A method and system for analyzing SEEG connections under brain division

Abstract

The application discloses a SEEG connection analysis method and system under brain region division, and relates to the technical field of computer vision, and the method comprises the following steps: constructing a three-dimensional brain region model based on a Destrieux atlas and an automatic segmentation algorithm by fusing MRI and CT images of a patient, performing multi-modal registration to calibrate the spatial position of SEEG electrodes, and generating a first-level electrode file with anatomical labels. A virtual electrode point is generated by adopting a bipolar connection, and the brain region attribution of the electrode is dynamically corrected by combining spatial registration and prior implantation information to form a three-level label file. Accordingly, SEEG electrode pairs are grouped, sorted and identified according to brain regions, brain region-level summary signals are output, and the connectivity analysis of electrode pairs across brain regions is supported, so that the spatial interpretability and analysis accuracy of SEEG data in brain network research are improved. Through the multi-modal image fusion, virtual electrode construction and prior information guided brain region attribution correction technology, the application realizes the accurate grouping of SEEG data according to anatomical brain regions.

Description

Technical Field

[0001] This invention relates to the field of computer vision technology, specifically to a method and system for SEEG connectivity analysis based on brain region segmentation. Background Technology

[0002] Stereotactic electroencephalography (SEEG) is an important technique that directly records deep brain electrical activity using implanted intracranial electrodes. It is widely used in epileptic focus localization and brain functional network research. Current SEEG data analysis typically uses electrodes or electrode pairs as basic units, calculates functional connectivity indices (such as WPLI, PLV, etc.) based on their signal characteristics, and generates connectivity matrices for visualization.

[0003] However, existing methods generally ignore the anatomical context of the brain regions where electrodes are located, relying solely on electrode numbers or names for analysis, resulting in data organization lacking biological significance. Different electrodes may be located in the same brain region, but due to disordered naming, they are treated as independent units, making it difficult to reflect coordinated activities at the brain region level. Furthermore, individual electrodes are susceptible to noise or artifacts, affecting the stability and accuracy of the analysis results.

[0004] Furthermore, because the electrode arrangement was not categorized by brain region, the functional connectivity matrix presented a chaotic pattern, and the inter-brain connectivity structure was unclear, severely reducing the readability of the results and the efficiency of clinical interpretation. More importantly, the electrode implantation protocols varied greatly among different patients, and the lack of a data organization method based on a unified brain region framework made it difficult to achieve standardized comparisons and population analyses across subjects.

[0005] Therefore, existing SEEG analysis methods have significant limitations in terms of interpretability, anti-interference ability, and generalizability. There is an urgent need for a method that can integrate brain region anatomical information, achieve structured sorting of electrode data, and tissue preprocessing to improve analytical accuracy and clinical application value. Summary of the Invention

[0006] To address the aforementioned issues, this invention proposes a SEEG data preprocessing method and system for brain region function analysis. Through a three-level label correction technique guided by multimodal image fusion, virtual electrode construction, and prior information, the method achieves accurate attribution and structured organization of SEEG data at the anatomical brain region level, significantly improving the spatial accuracy and interpretability of brain function analysis.

[0007] On the one hand, a SEEG connectivity analysis method based on brain region segmentation includes:

[0008] S1. Acquire the patient's brain MRI image data and raw SEEG data, and automatically segment the MRI image data based on the Destrieux brain region atlas and the automatic segmentation algorithm of cortical and subcortical structures to generate a three-dimensional brain region structure model containing multiple anatomical brain region labels.

[0009] S2, perform multimodal image registration and fusion of the three-dimensional brain region structure model and the patient's CT data. Based on the registration results, mark the three-dimensional spatial coordinates of each SEEG electrode point in the original SEEG data on the fused image, and assign a corresponding brain region anatomical label to each SEEG electrode point to obtain a first-level electrode label file.

[0010] S3, construct bipolar electrode pair signals from the original SEEG data using bipolar leads, generate bipolar leaded SEEG data, use the geometric midpoint of the bipolar electrode pair signal as the virtual electrode point, re-register based on the spatial position of the virtual electrode point in the fused image, generate brain region label information based on the registration result and the primary electrode label file, and save it as a secondary electrode label file.

[0011] S4. Using the virtual electrode point as the center, search for brain regions within a preset radius. If multiple candidate brain regions exist, the virtual electrode point is an electrode point with uncertain attribution. For electrode points with uncertain attribution, use the preoperative expected electrode implantation location as prior information, and combine it with the spatial distance score from the electrode to the candidate brain region, the brain region voxel overlap ratio score, and the segmentation confidence score to obtain a weighted attribution score for the candidate brain region. Based on the attribution score of the candidate brain region and the preoperative expected location, determine the main attribution brain region of the electrode point, thereby generating a tertiary electrode label file that has been corrected by prior information based on the secondary electrode label file.

[0012] S5, based on the three-level electrode tag file, adds the main brain region identifier to each bipolar electrode pair in the SEEG data of bipolar leads, generates electrode pair tags with brain region identifiers, and groups and regularizes all electrode pairs with brain region identifiers according to brain region identification, generating bipolar lead SEEG data with electrode pairs arranged according to brain region tissue.

[0013] S6, the SEEG data of bipolar leads arranged by the electrode pairs according to the brain region tissue is preprocessed and used as input, the signals of each electrode pair according to the brain region are summarized and processed, and the summarized signal is output as brain region level data.

[0014] S7, based on the three-level electrode tag file, extracts and groups the corresponding electrode pairs at the electrode level according to brain regions to form electrode pair groups for each brain region; performs pairwise pairing processing on electrode pair groups between any two brain regions and calculates the correlation or connectivity index of the pairing.

[0015] Furthermore, in S1, the automatic segmentation algorithm uses statistical parameter mapping to automatically segment MRI image data, and verifies the brain region boundaries of the segmentation results through color coding visualization.

[0016] Furthermore, in S2, the method for generating the primary electrode label file is as follows: the three-dimensional brain region structure model is aligned with the patient's CT data using a multimodal image registration algorithm, and the three-dimensional spatial coordinates of each electrode are calibrated in the CT volume based on the registration results. With each electrode point as the center, candidate brain regions are searched within a preset radius, one or more candidate brain region labels are added to the electrode point, and the assignment probability of each candidate label is recorded. The spatial coordinates of the electrode point, the candidate brain region labels and their assignment probabilities are saved together to generate the primary electrode label file.

[0017] Furthermore, in S3, the bipolar lead is used to construct bipolar electrode pair signals by matching the original SEEG data in a predetermined manner, and calculate the geometric midpoint of the bipolar electrode pair signals as virtual electrode points. The secondary electrode tag file is generated based on the spatial location of the virtual electrode points in the fused image and records the corresponding brain region tag information.

[0018] Furthermore, in S4, the spatial distance score is calculated as follows: the shortest Euclidean distance from the virtual electrode point to the surface of the candidate brain region or the set of voxels is calculated, and the shortest Euclidean distance is monotonically normalized and mapped to a preset radius to obtain the spatial distance score.

[0019] Furthermore, in S4, the brain region voxel overlap ratio score is calculated as follows: the number of voxels in the sphere and the number of voxels overlapping with the candidate brain region are counted within a sphere with the virtual electrode point as the center and a preset radius. The ratio of the number of voxels overlapping with the candidate brain region to the number of voxels in the sphere is used as the brain region voxel overlap ratio score.

[0020] Furthermore, in S4, the weighted calculation method for the attribution score of the candidate brain region is as follows: the primary attribution brain region of the electrode point is determined based on the attribution score of the candidate brain region and the preoperative expected location. By customizing the weights among the given spatial distance score, brain region voxel overlap ratio score, segmentation confidence score, and preoperative expected location, the attribution score of each candidate brain region of the electrode point is calculated after the sum of the weights of the four factors is 1. The subsequent brain regions are then sorted according to the attribution scores to obtain the top two candidate brain regions. If the difference between the two attribution scores exceeds the set value or there is no preoperative expected location, the brain region with the highest attribution score is selected as the primary attribution brain region; otherwise, the preoperative expected location is used as the primary attribution brain region.

[0021] Furthermore, in S5, during the process of grouping and arranging all electrode pairs in the original SEEG data according to brain region affiliation, the process is carried out in a preset anatomical order or a user-defined order through the interface. The specific order can be configured as from the cerebral cortex to deep nuclei, from the prefrontal cortex to the occipital lobe, or a custom grouping arrangement.

[0022] On the other hand, a SEEG connectivity analysis system based on brain region segmentation includes:

[0023] The segmentation module is used to acquire the patient's brain MRI image data and raw SEEG data, and automatically segment the MRI image data based on the Destrieux brain region atlas and the automatic segmentation algorithm of cortical and subcortical structures to generate a three-dimensional brain region structure model containing multiple anatomical brain region labels.

[0024] The primary electrode label generation module is used to perform multimodal image registration and fusion of the three-dimensional brain region structure model and the patient's CT data. Based on the registration results, the three-dimensional spatial coordinates of each SEEG electrode point in the original SEEG data are marked on the fused image, and a corresponding brain region anatomical label is assigned to each SEEG electrode point to obtain the primary electrode label file.

[0025] The secondary electrode tag generation module constructs bipolar electrode pair signals from the original SEEG data using bipolar leads, generates bipolar leaded SEEG data, uses the geometric midpoint of the bipolar electrode pair signal as a virtual electrode point, re-registers based on the spatial position of the virtual electrode point in the fused image, generates brain region tag information based on the registration result and the primary electrode tag file, and saves it as a secondary electrode tag file.

[0026] The Level 3 electrode tag generation module searches for brain regions within a preset radius, using the virtual electrode point as the center. If multiple candidate brain regions exist, the virtual electrode point is considered to have an uncertain attribution. For electrode points with uncertain attribution, the preoperatively expected electrode implantation location is used as prior information. Combined with the spatial distance score from the electrode to the candidate brain region, the brain region voxel overlap ratio score, and the segmentation confidence score, a weighted attribution score for the candidate brain region is obtained. Based on the attribution score of the candidate brain region and the preoperatively expected location, the main brain region to which the electrode point belongs is determined, thereby generating a Level 3 electrode tag file that has been corrected by prior information based on the Level 2 electrode tag file.

[0027] SEEG data of bipolar leads with electrode pairs arranged according to brain region tissue is based on a three-level electrode tag file. In the SEEG data of bipolar leads, the main brain region identifier is added to each bipolar electrode pair to generate electrode pair tags with brain region identifiers. All electrode pairs with brain region identifiers are grouped and arranged in a regular manner according to brain region identification to generate SEEG data of bipolar leads with electrode pairs arranged according to brain region tissue.

[0028] The input-output construction module takes the SEEG data of bipolar leads arranged by the electrode pairs according to the brain region tissue as input after preprocessing, performs summary processing on the signals of each electrode pair according to the brain region, and outputs the summary signal as brain region-level data.

[0029] The pairing calculation module, based on the three-level electrode tag file, extracts and groups the corresponding electrode pairs at the electrode level according to brain regions, forming electrode pair groups for each brain region; it performs pairwise pairing processing on electrode pair groups between any two brain regions and calculates the correlation or connectivity index of the pairing.

[0030] The present invention adopts the above technical solution and has the following beneficial effects:

[0031] (1) This invention integrates MRI and CT images, combines Destrieux brain region atlas for automatic segmentation and multimodal registration, and introduces bipolar leads to construct virtual electrode points, which effectively corrects the deviation of traditional single electrode positioning and improves the accuracy of matching electrode spatial position with corresponding brain regions.

[0032] (2) In view of the uncertainty of the attribution caused by electrodes crossing multiple brain regions, the present invention innovatively introduces the expected implantation location before surgery as prior information, prioritizes the determination of the main brain regions to which the electrodes belong, and significantly improves the rationality and stability of label allocation in complex situations.

[0033] (3) Based on the revised three-level electrode tag file, the present invention groups and arranges the electrode pairs according to brain regions and generates brain region-level summary signals. This not only realizes the orderly management of data, but also provides a standardized and highly interpretable data foundation for subsequent functional connectivity and network dynamic analysis between brain regions. Attached Figure Description

[0034] Figure 1 This is a flowchart of the SEEG connectivity analysis method based on brain region segmentation according to an embodiment of the present invention;

[0035] Figure 2 This is a schematic diagram of a segmented brain model according to an embodiment of the present invention;

[0036] Figure 3 This is a schematic diagram illustrating the addition of a new brain region prefix to the electrode pair according to an embodiment of the present invention;

[0037] Figure 4 This is a schematic diagram comparing the brain region prefix before and after adding it according to an embodiment of the present invention;

[0038] Figure 5 This is a schematic diagram showing the before and after rearrangement of brain regions according to an embodiment of the present invention;

[0039] Figure 6 This is a schematic diagram of the WPLI matrix of the original electrode pair arrangement in an embodiment of the present invention;

[0040] Figure 7 This is a schematic diagram of the WPLI matrix of electrode pairs arranged according to brain regions in an embodiment of the present invention;

[0041] Figure 8This is a diagram of the SEEG connectivity analysis system under brain region division in an embodiment of the present invention. Detailed Implementation

[0042] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0043] like Figure 1 As shown, the present invention provides a SEEG connectivity analysis method based on brain region segmentation, comprising:

[0044] S1. Acquire the patient's brain MRI image data and raw SEEG data, and automatically segment the MRI image data based on the Destrieux brain region atlas and the automatic segmentation algorithm of cortical and subcortical structures to generate a three-dimensional brain region structure model containing multiple anatomical brain region labels.

[0045] Specifically, in S1, the automatic segmentation algorithm uses statistical parameter mapping to automatically segment MRI image data, and verifies the brain region boundaries of the segmentation results through color coding visualization.

[0046] Specifically, in this embodiment, brain region segmentation is first performed. Patient MRI data is collected, preprocessed on a computing platform, converted to a standard volumetric image format, and an automatic segmentation operation is initiated. A three-dimensional brain region model containing multiple anatomical brain region labels is generated using an automatic segmentation algorithm based on the Destrieux (2009) atlas combined with subcortical structures. The results are then output using a color-coded visualization method, such as... Figure 2 The image shows the segmented brain model (including brain region labels), where different brain regions are marked with different colors, such as green for the left thalamus (label 10), blue for the left globus pallidus (label 13), and pink for the right inferior lateral ventricle (label 44). The confidence index is obtained through the probability mapping of the segmentation algorithm itself (if SPM is used, the probability map can be used directly), or through cross-validation based on multiple resampling / a small amount of manual annotation, to verify the clarity of brain region boundaries and the accuracy of segmentation.

[0047] S2 performs multimodal image registration and fusion of the three-dimensional brain region structure model and the patient's CT data. Based on the registration results, the three-dimensional spatial coordinates of each SEEG electrode point in the original SEEG data are marked on the fused image, and a corresponding brain region anatomical label is assigned to each SEEG electrode point to obtain a first-level electrode label file.

[0048] Specifically, the method for generating the primary electrode label file is as follows: The three-dimensional brain region structure model is aligned with the patient's CT data using a multimodal image registration algorithm. Based on the registration results, the three-dimensional spatial coordinates of each electrode are calibrated in the CT volume. With each electrode point as the center, candidate brain regions are searched within a preset radius. One or more candidate brain region labels are added to the electrode point, and the assignment probability of each candidate label is recorded. The spatial coordinates of the electrode point, the candidate brain region labels, and their assignment probabilities are saved together to generate the primary electrode label file.

[0049] Specifically, during the calibration process, the segmented brain region structural images and the patient's CT data are imported into the image analysis workflow. Multimodal image spatial alignment is achieved through image registration methods based on statistical imaging. On this basis, the three-dimensional coordinates of each SEEG electrode point are manually or automatically calibrated, and corresponding brain region labels (such as "ASEG_prob_DKT_White_L" and "Frontal_Inf_Opercular_L") are assigned according to their spatial location, generating a label file containing electrode name, coordinates, brain region label, and assignment probability.

[0050] Specifically, the electrode calibration process includes three steps: data import and registration, electrode location calibration, and label file export. First, the structural images generated from brain region segmentation and CT data are imported into the image analysis workflow, and spatial alignment of multimodal images is achieved through image registration methods based on statistical imaging. Next, based on the registered data, the electrode point coordinates are manually calibrated, and a specific brain region label (e.g., “ASEG_prob_DKT_White_L”, “Frontal_Inf_Opercular_L”) is assigned to each electrode point according to brain region anatomy information. Finally, a label file containing the electrode name, brain region label, coordinates, and probability values ​​is generated. Electrode names such as “SCS_0 (0.154, 32.524, 41.421)” and their corresponding brain region labels cover regions such as the frontal lobe, temporal lobe, parietal lobe, and white matter structures, ensuring that detailed electrode point information and their probability values ​​for location within the brain region are recorded.

[0051] S3. Bipolar leads are used to construct bipolar electrode pair signals from the original SEEG data to generate bipolar leaded SEEG data. The geometric midpoint of the bipolar electrode pair signal is used as a virtual electrode point. Based on the spatial position of the virtual electrode point in the fused image, re-registration is performed. Brain region label information is generated based on the registration results and the primary electrode label file and saved as a secondary electrode label file.

[0052] Specifically, bipolar leads are used to construct bipolar electrode pair signals by matching raw SEEG data in a predetermined manner. Based on surgical records or electrode naming rules (e.g., adjacent contact points A1-A2, B1-B2, etc.), bipolar references are constructed, and the geometric midpoint of the bipolar electrode pair signals is calculated as a virtual electrode point. The secondary electrode tag file is generated based on the spatial location of the virtual electrode point in the fused image and records the corresponding brain region tag information.

[0053] Specifically, in this embodiment, the virtual electrode point is obtained by taking the arithmetic mean of the coordinates of the two electrode points as the virtual electrode point coordinates. Based on the spatial position of the virtual electrode point in the fused image, the registration is re-performed, and the previous first-level electrode labeling steps are repeated twice according to the newly generated virtual electrode points.

[0054] S4. Using the virtual electrode point as the center, search for brain regions within a preset radius. If multiple possible candidate brain regions exist, the virtual electrode point is an electrode point with uncertain attribution. For electrode points with uncertain attribution, use the preoperative expected electrode implantation location as prior information, and combine it with the spatial distance score from the electrode to the candidate brain region, the brain region voxel overlap ratio score, and the segmentation confidence score to obtain a weighted attribution score for the candidate brain region. Based on the attribution score of the candidate brain region and the preoperative expected location, determine the main attribution brain region of the electrode point, thereby generating a tertiary electrode label file that has been corrected by prior information based on the secondary electrode label file.

[0055] Specifically, the spatial distance score S_d is calculated as follows: the shortest Euclidean distance d from the virtual electrode point to the surface of the candidate brain region c or the voxel set is calculated, and a monotonically normalized mapping S_d = max(0, 1 - d / R_search), (where R_search is a preset radius), is used to obtain the spatial distance score. The brain region voxel overlap ratio score S_o is calculated as follows: the number of voxels N_sphere and the number of voxels N_overlap that overlap with the candidate brain region are counted within a sphere with the virtual electrode point as the center and a preset radius. The ratio of the number of voxels overlapping with the candidate brain region to the number of voxels in the sphere, S_o = (N_overlap / N_sphere), is used as the brain region voxel overlap ratio score. The segmentation confidence score S_c is calculated as follows: the mean of the probability value or confidence value of the candidate brain region during segmentation mapped onto the sphere, in the range [0,1]. The preoperative prior score S_p is calculated as follows: if the electrode point has a planned brain region, then the preoperative prior score S_p for the planned brain region is 1; if there is no preoperative prior or the planned brain region is not in the planned brain region, then the preoperative prior score S_p for the planned brain region is 0. The weighted calculation method for the attribution score of candidate brain regions is as follows: Based on the attribution score of the candidate brain region and the preoperative expected location, the primary attribution brain region for the electrode point is determined. A custom weighting is used between the given spatial distance score, brain region voxel overlap ratio score, segmentation confidence score, and preoperative expected location. The weights are defined as w_d, w_o, w_c, and w_p, satisfying w_d + w_o + w_c + w_p = 1. The attribution score for each candidate brain region at the electrode point is calculated as Score = w_d * S_d + w_o * S_o + w_c * S_c + w_p * S_p. Subsequent brain regions are then ranked according to their attribution scores, obtaining the top two candidate brain regions. If the difference between the two attribution scores exceeds a set value Δ or there is no preoperative expected location, the brain region with the highest attribution score is selected as the primary attribution brain region; otherwise, the preoperative expected location is used as the primary attribution brain region.

[0056] S5, based on the three-level electrode tag file, adds the main brain region identifier to each bipolar electrode pair in the SEEG data of bipolar leads, generates electrode pair tags with brain region identifiers, and groups and regularizes all electrode pairs with brain region identifiers according to brain region identification, generating SEEG data of bipolar leads with electrode pairs arranged according to brain region tissue.

[0057] Specifically, during the grouping and regular arrangement of all electrode pairs with brain region identifiers, brain region prefixes are added to each bipolar electrode pair according to brain region affiliation (such as "HippocampusL-A1-A2" or "InsulaR-B3-B4"). This process follows a preset anatomical order or a user-defined order through the interface. The specific order can be configured as from the cerebral cortex to deep nuclei, from the prefrontal lobe to the occipital lobe, or a custom grouping arrangement.

[0058] S6 takes the SEEG data of bipolar leads arranged by the electrode pairs according to the brain region tissue as input after preprocessing, performs summary processing on the signals of each electrode pair according to the brain region, and outputs the summary signal as brain region-level data.

[0059] Specifically, in the SEEG data preprocessing process, the raw data first undergoes basic preprocessing, including downsampling to reduce data volume, filtering to extract target frequency band signals, and removing bad leads. Next, bipolar reference processing is employed, using the midpoint between two electrode points as new virtual electrode points, and recalibrating to obtain brain region labels for these virtual electrode points. For brain region label addition and electrode pair sorting, based on the matching relationship between electrode coordinates and brain region labels (e.g., "Transverse temporal_L", "Insula L"), a brain region prefix is ​​added to each bipolar electrode pair (e.g., optimizing "A1-A2" to "Amygdala L-A1-A2") to achieve precise association (e.g., ...). Figure 3 Adding new brain region prefixes to electrode pairs to achieve structured optimization of label information (e.g.) Figure 4 (Comparison before and after adding brain region prefixes). Subsequently, sorting code is written to parse the brain region prefixes in electrode pair names (e.g., "Transverse temporal L", "Hippocampus L"), classifies and groups them by brain region using a sorting algorithm, and groups and sorts the electrode pairs according to a custom anatomical order (e.g., from cortex to deep nuclei), ensuring that electrode pairs with the same brain region are grouped together (e.g., ...). Figure 5 (Comparison before and after rearranging based on brain regions). The image shows a comparison before and after the sorting. The left side is the original list of electrode pairs, and the right side is the structured result after sorting by brain region prefix. Figure 5 It demonstrates the effect of rearranging brain regions, and Figure 4 and Figure 3The document demonstrates the comparison before and after adding brain region prefixes, as well as the process of adding new brain region prefixes to electrode pairs. The electrode calibration process includes three steps: data import and registration, electrode location calibration, and label file export. First, the structural images generated by brain region segmentation and CT data are imported into the image analysis workflow, and spatial alignment of multimodal images is achieved through image registration methods based on statistical imaging. Next, based on the registered data, the electrode point coordinates are manually or automatically calibrated, and a specific brain region label (e.g., "ASEG_prob_DKT_White_L", "Frontal_Inf_Opercular_L") is assigned to each electrode point according to brain region anatomy information. Finally, a label file containing electrode name, brain region label, coordinates, and probability value is generated. Electrode names such as "SCS_0 (0.154, 32.524, 41.421)" and their corresponding brain region labels cover regions such as the frontal lobe, temporal lobe, and parietal lobe, as well as white matter structures, ensuring that detailed electrode point information and their probability values ​​for location within the brain region are recorded.

[0060] Specifically, in this embodiment, data from the 128-132 second time period and the 30-80 Hz frequency band (gamma band) of the SEEG data were selected to calculate the weighted phase lag index (WPLI) between all electrode pairs. This was used to analyze functional connectivity between brain regions and to plot the WPLI matrix. Figure 6 (As shown in the original electrode pair arrangement), the matrix has no obvious grouping, and the brain region connectivity patterns are blurred; while as... Figure 7 As shown in the diagram (after brain region sorting), the data are grouped by brain region (e.g., White L, Superior Temporal L, Hippocampus L, etc.) and color-coded (blue 0.1 - red 0.5) to visually reflect phase synchronization intensity. This makes the connectivity patterns between brain regions clearly distinguishable, significantly improving the readability and interpretability of the results. This processing not only enhances visual discriminability but also makes the functional connectivity between different brain regions clearer, facilitating in-depth research.

[0061] S7, based on the three-level electrode tag file, extracts and groups the corresponding electrode pairs at the electrode level according to brain regions to form electrode pair groups for each brain region; performs pairwise pairing processing on electrode pair groups between any two brain regions and calculates the correlation or connectivity index of the pairing.

[0062] Specifically, when calculating indicators, multiple indicators can be calculated in parallel and are supported, including but not limited to phase-related indicators: Phase Lock Value (PLV), Weighted Phase Lag Index (WPLI), and Phase Lag Index (PLI); frequency-related indicators: coherence and power correlation; and directional indicators such as partially directional coherence (PDC) and directional transfer function (DTF) based on MVAR.

[0063] In summary, this invention offers the following advantages: improved interference resistance and accuracy: by integrating electrode signals through brain region grouping, the impact of noise from individual electrode points on the results is reduced, improving analytical precision, more accurately locating pathological signals, and reducing the risk of misjudgment; anatomical background association: binding electrode pairs with brain region labels directly links SEEG data to brain anatomical structures, enhancing the biological significance of the data and facilitating the understanding of the relationship between brain function and pathological mechanisms; standardization and commonality revelation: by standardizing data among different subjects based on the analysis pattern of brain regions, it can reveal common activity patterns specific to brain regions, providing reliable support for brain function research and clinical applications (such as epileptic focus localization).

[0064] like Figure 8 As shown, this embodiment also discloses a SEEG connectivity analysis system based on brain region segmentation, comprising:

[0065] The segmentation module 81 is used to acquire the patient's brain MRI image data and raw SEEG data, and automatically segment the MRI image data based on the Destrieux brain region atlas and the automatic segmentation algorithm of cortical and subcortical structures to generate a three-dimensional brain region structure model containing multiple anatomical brain region labels.

[0066] The primary electrode label generation module 82 is used to perform multimodal image registration and fusion of the three-dimensional brain region structure model and the patient's CT data. Based on the registration result, the three-dimensional spatial coordinates of each SEEG electrode point in the original SEEG data are marked on the fused image, and a corresponding brain region anatomical label is assigned to each SEEG electrode point to obtain the primary electrode label file.

[0067] The secondary electrode tag generation module 83 constructs bipolar electrode pair signals from the original SEEG data using bipolar leads, generates bipolar leaded SEEG data, uses the geometric midpoint of the bipolar electrode pair signal as a virtual electrode point, re-registers based on the spatial position of the virtual electrode point in the fused image, generates brain region tag information based on the registration result and the primary electrode tag file, and saves it as a secondary electrode tag file.

[0068] The Level 3 electrode tag generation module 84 searches for brain regions within a preset radius, using the virtual electrode point as the center. If multiple candidate brain regions exist, the virtual electrode point is an electrode point with uncertain attribution. For electrode points with uncertain attribution, the expected electrode implantation location before surgery is used as prior information. Combined with the spatial distance score from the electrode to the candidate brain region, the brain region voxel overlap ratio score, and the segmentation confidence score, a weighted attribution score for the candidate brain region is obtained. Based on the attribution score of the candidate brain region and the expected location before surgery, the main brain region to which the electrode point belongs is determined, thereby generating a Level 3 electrode tag file that has been corrected by prior information based on the Level 2 electrode tag file.

[0069] The ordered electrode pair generation module 85, based on the three-level electrode tag file, adds the main brain region identifier to each bipolar electrode pair in the SEEG data of the bipolar lead, generates electrode pair tags with brain region identifiers, and groups and regularizes all electrode pairs in the original SEEG data according to brain region attribution, generating an ordered electrode pair sequence organized by brain region.

[0070] The input / output construction module 86 takes the preprocessed raw SEEG data corresponding to the ordered electrode pair sequence as input, performs summary processing on the signals of each electrode pair according to brain region, and outputs the summary signal as brain region-level data.

[0071] The pairing calculation module 87, based on the three-level electrode tag file, extracts and groups the corresponding electrode pairs at the electrode level according to brain regions to form electrode pair groups for each brain region; performs pairwise pairing processing on electrode pair groups between any two brain regions, and calculates the correlation or connectivity index of the pairing.

[0072] A specific implementation of a SEEG connectivity analysis system based on brain region segmentation is described in this embodiment. The SEEG connectivity analysis method based on brain region segmentation will not be described again in this embodiment.

[0073] Although the invention has been specifically shown and described in conjunction with preferred embodiments, those skilled in the art should understand that various changes in form and detail may be made to the invention without departing from the spirit and scope of the invention as defined in the appended claims, all of which shall be within the scope of protection of the invention.

Claims

1. A SEEG connectivity analysis method based on brain region segmentation, characterized in that, Includes the following steps: S1. Acquire the patient's brain MRI image data and raw SEEG data, and automatically segment the MRI image data based on the Destrieux brain region atlas and the automatic segmentation algorithm of cortical and subcortical structures to generate a three-dimensional brain region structure model containing multiple anatomical brain region labels. S2, perform multimodal image registration and fusion of the three-dimensional brain region structure model and the patient's CT data. Based on the registration results, mark the three-dimensional spatial coordinates of each SEEG electrode point in the original SEEG data on the fused image, and assign a corresponding brain region anatomical label to each SEEG electrode point to obtain a first-level electrode label file. S3, construct bipolar electrode pair signals from the original SEEG data using bipolar leads, generate bipolar leaded SEEG data, use the geometric midpoint of the bipolar electrode pair signal as the virtual electrode point, re-register based on the spatial position of the virtual electrode point in the fused image, generate brain region label information based on the registration result and the primary electrode label file, and save it as a secondary electrode label file. S4. Using the virtual electrode point as the center, search for brain regions within a preset radius. If multiple candidate brain regions exist, the virtual electrode point is an electrode point with uncertain attribution. For electrode points with uncertain attribution, use the preoperative expected electrode implantation location as prior information, and combine it with the spatial distance score from the electrode to the candidate brain region, the brain region voxel overlap ratio score, and the segmentation confidence score to obtain a weighted attribution score for the candidate brain region. Based on the attribution score of the candidate brain region and the preoperative expected location, determine the main attribution brain region of the electrode point, thereby generating a tertiary electrode label file that has been corrected by prior information based on the secondary electrode label file. S5, based on the three-level electrode tag file, adds the main brain region identifier to each bipolar electrode pair in the SEEG data of bipolar leads, generates electrode pair tags with brain region identifiers, and groups and regularizes all electrode pairs with brain region identifiers according to brain region identification, generating bipolar lead SEEG data with electrode pairs arranged according to brain region tissue. S6, the SEEG data of bipolar leads arranged by the electrode pairs according to the brain region tissue is preprocessed and used as input, the signals of each electrode pair according to the brain region are summarized and processed, and the summarized signal is output as brain region level data. S7, based on the three-level electrode tag file, extracts and groups the corresponding electrode pairs at the electrode level according to brain regions to form electrode pair groups for each brain region; performs pairwise pairing processing on electrode pair groups between any two brain regions and calculates the correlation or connectivity index of the pairing.

2. The SEEG connectivity analysis method based on brain region segmentation according to claim 1, characterized in that, In S1, the automatic segmentation algorithm uses statistical parameter mapping to automatically segment MRI image data, and verifies the brain region boundaries of the segmentation results through color coding visualization.

3. The SEEG connectivity analysis method based on brain region segmentation according to claim 1, characterized in that, In S2, the generation of the primary electrode label file is as follows: The three-dimensional brain region structure model is aligned with the patient's CT data using a multimodal image registration algorithm. Based on the registration results, the three-dimensional spatial coordinates of each electrode are calibrated in the CT volume. With each electrode point as the center, candidate brain regions are searched within a preset radius. One or more candidate brain region labels are added to the electrode point, and the assignment probability of each candidate label is recorded. The spatial coordinates of the electrode point, the candidate brain region labels and their assignment probabilities are saved together to generate the primary electrode label file.

4. The SEEG connectivity analysis method based on brain region segmentation according to claim 1, characterized in that, In S3, the bipolar lead is used to construct bipolar electrode pair signals by matching the raw SEEG data in a predetermined manner, and calculate the geometric midpoint of the bipolar electrode pair signals as virtual electrode points. The secondary electrode tag file is generated based on the spatial location of the virtual electrode points in the fused image and records the corresponding brain region tag information.

5. The SEEG connectivity analysis method based on brain region segmentation according to claim 1, characterized in that, In S4, the spatial distance score is calculated as follows: the shortest Euclidean distance from the virtual electrode point to the surface of the candidate brain region or the set of voxels is calculated, and the shortest Euclidean distance is monotonically normalized and mapped to a preset radius to obtain the spatial distance score.

6. The SEEG connectivity analysis method based on brain region segmentation according to claim 1, characterized in that, In S4, the brain region voxel overlap ratio score is calculated as follows: the number of voxels in the sphere and the number of voxels overlapping with the candidate brain region are counted in a sphere with the virtual electrode point as the center and the radius as the preset radius. The ratio of the number of voxels overlapping with the candidate brain region to the number of voxels in the sphere is used as the brain region voxel overlap ratio score.

7. The SEEG connectivity analysis method based on brain region segmentation according to claim 1, characterized in that, In S4, the weighted calculation method for the attribution score of the candidate brain region is as follows: the primary attribution brain region of the electrode point is determined based on the attribution score of the candidate brain region and the preoperative expected location. By defining the weights among the given spatial distance score, brain region voxel overlap ratio score, segmentation confidence score and preoperative expected location, the attribution score of each candidate brain region of the electrode point is calculated after the sum of the weights of the four factors is 1. The subsequent brain regions are then sorted according to the attribution scores to obtain the top two candidate brain regions. If the difference between the attribution scores of the two regions exceeds the set value or there is no preoperative expected location, the brain region with the highest attribution score is selected as the primary attribution brain region; otherwise, the preoperative expected location is used as the primary attribution brain region.

8. The SEEG connectivity analysis method based on brain region segmentation according to claim 1, characterized in that, In S5, during the process of grouping and arranging all electrode pairs in the original SEEG data according to brain region affiliation, the process is carried out in a preset anatomical order or a user-defined order through the interface. The specific order can be configured as from the cerebral cortex to deep nuclei, from the prefrontal lobe to the occipital lobe, or a custom grouping arrangement.

9. A SEEG connectivity analysis system based on brain region segmentation, characterized in that, include: The segmentation module is used to acquire the patient's brain MRI image data and raw SEEG data, and automatically segment the MRI image data based on the Destrieux brain region atlas and the automatic segmentation algorithm of cortical and subcortical structures to generate a three-dimensional brain region structure model containing multiple anatomical brain region labels. The primary electrode label generation module is used to perform multimodal image registration and fusion of the three-dimensional brain region structure model and the patient's CT data. Based on the registration results, the three-dimensional spatial coordinates of each SEEG electrode point in the original SEEG data are marked on the fused image, and a corresponding brain region anatomical label is assigned to each SEEG electrode point to obtain the primary electrode label file. The secondary electrode tag generation module constructs bipolar electrode pair signals from the original SEEG data using bipolar leads, generates bipolar leaded SEEG data, uses the geometric midpoint of the bipolar electrode pair signal as a virtual electrode point, re-registers based on the spatial position of the virtual electrode point in the fused image, generates brain region tag information based on the registration result and the primary electrode tag file, and saves it as a secondary electrode tag file. The Level 3 electrode tag generation module searches for brain regions within a preset radius, using the virtual electrode point as the center. If multiple candidate brain regions exist, the virtual electrode point is considered to have an uncertain attribution. For electrode points with uncertain attribution, the preoperatively expected electrode implantation location is used as prior information. Combined with the spatial distance score from the electrode to the candidate brain region, the brain region voxel overlap ratio score, and the segmentation confidence score, a weighted attribution score for the candidate brain region is obtained. Based on the attribution score of the candidate brain region and the preoperatively expected location, the main brain region to which the electrode point belongs is determined, thereby generating a Level 3 electrode tag file that has been corrected by prior information based on the Level 2 electrode tag file. SEEG data of bipolar leads with electrode pairs arranged according to brain region tissue is based on a three-level electrode tag file. In the SEEG data of bipolar leads, the main brain region identifier is added to each bipolar electrode pair to generate electrode pair tags with brain region identifiers. All electrode pairs with brain region identifiers are grouped and arranged in a regular manner according to brain region identification to generate SEEG data of bipolar leads with electrode pairs arranged according to brain region tissue. The input-output construction module takes the SEEG data of bipolar leads arranged by the electrode pairs according to the brain region tissue as input after preprocessing, performs summary processing on the signals of each electrode pair according to the brain region, and outputs the summary signal as brain region-level data. The pairing calculation module, based on the three-level electrode tag file, extracts and groups the corresponding electrode pairs at the electrode level according to brain regions, forming electrode pair groups for each brain region; it performs pairwise pairing processing on electrode pair groups between any two brain regions and calculates the correlation or connectivity index of the pairing.

Images