference between the directional changes of each two adjacent groups. The change value is recorded and the corresponding slice number position is marked. To improve the recognition accuracy, a three-slice sliding window is set up to calculate the average value and the maximum value of the two angle changes in three consecutive slices. The maximum value and the average value are multiplied by a preset coefficient and compared. This coefficient is generally set to 2 times. That is, a change exceeding twice the average change amplitude is considered an abnormal change. All change points that meet this standard are marked as directional change trajectory segments, and the start and end numbers of the trajectory segments in the sequence are stored. In actual operation, if the analysis is of the occipital lobe cortex edge region, attention should be paid to whether the change trajectory is concentrated in a certain depth range or whether it frequently alternates and jumps at multiple directional angles. This serves as a preliminary basis for identifying local fracture structures or abnormal activities, providing a data entry point for subsequent jump continuity analysis and spatial marking work, and obtaining the distribution of directional change trajectory segments.

[0082] S303: Based on the number of consecutive jumps between adjacent abrupt change segments in the distribution of abrupt change trajectory segments, the jump segments are compared segment by segment in spatial location to identify the path regions in the trajectory where consecutive jumps occur. The regions are then located and marked in three-dimensional space according to image slice coordinates to obtain the spatial distribution of structural fractures.

[0083] The spatial distribution values ​​of the continuous jump path region are calculated using the following formula:

[0084]

[0085] Where SA represents the spatial distribution value of the continuous jump path region, P i P represents the spatial coordinates of the i-th jump path. i+1 |P represents the spatial coordinates of the (i+1)th jump path. i+1 -P i | represents the absolute distance between the i-th and i+1-th jump paths, A i B represents the jump amplitude of the i-th jump path. i The jump stability coefficient represents the jump path of the i-th segment. C is the ratio of the jump amplitude to the stability coefficient. i The weighting factor represents the i-th jump path, and N represents the total number of jump paths;

[0086] Meaning of parameters and derivation of formulas:

[0087] P i and P i+1 Let represent the spatial coordinates of the jump paths for the i-th and (i+1)-th segments, respectively. These coordinates are obtained through a three-dimensional spatial positioning system (such as laser scanning, stereo vision, or GPS). Let the starting coordinates of the first segment be P1 = (x1, y1, z1), and the starting coordinates of the second segment be P2 = (x2, y2, z2). Then the Euclidean distance between the two points is:

[0088]

[0089] A i : Represents the jump amplitude of the i-th jump path, calculated by analyzing the curvature change or velocity change rate of the path. If the velocity change rate of the first jump path is Δv1, then the jump amplitude is:

[0090] A1 = Δv1;

[0091] B i : Represents the jump stability coefficient of the i-th jump path, evaluated by the smoothness, continuity, or amplitude decay of the path. If the smoothness index of the first path segment is S1, then the stability coefficient is .

[0092] B1 = S1;

[0093] C i : Represents the weighting factor of the i-th jump path, assigned a value based on the path's importance, frequency, or other relevant factors. If the importance weight of the first path segment is W1, then the weighting factor is:

[0094] C1 = W1;

[0095] Formula calculation derivation:

[0096] Calculate the distance difference for each jump path:

[0097] For the first and second jump paths, calculate the Euclidean distance between their spatial coordinates:

[0098]

[0099] Calculate the ratio of jump amplitude to stability coefficient:

[0100] For the first jump path, calculate the ratio of jump amplitude to stability coefficient:

[0101]

[0102] Calculate the weighted difference for each jump path:

[0103] For the first jump path, calculate its weighted difference:

[0104]

[0105] The weighted difference of cumulative jump paths:

[0106] For N jump paths, calculate their sum:

[0107] Example calculation:

[0108] There are 3 jump paths set up, with the following parameters:

[0109] Segment 1: P1 = (0, 0, 0), P2 = (1, 1, 1), A1 = 2, B1 = 1, C1 = 1; Segment 2: P2 = (1, 1, 1), P3 = (2, 2, 2), A2 = 3, B2 = 2, C2 = 1.5; Segment 3: P3 = (2, 2, 2), P4 = (3, 3, 3), A3 = 4, B3 = 3, C3 = 2; Calculate the Euclidean distance of each jump path:

[0110] Paragraph 1:

[0111]

[0112] Paragraph 2:

[0113]

[0114] Paragraph 3:

[0115]

[0116] Calculate the ratio of jump amplitude to stability coefficient for each jump path segment:

[0117] Paragraph 1:

[0118]

[0119] Paragraph 2:

[0120]

[0121] Paragraph 3:

[0122]

[0123] Calculate the weighted difference for each jump path:

[0124] Paragraph 1:

[0125]

[0126] Paragraph 2:

[0127]

[0128] Paragraph 3:

[0129]

[0130] The weighted difference of cumulative jump paths:

[0131] SA=-0.268+0.348+0.798=0.878;

[0132] The results indicate that the spatial distribution value of the continuous jump path region in the analyzed trajectory is 0.878, reflecting the continuity and stability of the path.

[0133] The specific steps for obtaining the set of continuous structure fillers are as follows:

[0134] S401: Call the directional path segments that have not broken in the spatial distribution of structural fractures, collect the directional vector values ​​of adjacent voxels within the boundary range, determine the extension direction of the path segments, connect the three-dimensional spatial paths according to the voxel points with the same direction, and generate a three-dimensional directional path voxel set.

[0135] Path segments with stable structural orientations are prioritized and used as the starting point for subsequent 3D extension analysis. During the operation, the path segments are located in the 3D image data, and their boundary voxel information is extracted by accessing the corresponding slice number and coordinate region. Within the boundary range of each path segment, the program traverses multiple adjacent voxel points to obtain the orientation vector value carried by each voxel and analyzes its matching degree with the main direction of the path. To determine the extension direction of the path segment, an orientation consistency threshold needs to be set, and only voxel points with deviations from the main direction of the path within a specified range are retained for subsequent connection. When processing the frontal cortex orientation segment, only neighboring voxel points with a difference of no more than a specified angle from the original direction are retained. 3D path connections are performed on voxel points with consistent orientations. The connection process is based on the relative position of voxels in the X, Y, and Z axes space, and is executed according to the rule of prioritizing connections with consistent orientations and the shortest spatial distance. Adjacent voxels are gradually expanded to form a complete 3D orientation path. During the path generation process, the index relationship and voxel number of each connection chain are recorded to generate a 3D orientation path voxel set.

[0136] S402: Based on the gaps in space of the path segments in the three-dimensional directional path voxel set, extract the directional information of the boundary voxels of the segments, and fill them in the up-down, front-back and left-right directions by staggering. Calculate the connectivity probability of the path units, merge the local path units according to the splicing results, and generate a continuous structure filling set.

[0137] The formula for calculating the connectivity probability of a path unit is:

[0138]

[0139] Among them, P ab V represents the connectivity probability between voxels a and b. a V represents the volume of voxel a in space. b d represents the volume of voxel b in space. a Represents the directional information of voxel a, d b Represents the directional information of voxel b;

[0140] Meaning of parameters and derivation of formulas:

[0141] Volume measurement (V) a and V b ):

[0142] The volume of a voxel is obtained through imaging techniques such as magnetic resonance imaging (MRI) or computed tomography (CT). Image segmentation techniques can be used to identify the boundaries of voxels, allowing for the calculation of their volume. The unit of volume is typically cubic millimeters (mm). 3 );

[0143] Directional information extraction (d a and d b ):

[0144] The orientation information of voxels can be obtained through diffusion tensor imaging (DTI) technology. DTI can provide the principal diffusion direction of each voxel, which is represented as a unit vector. By analyzing the direction, the orientation information of the voxel can be obtained.

[0145] Once the volume and orientation information of the voxels are obtained, the connectivity probability can be calculated using the formula described above. The specific calculation process is as follows:

[0146] Calculate the absolute value of the difference between the volumes of voxels a and b:

[0147] |V a -V b |;Set V a =100mm 3 V b =120mm 3 ,but:

[0148] |V a -V b |=|100-120|=20mm 3 ;

[0149] Calculate the square root of the sum of the squares of the volumes of voxels a and b:

[0150]

[0151] Calculate the absolute value of the difference between the directional information of voxel a and voxel b:

[0152] |d a -d b |;Set d a =0.8, d b =0.9, then:

[0153] |d a -d b |=|0.8-0.9|=0.1;

[0154] Calculate the absolute value of the sum of the directional information of voxels a and b:

[0155] |d a +d b |=|0.8+0.9|=1.7;

[0156] Calculate the connectivity probability of a path unit:

[0157]

[0158] The result indicates that the connectivity probability between voxels a and b is 0.0075, suggesting a low level of connectivity between them.

[0159] The specific steps for obtaining the spatial hierarchical structure map of neuroimaging are as follows:

[0160] S501: Based on the extension direction and connection relationship of the path segments in the continuous structure infill set, extract the direction vector of the path unit in three-dimensional coordinates, and determine the belonging relationship based on the position between the connection points, divide the set of structural blocks with primary and secondary connection relationships, and generate the spatial connection level value of the structural block.

[0161] For each path segment, direction extraction and connection relationship analysis are required. The process retrieves the start point, end point, and path direction of each path segment from the 3D image data. Based on the arrangement order of multiple voxel points in the path segment, a direction vector is generated to reflect the extension trend of the path in 3D space. The connection point information between path segments is read to determine whether two path segments are directly connected or connected through intermediate voxels. In this way, a connection map between all path units is constructed. Then, based on the positional hierarchy between connection points, such as a path segment with the start point at the top and the extension direction downward, it is considered a main path, and the remaining path segments connected to its end or branch point are considered secondary paths. The path segments are divided into block-level sets with a master-slave structure. Each block set consists of a main path and subordinate paths, and is assigned a hierarchical number to indicate its positional relationship in the overall structure. Based on the affiliation of each group of blocks and the connection map structure, it is stored as a 3D image index for subsequent layer analysis, generating spatial connection hierarchy values ​​for the block.

[0162] S502: Call the spatial connection hierarchy value of the structural block, identify the stacking order and staggered position between structural blocks according to the spatial arrangement order between the starting path unit and the subordinate path unit in the connection relationship, calculate the degree of staggered position according to the overlap rate of the structural blocks in the coordinate axis direction, and obtain the distribution value of the structural connection staggered order.

[0163] The spatial interlacing and stacking relationships between different structural blocks will be further identified. The spatial path starting point and the spatial arrangement order of the subordinate path extension segments of each structural block will be read, and their starting and ending positions in the X, Y, and Z coordinate axes will be recorded. Using the path connection points as alignment references, the arrangement direction of the starting path unit of the structural block and its connected subordinate path segments in three-dimensional space will be compared one by one to confirm whether there are overlapping intervals and the degree of intersection in the extension direction. If two structural blocks have a high proportion of voxel regions overlapping in a certain coordinate axis direction, it can be identified as structural interlacing. The overlap rate is calculated by the ratio of the number of overlapping voxel numbers to the total path length, and the comparison is carried out in the three axes to identify the degree of interlacing. The interlacing relationship between structural blocks is recorded item by item and organized in the form of a list. Each item corresponds to the interlacing situation and its sequence number between a group of structural blocks. The result is used to represent the interweaving relationship of the entire group of path segments in space, laying the foundation for constructing a complete spatial hierarchical structure and generating the structural connection interlacing sequence distribution value.

[0164] S503: Based on the spatial location and hierarchical division results of the structural blocks in the staggered distribution value of structural connections, the structural blocks are sequentially located in the spatial three-axis coordinate system according to the connection order, and the differentiated hierarchical labels are assigned to the corresponding location regions of the structural blocks to obtain the spatial hierarchical structure map of the neuroimage.

[0165] Further localization of the specific distribution of neural structural blocks in three-dimensional space is required. Following the path connection order, each structural block is sequentially imported according to its number, and its actual voxel distribution area is reconstructed in the X, Y, and Z coordinate axes. During this process, the main path and subordinate paths of each structural block are sorted according to the connection order and their coordinates are marked. The overall structural block extends outward from the main path as the center, forming a tree-like or branch-like spatial distribution map. The hierarchical information of the structural blocks is also imported simultaneously. Each block is assigned a corresponding differentiated label value according to its hierarchical number in the connection map. The label value is reflected in the label change in the voxel value domain. For example, the main path is set as a first-level hierarchical value, and the subordinate path is set as a second-level value. Different color mappings or numerical identifiers are used in the annotation layer to distinguish different spatial levels. The structural block position coordinates and hierarchical numbers are integrated, and the map can be displayed in a three-dimensional visualization platform. This allows the neural boundary structure to present a clear hierarchical relationship and spatial connection direction in the three-axis coordinates, providing a complete structural localization basis for subsequent analysis and obtaining a spatial hierarchical structure map of neural images.

[0166] Please see Figure 2 A neurosurgical imaging diagnostic system based on image processing, the system comprising:

[0167] The section node extraction module obtains gray matter edge nodes in coronal, sagittal, and horizontal triaxial image sections, detects the brightness change gradient of gray matter region edges within the slice, collects the location of gray value changes, identifies whether edge feature trends are present, and generates a symmetrical path unit sequence.

[0168] The path symmetry construction module is based on the symmetric path unit sequence. It pairs nodes in the three-axis direction according to their spatial location, calculates the difference value of the directional projection trend of the node pair in the differential axis, filters nodes that meet the directional symmetry relationship, and generates a neural boundary directional trajectory map.

[0169] The edge direction recognition module calls the neural boundary direction trajectory map, collects the edge direction vectors within the neighborhood, and compares them with the direction vectors of typical brain regions in the anatomical reference direction group in turn to determine whether the direction consistency meets the continuity standard and generates the spatial distribution of structural breaks.

[0170] The continuous structural filling module determines whether there is a significant change in direction between adjacent vectors based on the spatial distribution of structural fractures and marks the fracture area in the trajectory accordingly. After excluding fracture segments, it calls the directional path segments that have not undergone a sudden change, detects the angle difference between the path endpoint direction and the path direction of the surrounding area, and generates a set of continuous structural filling bodies.

[0171] The spatial structure partitioning module calls the direction information of each path in the continuous structure filling set, determines the directional relationship between the main axis direction and the three-dimensional coordinate axis, sets the hierarchical label based on the typical orientation of the brain region structure, detects the degree of spatial position overlap and connection order between paths, and generates a neural imaging spatial hierarchy structure map.

[0172] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A neurosurgical imaging diagnostic method based on image processing, characterized in that, Includes the following steps: S1: Obtain gray matter edge nodes in the coronal, sagittal, and horizontal three-axis image sections, construct node pair sets according to the axis, compare the spatial projection trends of multiple node pairs in the three-axis directions, and if the directional extensions are mutually symmetrical, they are identified as candidate regions for structural continuous paths, and a sequence of symmetrical path units is generated. S2: Call the gray matter edge points of the calibrated region in the symmetric path unit sequence, collect the neighborhood direction vectors, and compare them with the standard anatomical direction group item by item. If the directions tend to be consistent, the points are retained. Identify edge path segments with consistent directions and generate a neural boundary direction trajectory map. S3: Based on the brain lobe edge region within the neural boundary direction trajectory map, extract the nerve fiber direction vectors from the continuous slices, construct a rotation trajectory sequence according to the slice order, analyze the direction change trend between adjacent vectors, identify trajectory segments with abrupt direction changes, and generate the spatial distribution of structural fractures. S4: Call the directional path segments that have not broken in the spatial distribution of the structural fracture, combine them with the neighborhood direction vector to construct three-dimensional path voxels, fill the gap segments by the path direction intersection, and generate a continuous structural filling body set. The symmetrical path unit sequence includes structural extension direction identifiers, spatial projection correspondences, and node pair set numbers. The neural boundary direction trajectory map includes edge direction annotation sequences, distribution of direction consistency segments, and a table of eliminated point indexes. The spatial distribution of structural fractures includes jump trajectory identifier areas, tension abnormality block numbers, and fracture trend extension markers. The continuous structural infill set includes path projection connection groups, staggered filling path segments, and voxel-level connected units.

2. The neurosurgical imaging diagnostic method based on image processing according to claim 1, characterized in that, The specific steps for obtaining the symmetric path unit sequence are as follows: S101: Obtain gray matter edge nodes in the coronal, sagittal and horizontal three-axis image sections, identify pixels whose signal intensity change rate at the edge of the section exceeds the gray matter boundary change value, connect multiple adjacent pixels to construct a continuous boundary path, extract it as an edge node sequence, and generate a three-axis gray matter edge node set; S102: Call the three-axis gray matter edge node set, pair the nodes in pairs according to the positional relationship of the multi-axis coordinate system, and remove combinations whose horizontal or vertical coordinate spacing exceeds the average node spacing multiple, retain the pairing combinations that meet the node distance requirements, and obtain the displacement distribution of the paired node pairs. S103: Based on the spatial coordinates of the node pair in the three-axis direction in the displacement distribution of the paired node pair, calculate the projection change direction and path expansion degree of the node pair on the three axes, and compare the angle difference between the projection vectors in the differentiated direction with the angle spacing of the three-axis symmetry direction to screen out the path region with symmetrical extension direction and establish a symmetric path unit sequence.

3. The neurosurgical imaging diagnostic method based on image processing according to claim 2, characterized in that, The specific steps for obtaining the neural boundary direction trajectory map are as follows: S201: Based on the edge nodes of the calibrated region in the symmetric path unit sequence, call the spatial position of the edge nodes in the grayscale image, extract the grayscale edge gradient change direction in the pixel neighborhood, and generate the edge node neighborhood direction vector group. S202: Call the neighborhood direction vector group of the edge node, calculate the direction angle difference between the direction vector and the three main directions of the standard anatomical direction group, and sort the angle difference according to the main direction to obtain the direction-consistent edge node set; S203: Based on the spatial arrangement of the nodes in the directional consistent edge node cluster, connect the continuous nodes and remove the path segments whose lateral or longitudinal distance changes exceed the average edge segment length, combine them into directionally coherent edge path segments, and obtain the displacement map of continuous edge directional path segments. S204: Call the continuous edge direction path segment displacement map, locate and map the direction distribution vector of the path segment according to the spatial coordinates of the anatomical structure, and convert the mapping result into a direction vector trajectory to establish a nerve boundary direction trajectory map.

4. The neurosurgical imaging diagnostic method based on image processing according to claim 3, characterized in that, The specific steps for obtaining the spatial distribution of the structural fracture are as follows: S301: Based on the brain lobe edge region marked in the neural boundary direction trajectory map, extract the direction vectors of nerve fibers located in the region in the continuous slices, and arrange the direction vectors in sequence according to the slice number to obtain the nerve fiber rotation trajectory sequence. S302: Call the nerve fiber rotation trajectory sequence, calculate the angle change value between adjacent direction vectors, and compare the average value and the maximum value of the direction change in three consecutive slices to filter the trajectory segments whose direction change amplitude exceeds twice the average change range, and obtain the distribution of direction change trajectory segments. S303: Based on the number of consecutive jumps between adjacent abrupt change segments in the distribution of the directional abrupt change trajectory segments, the jump segments are compared segment by segment in spatial position to identify the path regions in the trajectory where consecutive jumps occur, and the regions are located and marked in three-dimensional space according to image slice coordinates to obtain the spatial distribution of structural fractures. The spatial distribution values ​​of the continuous jump path region are calculated using the following formula: ； in, Spatial distribution values ​​representing regions with continuous jump paths. Representing the Spatial coordinates of the jump path Representing the Spatial coordinates of the jump path Indicates the first and the The absolute distance between jump paths. Representing the The jump range of the jump path. Representing the The jump stability coefficient of a segment jump path. It is the ratio of the jump amplitude to the stability coefficient. Representing the Weighting factor for segment jump paths, This represents the total number of jump segments.

5. The neurosurgical imaging diagnostic method based on image processing according to claim 4, characterized in that, The specific steps for obtaining the continuous structure infill set are as follows: S401: Call the directional path segments that have not broken in the spatial distribution of the structural fracture, collect the directional vector values ​​of adjacent voxels within the boundary range, determine the extension direction of the path segments, connect the three-dimensional spatial paths according to the voxel points with the same direction, and generate a three-dimensional directional path voxel set. S402: Based on the gaps in space of the path segments in the three-dimensional directional path voxel set, extract the directional information of the boundary voxels of the segments, and fill them by staggering the vertical, front-back, and left-right directions. Calculate the connectivity probability of the path units, and merge the local path units according to the splicing results to generate a continuous structure filling set.

6. The neurosurgical imaging diagnostic method based on image processing according to claim 5, characterized in that, The formula for calculating the connectivity probability of a path unit is: ； in, Representative voxels With voxels The probability of connectivity between them Representative voxels Volume in space Representative voxels Volume in space Representative voxels Directional information, Representative voxels Directional information.

7. The neurosurgical imaging diagnostic method based on image processing according to claim 1, characterized in that, The method further includes step S5: S5: Based on the path direction and connection relationship in the continuous structure filler set, divide the spatial hierarchy of the structural blocks, identify the structural connection hierarchy and spatial interlacing order, count the connection order and interlacing position, and generate a neural image spatial hierarchy structure map. The neuroimaging spatial hierarchy diagram includes a spatial hierarchy index matrix, connection structure affiliation clusters, and staggered orientation mapping relationships.

8. The neurosurgical imaging diagnostic method based on image processing according to claim 7, characterized in that, The specific steps for obtaining the neural image spatial hierarchy structure map are as follows: S501: Based on the extension direction and connection relationship of the path segments in the continuous structure infill set, extract the direction vector of the path unit in three-dimensional coordinates, and determine the belonging relationship based on the position between the connection points, divide the structure block set with primary and secondary connection relationship, and generate the structure block spatial connection level value. S502: Call the spatial connection hierarchy value of the structural block, identify the stacking order and staggered position between structural blocks according to the spatial arrangement order between the starting path unit and the subordinate path unit in the connection relationship, and calculate the degree of staggered position according to the overlap rate of the structural blocks in the coordinate axis direction to obtain the distribution value of the structural connection staggered order. S503: Based on the spatial position and hierarchical division results of the structural blocks in the staggered distribution value of the structural connection, the structural blocks are sequentially located in the spatial three-axis coordinate system according to the connection order, and the differentiated hierarchical markers are assigned to the corresponding position areas of the structural blocks to obtain the spatial hierarchical structure map of the neural image.

9. A neurosurgical imaging diagnostic system based on image processing, characterized in that, The system is used to implement the image processing-based neurosurgical imaging diagnostic method according to any one of claims 1-8, the system comprising: The section node extraction module obtains gray matter edge nodes in coronal, sagittal, and horizontal triaxial image sections, detects the brightness change gradient of gray matter region edges within the slice, collects the location of gray value changes, identifies whether edge feature trends are present, and generates a symmetrical path unit sequence. The path symmetry construction module, based on the symmetric path unit sequence, pairs nodes in the three-axis direction according to their spatial location, calculates the difference value of the directional projection trend of the node pair in the differential axis, filters nodes that meet the directional symmetry relationship, and generates a neural boundary directional trajectory map. The edge direction recognition module calls the neural boundary direction trajectory map, collects the edge direction vectors within the neighborhood, and compares them with the direction vectors of typical brain regions in the anatomical reference direction group in turn to determine whether the direction consistency meets the continuity standard and generates the spatial distribution of structural breaks. The continuous structural filling module determines whether there is a significant change in direction between adjacent vectors based on the spatial distribution of the structural fracture and marks the fracture area in the trajectory accordingly. After excluding the fracture segment, it calls the directional path segment that has not undergone a change in direction, detects the angle difference between the path endpoint direction and the path direction of the surrounding area, and generates a set of continuous structural filling bodies. The spatial structure partitioning module calls the direction information of each path in the continuous structure filling set, determines the direction belonging relationship between the main axis direction and the three-dimensional coordinate axis, sets the belonging level label according to the typical orientation of the brain region structure, detects the degree of spatial position overlap and connection order between paths, and generates a neuroimaging spatial hierarchy structure map. The symmetrical path unit sequence includes structural extension direction identifiers, spatial projection correspondences, and node pair set numbers. The neural boundary direction trajectory map includes edge direction annotation sequences, distribution of direction consistency segments, and a table of eliminated point indexes. The spatial distribution of structural fractures includes jump trajectory identifier areas, tension abnormality block numbers, and fracture trend extension markers. The continuous structural infill set includes path projection connection groups, staggered filling path segments, and voxel-level connected units.

Images