Hippocampal surface reference map mapping method, mechanism evaluation method and medium

Abstract

The application relates to a hippocampus surface reference atlas mapping method, a mechanism evaluation method and a medium in the technical field of neuroimaging. The hippocampus surface reference atlas mapping method comprises the following steps: mapping an external reference atlas to a reference hippocampus surface grid of a preset template brain to obtain the reference hippocampus surface grid and atlas value distribution thereof. A vertex correspondence relationship between a hippocampus surface grid of a subject and the reference hippocampus surface grid is established, the atlas value distribution on the reference hippocampus surface grid is transferred to the hippocampus surface grid of the subject, and a hippocampus surface reference atlas is obtained. According to the application, the external reference atlas is first mapped to the reference hippocampus surface grid of the preset template brain to obtain the atlas value distribution on the reference hippocampus surface grid, then based on the vertex correspondence relationship between the hippocampus surface grid of the subject and the reference hippocampus surface grid, the hippocampus surface reference atlas is obtained, and therefore the stability of atlas mapping and the reliability of statistical significance evaluation in hippocampus mechanism interpretation are improved.

Description

Technical Field

[0001] This invention relates to the field of neuroimaging technology, and in particular to a method for mapping reference maps of the hippocampus surface, a method for evaluating mechanisms, and a medium. Background Technology

[0002] In recent years, the fields of neuroimaging and neuroscience have accumulated and published a large number of brain atlases. Researchers increasingly tend to compare their own obtained disease variation patterns, individual deviation distribution maps, or other statistical results with these external reference atlases to obtain more biologically meaningful mechanistic explanations. To address the need for "cross-coordinate system access / conversion / comparison of brain atlases," existing tools have proposed analytical frameworks that generate high-quality transformations between multiple coordinate systems, provide native spatial reference atlas libraries, and consider spatial autocorrelation when assessing atlas similarity. However, existing general frameworks mostly focus on the whole brain or the cortex. For subcortical structures like the hippocampus, which are small in volume, highly folded, and exhibit significant long-axis organizational regularity, achieving a standardized closed loop of "external atlas → hippocampal surface → individual mechanism explanation" still faces significant technical obstacles: the hippocampus is a paleocortical structure with obvious folding structures and significant individual differences, especially in the head region where folding is more complex. Traditional voxel registration or coarse template deformation cannot guarantee a stable one-to-one correspondence within the hippocampus, leading to alignment difficulties. Existing reference maps are usually in the form of whole-brain voxels. When used directly on the hippocampus, the scale and boundaries of the map may not match the hippocampus, resulting in significant differences in the results. Summary of the Invention

[0003] To address the technical problem of unstable mapping of existing external reference maps to the hippocampal surface, this invention provides a method for mapping reference maps to the hippocampal surface, a mechanism evaluation method, and a medium.

[0004] In a first aspect, the present invention provides a method for mapping a hippocampal surface reference map, which is used to map at least one external reference map to a hippocampal surface domain with fine granularity to obtain a hippocampal surface reference map characterized in units of grid cells and / or vertices. The mapping method includes: A reference hippocampal surface mesh of a preset template brain is obtained, the reference hippocampal surface mesh having a fixed topological structure and vertex sequence. At least one external reference map is obtained. After registering or transforming the external reference map to a standard space consistent with the preset template brain, the external reference map is projected onto the reference hippocampal surface mesh using voxel-to-surface mapping and / or surface-to-surface mapping to obtain the map value distribution on the reference hippocampal surface mesh. A subject's hippocampal surface mesh is obtained, and the subject's hippocampal surface mesh is made to have a consistent topological structure and vertex correspondence with the reference hippocampal surface mesh. Based on the vertex correspondence, the map value distribution on the reference hippocampal surface mesh is transferred to the subject's hippocampal surface mesh to obtain a hippocampal body surface reference map.

[0005] Secondly, this invention also proposes a mechanism evaluation method based on a hippocampal surface reference map, comprising: first, obtaining a hippocampal surface reference map using the hippocampal surface reference map mapping method described in the first aspect; then, acquiring a subject's hippocampal deviation map and aligning it with the vertex sequence of the grid in the hippocampal surface reference map; calculating the correlation effect size between the subject's hippocampal deviation map and the hippocampal surface reference map to obtain an observational statistic. The correlation effect size includes a correlation coefficient, a regression coefficient, and a similarity index; constructing a constraint permutation that preserves the spatial structure in the hippocampal surface domain, generating a permutation null distribution, and obtaining a corrected significance result accordingly; and outputting the corrected significance result and / or the observational statistic for mechanism evaluation.

[0006] Thirdly, the present invention provides a computer-readable storage medium storing a computer program / instructions. When the computer program / instructions are executed by a processor, they implement the steps of the hippocampal surface reference map mapping method in the first aspect.

[0007] The beneficial effects of this invention are as follows: This invention first maps an external reference map onto a reference hippocampal surface grid of a preset template brain to obtain the map value distribution on the reference hippocampal surface grid. Then, based on the vertex correspondence between the subject's hippocampal surface grid and the reference hippocampal surface grid, a reference map of the hippocampal body surface is obtained. This makes the map mapping stable and statistically significant in the explanation of hippocampal mechanisms, and has significant practical value and prospects for promotion and application. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a flowchart of the method for mapping reference maps on the surface of the hippocampus; Figure 2 This is a schematic diagram of an external reference map mapped onto a reference hippocampus surface mesh; Figure 3 This is a flowchart of a mechanism evaluation method based on a reference map of the hippocampus surface. Detailed Implementation

[0010] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "or / and" as used herein includes any and all combinations of one or more of the associated listed items.

[0012] This invention provides a method for hippocampal surface reference map mapping and mechanism assessment based on magnetic resonance imaging (MRI) and standard spatial brain atlas data. This method forms a complete closed loop: "external reference map → hippocampal surface mapping → mechanism assessment with individual deviation maps → spatial autocorrelation correction of the hippocampal surface domain → structured reporting and version tracking." The hippocampal surface reference map mapping and mechanism assessment method described in this invention can be applied to brain atlas mapping in mammals, and mainly involves the following steps: S1. Surface representation of hippocampal structure and establishment of cross-individual correspondence.

[0013] Given the small size, complex folding, and significant individual differences of the hippocampus, this study automatically segments, reconstructs, and generates a surface mesh of the subject's hippocampus (including related structures such as the dentate gyrus). A reference hippocampal surface / mesh consistent with external brain atlases can be provided as a control. Furthermore, a stable correspondence between the hippocampal surfaces is established through intrinsic coordinates / parameter domains or equivalent topological constraints, ensuring that the subject's hippocampus and the reference hippocampus (as well as different spatial representations such as anatomical and unfolded surfaces) have a consistent vertex sequence and a comparable spatial localization basis.

[0014] S2. Acquisition and standardization of external reference maps.

[0015] Externally available or self-built brain atlases are used as input for "external reference atlases." These atlases can be in voxel format (e.g., 3D / 4D NIfTI) or standard surface format, and their content may include, but is not limited to: reference atlases related to neurotransmitters / receptors / transporters, atlases related to mitochondrial function or energy metabolism, spatial maps of gene expression, atlases of myelin / iron deposition / perfusion / microstructure proxy maps, distribution maps of pathological proteins or tracers, functional connectivity maps, network gradient maps, or spatial maps related to cognition. The external reference atlases undergo uniform spatial coordinate system alignment, resolution and intensity scale normalization, mask constraints, and metadata archiving to ensure reusability and traceability.

[0016] S3. Constrained mapping and distortion correction of voxel / surface maps to the hippocampal surface.

[0017] External reference maps are mapped from their respective whole-brain voxel space or reference surface space to their corresponding reference hippocampal surface vertices, forming a hippocampal surface reference map distribution. This distribution can then be uniformized into the subject's hippocampal surface vertex sequence based on established correspondences. The mapping process includes, but is not limited to: sampling constraints based on hippocampal tissue masks or hippocampal "ribbon regions," interpolation / kernel sampling strategies, partial volume effect suppression, and handling of missing / outlier values. Furthermore, area weights or distortion correction factors can be introduced for the hippocampal unfolded representation to reduce the spurious effects of geometric unfolding or local mesh density differences on the mapping results, thereby obtaining more stable and comparable vertex-level map features.

[0018] S4, Mechanism Assessment and Spatial Statistical Correction.

[0019] After obtaining the hippocampal surface reference map, its mechanistic association with the abnormal characterization of the subject's hippocampus is assessed. This abnormal characterization can be a hippocampal apex-level deviation metric (e.g., deviation value, abnormality probability, abnormality score, abnormality distribution map, etc. obtained from a normative model), or a difference / deviation map generated by other control methods. Mechanism assessment may include correlation analysis, regression analysis, similarity matching, weighted correlation, partition / axial summary statistics, etc., and further employs spatial statistical correction methods suitable for the hippocampal surface domain (e.g., permutation tests / rotational permutations / cyclic shifts / block permutations, etc., that preserve spatial autocorrelation structure or geometric constraints) to obtain corrected significance, confidence intervals, and credibility indices, reducing the risk of false positives caused by spatial autocorrelation.

[0020] S5, structured visual report generation and version tracking.

[0021] Based on the mapping results and mechanism evaluation results of S4, a structured visualization report is automatically generated. The report can include: external reference map name / version, mapping parameters and quality control results, correlation effect size and direction with the deviation map, significance and confidence after spatial correction, key region / axis summary, and visualization heatmap and charts. The reference map version, parameter configuration and output results are bound and archived to achieve auditable, reproducible and scalable deployment (local, server or containerized environment) engineering application.

[0022] In one embodiment, a hippocampal surface reference map mapping method is proposed, as described in processes S1 to S3 above. This method maps at least one external reference map to the hippocampal surface domain with fine-grained detail to obtain a hippocampal surface reference map characterized in units of mesh cells and / or vertices. Specifically, please refer to... Figure 1 Taking human brain mapping as an example, the mapping methods include: The process begins by obtaining a reference hippocampal surface mesh of a predefined template brain as input. This template brain can be a "standard brain" image derived from the average of numerous MRI scans of healthy adults, aligning each person's brain to the same standard space and facilitating data comparison and integration across subjects and studies. The template brain can be segmented using the HippUnfold tool or other hippocampal segmentation tools to obtain a reference hippocampal surface mesh of the standard hippocampus. This reference hippocampal surface mesh has a fixed topological structure and vertex sequence, serving as the basis for hippocampal surface domain mapping in external reference atlases.

[0023] At least one external reference map is acquired as input. The type of external reference map is not limited and can be selected according to the research direction, with the reference hippocampal surface mesh set to a triangular surface mesh. Further, after obtaining the external reference map, compliance checks and map standardization are performed, and metadata records are generated for the external reference map. The compliance check items include: file format (voxel / surface), data dimension (3D / 4D), data type (floating-point / integer), spatial information (affine matrix / voxel size / template space identifier), and numerical values ​​(NaN / Inf ratio, extreme value range, identification of all-zero or anomalous constants). This compliance check can be implemented through script rules, statistical summaries, or learned anomaly detection. The external reference map that passes the compliance check is standardized through spatial alignment preparation, resolution / scale unification, and intensity scale processing. Spatial alignment preparation ensures the integrity of the map's spatial information. Resolution / scale unification ensures that the voxel map can be recorded or resampled to a uniform voxel size. Intensity scaling is used to perform linear normalization, quantile truncation, z-score normalization, or rank normalization, and to solidify the parameters. Metadata records are generated for the normalized external reference map, preferably including the following fields: atlas_id (unique identifier), atlas_name, atlas_version, source, license; space (template space / version), resolution, modality (molecular / metabolic / genetic categories); preprocess (whether to smooth / normalize / truncate / handle missing data), checksum (checksum); date_added (entry date), notes (remarks).

[0024] Simultaneously, a hippocampal surface domain representation is established for each external reference map, and it can be stored according to the following structural organization: atlas_id / version / side(L / R) / surface_type(native / unfold) / density / data_vector.npy; atlas_id / version / qc_summary.json (coverage, outlier ratio, edge hints, etc.).

[0025] The hippocampal surface atlas library established in this manner can be used for batch calls in subsequent mechanism evaluations and supports version rollback and incremental additions. Through atlas standardization and metadata registration, new external reference atlases such as those for molecules, metabolism, energy, and gene expression can be continuously added without changing the main process, forming a scalable mechanism interpretation capability and avoiding the drawback of existing technologies requiring process rewriting when adding resources. For a selected external reference atlas, it needs to be registered or transformed to a standard space consistent with the preset template brain. Then, the external reference atlas is projected onto the reference hippocampal surface mesh through voxel-to-surface mapping and / or surface-to-surface mapping to obtain the atlas values ​​on the reference hippocampal surface mesh. t i Distribution. Specifically, when the external reference map is in voxel format, hippocampal surface mapping can be achieved using point sampling + kernel-weighted interpolation: First, calculate the first... in the reference hippocampal surface mesh. i vertices v i Sampling coordinates in the voxel space of the external reference map x i Then in x i The sampling point set Ω within the neighborhood i (Ω i Located within or under the constraints of the hippocampus ribbon mask, and using a kernel function. K (·) Weighted calculation of the firsti Graph values ​​of vertices t i :

[0026] In the formula, x Ω is the set of sampling points i The coordinates of the sampling points in the data. A ( x ) for external reference map in x The spectral values ​​at each vertex are obtained. The kernel function can be a trilinear kernel, a Gaussian kernel, or other interpolation / weighting methods. The neighborhood radius can be 1-3 voxels, and the kernel width can be 0.5-2.0 voxels. After traversing all vertices, the vertex-level spectral value distribution on the reference hippocampus surface mesh can be obtained.

[0027] On the other hand, hippocampal surface mapping can also be achieved using ribbon sampling, which is a multi-point averaging method along the normal: First define the vertices v i normal direction n i And the boundaries of the inner and outer surfaces (or gray banded areas). Then in x i + s m n i Upsampling M A point (e.g., can be taken) M =3~11), and the average value was obtained by averaging within the hippocampal ribbon mask. t i :

[0028] In the formula, A (·) represents the map value of the external reference map at the corresponding sampling point coordinates. s m For the first m Each sampling point along the normal direction n i The offset distance. s in This represents the inner sampling boundary. s out The outer sampling boundary is defined. When a sampling point falls outside the mask, it can be skipped, truncated to the boundary, or filled with the nearest valid point, and the missing proportion is recorded. After traversing all vertices, the vertex-level spectral value distribution on the reference hippocampus surface mesh is obtained. By using kernel interpolation, mask-constrained sampling, or normal multi-point sampling, the instability caused by boundary mixing and interpolation differences is reduced, making the mapping results more consistent and reliable.

[0029] When the external reference map is in a standard surface format, first convert the surface mesh of the external reference map to a standard space consistent with the reference hippocampus surface mesh, and then target the first... i vertices v i On the grid of the external reference map, determine the nearest point or the triangular facet corresponding to it, and obtain it using nearest neighbor interpolation or barycenter coordinate interpolation. v i spectral values t i This allows us to obtain the vertex-level spectral value distribution on the reference hippocampal surface mesh. For example... Figure 2 As shown, after spatial normalization and surface mapping, different external reference maps can all form vertex-level or patch-level map value distributions on the reference hippocampus surface grid. Figure 2 Different colors are used to represent the relative distribution intensity or standardized map values ​​of external reference maps at different locations on the hippocampal surface. Color differences are used to indicate the relative intensity of the same map across different hippocampal regions, rather than being limited to fixed absolute value thresholds. Here, *receptor* represents receptor-related maps, and *5-HT1A* represents the serotonin 1A receptor map. *mitochondria* represents mitochondrial-related maps, and CI, CII, and CIV represent mitochondrial respiratory chain complexes I, II, and IV-related maps, respectively. *MRC* represents the mitochondrial respiratory chain-related comprehensive map. *transporter* represents transporter-related maps. *DAT* represents the dopamine transporter map, and *NET* represents the norepinephrine transporter map. *cognition* represents cognitive-related maps, and *Cognitive PC1* represents the first principal component map obtained from principal component analysis of cognitive-related indicators. Figure 2 It can be seen that the present invention is not limited to a single molecule or a single modality map, but can project multiple external reference maps such as receptors, transporters, mitochondrial function, and cognitive-related spatial patterns onto the hippocampal surface domain, so that different maps can be displayed, compared and subsequently evaluated under the same hippocampal surface vertex sequence.

[0030] Furthermore, mapping quality control (QC) and anomaly handling are performed on the calculated vertex-level graph value distribution. The mapping quality control metrics include: coverage (proportion of valid vertices); outlier proportion; edge artifact hints (proportion of hippocampal endpoints / boundary regions); and missing value handling records. When mapping quality control fails, the vertex-level graph value distribution can be marked as "unreliable" for removal and remapping can be triggered (by changing the sampling kernel / mask / alignment method of the kernel function). The QC metrics can automatically identify and handle interpolation instability issues in endpoints / boundary regions, avoiding erroneous associations caused by endpoint artifacts in existing technologies. Moreover, the QC metrics can be written into the final output results, reports, or accompanying files, providing auditable evidence for each mechanism interpretation, which is a significant advantage compared to solutions that "only provide conclusions without explaining mapping quality."

[0031] The subject hippocampal surface mesh is obtained as input three. After processing, this subject hippocampal surface mesh has a consistent topological structure and vertex correspondence with the reference hippocampal surface mesh in the same standard space. Specifically, the method for establishing the vertex correspondence between the subject hippocampal surface mesh and the reference hippocampal surface mesh includes: unfolding the subject hippocampal surface mesh and the reference hippocampal surface mesh from folded surfaces to unfolded surfaces. A one-to-one correspondence is established on the two unfolded surfaces based on unfolding parameter coordinates / unfolded mesh index, and resampling or deformation alignment is performed to ensure that the two have a consistent vertex sequence. After alignment, it is converted back into a folded surface to obtain the mapped subject hippocampal surface mesh and the reference hippocampal surface mesh. To reduce distortion during the folding to unfolding process, area weights for vertices or faces can be introduced. w i For spectral values t i Perform correction: .

[0032] .

[0033] In the formula, Indicates the corrected spectral values t i . f This indicates multiplicative correction, logarithmic field correction, or normalization correction. Indicates the folded surface at the 1st i The neighborhood area of ​​each vertex. Indicates the unfolded surface at the 1st iThe neighborhood area of ​​each vertex. This area-weighted correction reduces the systematic impact of geometric distortion on effect size and significance, improves comparability across individuals and versions, and reduces the driving force of local geometric instability on overall mechanism conclusions, thereby improving the robustness of the results. This makes distortion correction and area weighting more consistent with the actual anatomical area and spatial structure, avoiding the risk of "geometric changes being misjudged as biological changes".

[0034] Based on the vertex correspondence between the subject's hippocampal surface grid and the reference hippocampal surface grid, the distribution of map values ​​on the reference hippocampal surface grid is correspondingly transferred to the subject's hippocampal surface grid, thus obtaining a fine-grained hippocampal surface reference map as output. The grid shape in both the subject's hippocampal surface grid and the hippocampal surface reference map can preferably be triangular. Based on this hippocampal surface reference map, the hippocampal deviation map and correlation analysis of a subject with a certain disease relative to the standard population can be analyzed. For example, taking an Alzheimer's disease patient and using a neurotransmitter (serotonin) reference map as an external reference map, the patient's left hippocampus is divided into 2048 patches in the hippocampal surface reference map. Through the subject's hippocampal deviation map, it can be determined how much each position of the hippocampus deviates from that of a normal person. Through the distribution of map values ​​for each patch, the characteristics of the corresponding external reference map for each position can be determined. It's not just about knowing "this patient's hippocampus is smaller," but about knowing "which parts of the hippocampus are more abnormal, which parts are relatively preserved, and how the abnormalities are distributed." Simultaneously, because an external reference map is mapped, each of the 2048 locations represents what biological characteristics this location is biologically predominantly aligned with. In other words, are the areas with the most severe abnormalities in the patient also areas with more pronounced serotonin characteristics? If so, it indicates that the hippocampal damage in the patient is not random but tends to concentrate in regions with a specific biological background. This hippocampal surface reference map thus provides a "mechanistic clue": some areas in the hippocampus may be inherently more vulnerable, or, when disease occurs, these areas with specific biological properties are more likely to be affected first. Conversely, if it is found that the most severe hippocampal atrophy in the patient is not in areas with high serotonin characteristics, or even significantly avoids these areas, this result is equally significant, indicating that the serotonin-related external reference map may not adequately explain the abnormal distribution in this patient. The direction of hippocampal damage in the patient may not primarily follow the serotonin line. Therefore, other types of "biological maps," i.e., external reference maps, such as dopamine, acetylcholine, energy metabolism, and mitochondrial-related external reference maps, can be used to see if these other types of maps correspond to the patient's abnormalities. This leads to a more comprehensive picture, rather than a simplistic conclusion, that is, a "biological match report card" for the patient: how closely serotonin and patient abnormalities are correlated, how closely dopamine is correlated, how closely acetylcholine is correlated, and how closely energy metabolism is correlated. Ultimately, it was found that the patient's hippocampal abnormalities had a moderate relationship with serotonin, but were more closely correlated with acetylcholine and energy metabolism patterns. Therefore, it can be more reasonably concluded that the patient's hippocampal damage pattern is more likely to occur in areas related to acetylcholine or high energy demand, rather than in dopamine-related areas.

[0035] In another embodiment, a mechanism evaluation method based on a hippocampal surface reference map is proposed. For example... Figure 3As shown, firstly, a hippocampal surface reference map is obtained using the hippocampal surface reference map mapping method described in the above embodiments. Next, the subject's hippocampal deviation map is acquired (the deviation map can be obtained through norm analysis, case-control difference analysis, follow-up change analysis, or other methods) and aligned with the vertex sequence of the grid in the hippocampal surface reference map. Subsequently, the correlation effect size between the subject's hippocampal deviation map and the hippocampal surface reference map is calculated, yielding the observed statistics. A constraint permutation that preserves the spatial structure is constructed in the hippocampal surface domain, generating a permutation null distribution, and the corrected significance result is obtained accordingly. This corrected significance result and / or observed statistics are output for mechanism evaluation. The result elements of this mechanism evaluation include: a list of most relevant maps, effect size, direction, CI (confidence interval), results and differences on the left and right sides respectively, axial summary (anterior / middle / posterior segments) or partition summary, confidence level: overall mapping coverage, distortion ratio, permutation stability, etc. Specifically, the observational statistics here are not generated independently, but are calculated by comparing the "subject's hippocampal deviation map" and the "hippocampal surface reference map" point by point on the same set of vertices. For example, the left hippocampal surface has 2048 vertices, each with two values: one is the subject's deviation at that location, and the other is the reference map value of a certain external reference map at that location. Correlation analysis is performed on these 2048 pairs of values. If the calculated Pearson correlation coefficient r = 0.36, then r = 0.36 can be used as the observational statistics for this mechanism evaluation. To determine the reliability of this correlation result, further calculation of the corrected significance is needed. Specifically, a permutation test is performed while maintaining the spatial structure of the hippocampal surface. For example, cyclic displacement along the major axis, surface block permutation, or permutation methods based on grid distance constraints are used to generate 10,000 random permutation results. After each permutation, a new correlation coefficient is calculated to form a null distribution. If only 128 out of 10,000 random results have a correlation strength greater than or equal to the true observed value of 0.36, then the corrected significance can be calculated as p = (128 + 1) / (10,000 + 1) = 0.0129. If multiple external reference maps are compared simultaneously, the FDR method can be further used for multiple comparison correction to obtain the final corrected significance result. In terms of implementation tools, NumPy, SciPy, and Statsmodels in Python can be used for correlation, regression, and FDR correction, combined with a custom hippocampal surface permutation script, or spatial constraint permutation can be completed based on the hippocampal surface vertex sequence and mesh structure output by HippUnfold. Therefore, in this embodiment, the observation statistics come from the correlation calculation between the true deviation map and the reference map. The corrected significance result comes from the permutation test that preserves the spatial structure and multiple comparison correction.

[0036] For any subject's brain scan, the hippocampal mesh can be obtained by segmenting the brain using the HippUnfold tool or other hippocampal segmentation tools. After processing with norm analysis, case-control analysis, or other anomaly detection methods, the subject's hippocampal deviation map is obtained. Since both the subject's hippocampal deviation map and the hippocampal surface reference map are registered to a standard space consistent with the template brain and have a consistent vertex sequence, cross-subject and same-vertex comparisons are possible. Finally, a structured report can be automatically generated based on the hippocampal surface reference map and mechanism assessment results. This report includes: subject information (ID, date, left / right side), external reference map information (version, source, space), mapping parameters and QC summary, mechanism assessment results, visualization images (deviation heatmap + most relevant atlas heatmap + axial summary / comparison image), confidence level and alerts (e.g., insufficient coverage, excessive distortion, unstable permutation, etc.). The report can be output in PDF, HTML, image (PNG / SVG) or structured JSON formats, and can simultaneously output a "machine-readable results file + human-readable report".

[0037] In another embodiment, a computer-readable storage medium is also proposed, which stores a computer program / instructions. When the computer program / instructions are executed by a processor, the steps of the hippocampal surface reference map mapping method and / or the mechanism evaluation method based on the hippocampal surface reference map described in the above embodiments are implemented. The computer-readable storage medium may include, but is not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0038] In another embodiment, a computer program product is also proposed, comprising a computer program / instructions. The computer program / instructions are used to cause a computer to perform the steps of the hippocampal surface reference map mapping method described in the above embodiments. The computer program / instructions may exist in a computer-readable medium in forms including, but not limited to, source files, executable files, and installation package files. Accordingly, the computer program instructions may be executed by a computer in ways including, but not limited to: the computer directly executing the instructions; the computer compiling the instructions and then executing the corresponding compiled program; the computer reading and executing the instructions; or the computer reading and installing the instructions and then executing the corresponding installed program.

[0039] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0040] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for mapping a reference map of the hippocampus surface, characterized in that, It is used to map at least one external reference map to the hippocampal surface domain in a fine-grained manner to obtain a hippocampal surface reference map characterized in units of grid cells and / or vertices; Mapping methods include: Obtain a reference hippocampal surface mesh of a preset template brain, wherein the reference hippocampal surface mesh has a fixed topological structure and vertex sequence; Obtain at least one external reference map; after registering or converting the external reference map to a standard space consistent with the preset template brain, project the external reference map onto the reference hippocampus surface grid through voxel-to-surface mapping and / or surface-to-surface mapping to obtain the map value distribution on the reference hippocampus surface grid. Obtain the hippocampal surface mesh of the subject and ensure that the subject's hippocampal surface mesh and the reference hippocampal surface mesh have the same topological structure and vertex correspondence; based on the vertex correspondence, transfer the map value distribution on the reference hippocampal surface mesh to the subject's hippocampal surface mesh to obtain the hippocampal body surface reference map.

2. The hippocampal surface reference map mapping method according to claim 1, characterized in that, External reference maps include: neurotransmitter / receptor / transporter related reference maps, mitochondrial function or energy metabolism related maps, gene expression spatial maps, myelin / iron deposition / perfusion / microstructure proxy maps, pathological protein or tracer distribution maps, functional connectivity, network gradient type maps or cognitive related spatial maps.

3. The hippocampal surface reference map mapping method according to claim 1, characterized in that, The acquired external reference maps undergo compliance checks and map standardization processing, and metadata records are generated for the external reference maps. The compliance check items include: file format, data dimensions, data types, spatial information, and numerical range; Atlas standardization processes include: spatial alignment preparation, resolution / scale unification, and intensity scale processing.

4. The hippocampal surface reference map mapping method according to claim 1, characterized in that, The external reference map is in voxel format and / or surface format; the subject hippocampal surface mesh and the reference hippocampal surface mesh are triangular surface meshes; the hippocampal surface reference map records map values ​​in units of vertices and / or faces of the triangular surface mesh.

5. The hippocampal surface reference map mapping method according to claim 1, characterized in that, When the external reference map is in voxel format, the methods for projecting the external reference map onto the reference hippocampus surface mesh include: Calculate the first reference hippocampal surface mesh i vertices v i Sampling coordinates in the reference map voxel space x i ; exist x i The sampling point set Ω within the neighborhood i and using kernel functions K (·) Weighted calculation of the first i Graph values ​​of vertices t i : ； In the formula, x Ω is the set of sampling points i Sampling points in; A ( x ) for external reference map in x The spectral values ​​at each vertex are obtained; after traversing all vertices, the vertex-level spectral value distribution on the reference hippocampus surface mesh is obtained; Alternatively, define vertices. v i normal direction n i and the inner and outer surface boundaries; in x i + s m n i Upsampling M The average value was obtained from several points within the hippocampal ribbon mask. t i : ； In the formula, A (·) represents the map value of the external reference map at the corresponding sampling point coordinates. s m For the first m Each sampling point along the normal direction n i offset distance, s in The inner sampling boundary, s out The outer sampling boundary; After traversing all vertices, the vertex-level spectral value distribution on the reference hippocampus surface mesh is obtained.

6. The hippocampal surface reference map mapping method according to claim 1, characterized in that, When the external reference map is in a standard surface format, the method for mapping the external reference map to obtain a reference hippocampus surface mesh includes: First, convert the surface mesh of the external reference map to a standard space consistent with the reference hippocampus surface mesh; then, for the first... i vertices v i On the grid of the external reference map, determine the nearest point or the triangular facet corresponding to it, and obtain it using nearest neighbor interpolation or barycenter coordinate interpolation. v i spectral values t i This allows us to obtain the vertex-level spectral value distribution on the reference hippocampal surface mesh.

7. The hippocampal surface reference map mapping method according to claim 1, characterized in that, Methods for establishing vertex correspondences between the subject's hippocampal surface mesh and the reference hippocampal surface mesh include: The subject's hippocampal surface mesh and the reference hippocampal surface mesh were unfolded from folded surfaces to unfolded surfaces; A one-to-one correspondence is established on the two unfolded surfaces based on the unfolded parameter coordinates / unfolded mesh index, and resampling or deformation alignment is performed to make the two have a consistent vertex sequence; after alignment, it is converted back into a folded surface to obtain the aligned subject hippocampal surface mesh and the reference hippocampal surface mesh.

8. A mechanism evaluation method based on hippocampal surface reference maps, characterized in that, It includes: First, a hippocampal surface reference map is obtained using the hippocampal surface reference map mapping method as described in any one of claims 1 to 7; then, the subject's hippocampal deviation map is obtained and aligned with the vertex sequence of the grid in the hippocampal surface reference map. The correlation effect size between the subject's hippocampal deviation map and the hippocampal surface reference map is calculated to obtain the observational statistics; the correlation effect size includes the correlation coefficient, regression coefficient, and similarity index. A constrained permutation that preserves the spatial structure is constructed in the hippocampal surface domain, generating a permutation null distribution, and a corrected significance result is obtained accordingly. The corrected significance result and / or observation statistics are output for mechanism evaluation.

9. The mechanism evaluation method based on hippocampal surface reference maps according to claim 8, characterized in that, Based on the mechanism evaluation results, a structured and visualized report is automatically generated. The report includes: the name / version of the external reference map, mapping parameters and quality control results, the effect size and direction of the correlation with the deviation map, the significance and confidence after spatial correction, key region / axis summary, and visualization heatmaps and charts.

10. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the hippocampal surface reference map mapping method as described in any one of claims 1 to 7.

Images