Method and apparatus for analyzing effects of olfactory loss on cerebral cortical structure

Abstract

The method and device for analyzing the influence of anosmia on the structure of cerebral cortex, by integrating sMRI data, image transcriptomics analysis and Mendelian randomization method, systematically explore the influence of anosmia on the structure of cerebral cortex. The method includes: (1) using 3.0T magnetic resonance scanning to obtain high-resolution structural magnetic resonance imaging of anosmia patients, and using FreeSurfer software to extract the cerebral cortical surface area and cortical thickness; (2) based on the FinnGen database, the genetic variation related to anosmia is obtained, combined with the genetic correlation data of the cerebral cortical surface area and cortical thickness of the ENIGMA database, the causal effect of anosmia on the cortical structure is evaluated by Mendelian randomization analysis; (3) based on the Allen Human Brain Atlas, the genes related to the changes of cerebral cortical surface area and cortical thickness are identified by using partial least squares regression, and the molecular mechanism is revealed by enrichment analysis and cell type specific analysis.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and more particularly to an analytical method and apparatus for analyzing the impact of loss of smell on the structure of the cerebral cortex. Background Technology

[0002] Anosmia is a common clinical symptom that can be caused by a variety of factors, including viral infections, head trauma, and neurodegenerative diseases. Anosmia not only significantly impacts patients' quality of life but may also trigger changes in brain structure. High-resolution structural magnetic resonance imaging (sMRI) can precisely capture morphological features of brain structures, such as gray matter volume (GMV), cortical thickness (TH), and surface area (SA), providing crucial data support for studying the impact of anosmia on brain structure. Studies have shown that patients with congenital anosmia have reduced GMV in the gyrus rectus and increased GMV in the orbitofrontal cortex. COVID-19-related anosmia may involve alterations in multiple genes and their related molecular pathways. Although existing research has revealed potential links between anosmia and brain structure, systematic studies on the effects of anosmia on cerebral cortex structure and its molecular mechanisms remain limited.

[0003] With the rapid development of neuroimaging and genomics technologies, imaging transcriptomics and Mendelian randomization (MR) analysis have provided new tools for studying the impact of olfactory loss on cerebral cortical structure. Imaging transcriptomics analysis, by integrating neuroimaging data with gene expression data, can reveal the association between brain structure and gene expression, thus providing new clues for understanding the molecular mechanisms of olfactory loss. This method can identify regional genes associated with altered spatial expression characteristics of neuroimaging biomarkers, providing a new perspective on the mechanisms of disease-related imaging biological changes, from microscopic molecular function to macroscopic brain functional structural abnormalities. Current research increasingly needs to further explore the spatial association between disease neuroimaging representations and brain gene expression to gain a deeper understanding of the genetic mechanisms of imaging biomarkers.

[0004] Mendelian randomization (MR) is a causal inference method based on genetic variation. It effectively avoids confounding factors and reverse causality problems in traditional observational studies, providing a powerful tool for exploring the causal relationship between olfactory loss and cerebral cortical structure. By selecting single nucleotide polymorphisms (SNPs) significantly associated with olfactory loss as instrumental variables, MR can assess the causal impact of olfactory loss on cortical surface area (SA) and cortical thickness (TH). In recent years, genome-wide association studies (GWAS) have identified a novel significant gene locus (rs11228623) associated with olfactory dysfunction, and this region is rich in olfactory receptor genes. Furthermore, combining GWAS data and expression quantitative trait loci (eQTL) data, summary-data-based Mendelian randomization (SMR) has achieved significant results in COVID-19 research, identifying 38 new genes associated with severe COVID-19. Based on the SMR method, risk genes associated with loss of smell can be further identified, and the role of these genes in changes in cortical structure can be explored. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, the technical problem to be solved by this invention is to provide an analytical method for the impact of loss of smell on the structure of the cerebral cortex. This method systematically explores the impact of loss of smell on the structure of the cerebral cortex by integrating sMRI data, image transcriptomics analysis and Mendelian randomization.

[0006] The technical solution of this invention is: an analytical method for the impact of loss of smell on the structure of the cerebral cortex, comprising the following steps:

[0007] (1) High-resolution structural magnetic resonance imaging of patients with loss of smell was obtained by 3.0T magnetic resonance scanning, and the surface area and thickness of the cerebral cortex were extracted by FreeSurfer software.

[0008] (2) Genetic variations related to loss of smell were obtained based on the FinnGen database, combined with

[0009] Genetic association data on cortical surface area and cortical thickness from the ENIGMA database were used to assess the causal effects of loss of smell on cortical structure using Mendelian randomization analysis.

[0010] (3) Based on the Allen brain map, partial least squares regression was used to identify genes related to changes in cerebral cortex surface area and cortex thickness, and the molecular mechanisms were revealed through enrichment analysis and cell type specific analysis.

[0011] This invention systematically explores the impact of loss of smell on the structure of the cerebral cortex by integrating sMRI data, image transcriptomics analysis and Mendelian randomization methods.

[0012] An analytical device for analyzing the impact of loss of smell on the structure of the cerebral cortex is also provided, comprising:

[0013] The extraction module is configured to acquire high-resolution structural magnetic resonance imaging of patients with loss of smell using 3.0T magnetic resonance scanning and to extract the surface area and thickness of the cerebral cortex using FreeSurfer software.

[0014] The assessment module is configured to obtain olfactory loss-related genetic variations based on the FinnGen database, combine them with genetic association data on cortical surface area and cortical thickness from the ENIGMA database, and use Mendelian randomization analysis to assess the causal effects of olfactory loss on cortical structure.

[0015] The analysis module, configured based on the Allen brain atlas, uses partial least squares regression to identify genes associated with changes in cerebral cortical surface area and cortical thickness, and reveals their molecular mechanisms through enrichment analysis and cell type-specific analysis. Attached Figure Description

[0016] Figure 1 A flowchart of the analytical method for the effect of loss of smell on the structure of the cerebral cortex according to the present invention is shown. Detailed Implementation

[0017] like Figure 1 As shown, this method for analyzing the impact of olfactory loss on the structure of the cerebral cortex includes the following steps:

[0018] (1) High-resolution structural magnetic resonance imaging of patients with loss of smell was obtained by 3.0T magnetic resonance scanning, and the surface area and thickness of the cerebral cortex were extracted by FreeSurfer software.

[0019] (2) Genetic variations related to loss of smell were obtained from the FinnGen database. Combined with the genetic association data of cortical surface area and cortical thickness from the ENIGMA database, Mendelian randomization analysis was used to evaluate the causal effects of loss of smell on cortical structure.

[0020] (3) Based on the Allen brain map, partial least squares regression was used to identify genes related to changes in cerebral cortex surface area and cortex thickness, and the molecular mechanisms were revealed through enrichment analysis and cell type specific analysis.

[0021] This invention systematically explores the impact of loss of smell on the structure of the cerebral cortex by integrating sMRI data, image transcriptomics analysis and Mendelian randomization methods.

[0022] Preferably, in step (2), the 3D-T1WI sequence is standardized and post-processed using FreeSurfer software, including: head motion correction, volume segmentation, spherical expansion, surface reconstruction, spherical registration, smoothing, visual inspection, and manual correction; the cortex is partitioned using DKT maps to generate regions of interest (ROIs), with each hemisphere containing 34 cortical ROIs, and SA and TH data of the 34 left hemisphere ROIs are extracted; age and gender are covariate-corrected using a general linear model, with the significance level set at P<0.05.

[0023] Preferably, in step (2), the Mendelian randomization analysis method uses SA and TH as outcome variables, selects single nucleotide polymorphisms (SNPs) that are significantly associated with loss of smell as instrumental variables, and analyzes the causal relationship between exposure factors and brain structural characteristics; evaluates the effectiveness of instrumental variables through heterogeneity test, uses sensitivity analysis to exclude the influence of potential confounding factors, and uses pleiotropy analysis to detect whether there is horizontal pleiotropy bias.

[0024] Preferably, in step (2), to identify genetic variations significantly associated with the exposure factor, the genome-wide significance threshold is set to P < 5 × 10⁻⁶. -8 To eliminate the effects of cascading imbalances, set r 2 The screening criteria included SNPs with a value <0.001 and a physical distance greater than 10 kb; association information related only to the exposure factor was extracted from the summary data of the outcome variables, and incompatible SNPs and genetic variations with palindromic structures were excluded during data matching; SNPs with incorrect causal association directions were removed using MR-Steiger filtering; the F-value of each SNP was calculated, and the F-value was estimated using the formula F = β* / SE. 2 The F-value measures the strength of the instrumental variable in explaining the risk factors of exposure. F < 10 indicates a weak instrumental variable, which is then removed.

[0025] Preferably, in step (2), in Mendelian randomization analysis, the inverse variance weighted method is used as the core analysis method. By calculating the effect value of each genetic variation and integrating it with inverse variance weight, the overall effect estimate is obtained. This method assumes that all instrumental variables are effective instruments. Multiple sensitivity analysis methods are used: First, the effect of a single SNP on the overall result is evaluated by leaving one out; second, the potential bias and heterogeneity of the instrumental variables are detected by funnel plot; finally, the effect size of each SNP on the cerebral cortex structure and olfactory dysfunction is displayed intuitively by forest plot; the heterogeneity analysis is quantitatively evaluated by Cochran's Q test. When the P value of the Q statistic is less than 0.05, it indicates that there is significant heterogeneity. At this time, the random effects IVW model is used for re-estimation; in order to detect level pleiotropy, both MR-PRESSO and MR-Egger methods are applied. When the detection result P < 0.05, the robustness of the result is re-evaluated after removing abnormal SNPs; the statistical significance level is set to P < 0.05.

[0026] Preferably, in step (3), brain region gene microarray expression data from the Allen Human Brain Atlas (AHBA) are obtained, and the AHBA is processed using the Abagen toolbox. The transcription data is mapped to 83 brain regions of the whole brain based on the DKT map. Gene probes are re-annotated, and probes whose expression intensity is lower than the background noise level in more than 50% of the samples are removed. The tissue samples are spatially located using the corrected MNI coordinates, and mapped to the nearest brain region with a radius of 2 mm. To reduce the misassignment rate, the sample-to-brain-region matching process follows a large-scale... Anatomical constraints on hemispheric division and cortical / subcortical partitioning; for samples that were not matched to a specific brain region through the above standard procedure, the nearest centroid method was used for supplementary allocation to ensure that all brain regions obtained the corresponding gene expression data; when multiple samples were assigned to the same brain region, the average expression value of each donor was calculated separately, the robust sigmoid function was applied to normalize the brain region expression profile of each donor, and it was transformed into a unit interval. The expression data of all donors were integrated to construct a gene expression matrix with brain regions as rows and 15,633 genes as columns.

[0027] Preferably, in step (3), PLS regression analysis is used to explore the relationship between the transcriptome of healthy brain and changes in cortical surface area and thickness in olfactory dysfunction. The PLS algorithm uses gene expression as the predictor variable and changes in SA and TH as the response variables. The first PLS component, PLS1, represents the maximum covariance between gene expression patterns and changes in SA / TH. The relative explained variance of 1-15 components is calculated, and significance is evaluated using a 1000-times permutation test. The standard error of gene PLS1 weights is estimated using bootstrap sampling. Z-scores are calculated to rank gene contributions. Replacement sampling is performed on 34 brain regions. With a 5% FDR threshold, genes are divided into genes with positive weight PLS+ and negative weight PLS- according to the positive or negative Z-score, thus obtaining a list of genes related to changes in SA / TH.

[0028] Preferably, in step (3), Mendelian randomization based on aggregated data is used as an extension of MR, integrating GWAS data and gene expression data to provide conditions for identifying the causal effects of gene expression; the heterogeneity-dependent tool test is used to distinguish between true causal relationships and genomic linkage disequilibrium effects; by integrating genome-wide association study data and quantitative trait locus data of olfactory dysfunction, the SMR method is used to identify risk genes associated with olfactory dysfunction; in the SMR analysis, SNPs associated with olfactory loss are used as instrumental variables to study the causal effects of exposure factors on the outcome, using default parameters and P SMR <0.05 was set as the significance level; to test the heterogeneity of IVs, a HEIDI test was performed to filter out P-values. HEIDI To explore the relationship between genes related to loss of smell and genes related to SA and TH in SMR analysis, a list of genes related to loss of smell and overlapping genes with SA and TH were obtained, and the expression association of these overlapping genes was then analyzed.

[0029] Preferably, in step (3), a systems biology analysis is performed on significant genes among SA-related genes and TH-related genes; gene ontology functional annotation is performed on the PLS1 negatively weighted genes using the clusterProfiler software package; pathway enrichment analysis is performed on the screened significantly differentially expressed genes using the Metascape online platform; key signaling pathways are visualized using the QuickGO tool; SA-related genes and TH-related genes are submitted to the Synapse gene ontology database SynGO for functional annotation, wherein cellular component information is visualized using a sunburst plot; based on STRING numbers... A protein-protein interaction network was constructed based on the database. Text mining sources were excluded during the analysis, and a high confidence threshold of 0.9 was used to screen for reliable interactions. To identify key regulatory nodes, the CytoHubba plugin of the Cytoscape platform was used for network topology analysis. Weighted cell type enrichment analysis was used to determine whether subsets of SA-related genes and TH-related genes were enriched in specific brain cell types. The results were validated using two independent human-derived cell type-specific gene datasets. First, single-cell transcription data from AHBA-seq were used, followed by repeated analysis using the DroNc-seq dataset to ensure that the results were not dependent on a single dataset.

[0030] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium. When executed, the program includes the steps of the methods of the above embodiments. The storage medium can be ROM / RAM, magnetic disk, optical disk, memory card, etc. Therefore, corresponding to the method of the present invention, the present invention also includes an analytical device for the impact of olfactory loss on the structure of the cerebral cortex. This device is typically represented in the form of functional modules corresponding to the steps of the method. The device includes:

[0031] The extraction module is configured to acquire high-resolution structural magnetic resonance imaging of patients with loss of smell using 3.0T magnetic resonance scanning and to extract the surface area and thickness of the cerebral cortex using FreeSurfer software.

[0032] The assessment module is configured to obtain olfactory loss-related genetic variations based on the FinnGen database, combine them with genetic association data on cortical surface area and cortical thickness from the ENIGMA database, and use Mendelian randomization analysis to assess the causal effects of olfactory loss on cortical structure.

[0033] The analysis module, configured based on the Allen brain atlas, uses partial least squares regression to identify genes associated with changes in cerebral cortical surface area and cortical thickness, and reveals their molecular mechanisms through enrichment analysis and cell type-specific analysis.

[0034] The verification results of the present invention are described in more detail below.

[0035] 1. Changes in SA and TH caused by anosmia and their MR verification

[0036] This study, using multimodal magnetic resonance imaging (MRI) data, revealed significant changes in cortical surface area and thickness in brain regions related to olfactory processing in patients with anosmia. Specifically, salivary osmosis (SA) was decreased in the left anterior cingulate cortex and superior temporal sulcus, while total thrombosis (TH) was significantly increased in the left cuneus, supramarginal gyrus, precentral gyrus, paracentral gyrus, anterior cingulate cortex, precuneus, superior frontal gyrus, caudal part of the middle frontal gyrus, and superior parietal lobule. These brain regions play crucial roles in the integration, processing, and transmission of olfactory information, and changes in their structure and morphology may be closely related to the loss of olfactory function. The increase in TH may be a compensatory mechanism to cope with the lack of olfactory information; for example, olfactory experts such as sommeliers and perfumers, through professional training, exhibit a significant increase in gray matter volume in the right insula and entorhinal cortex. To verify the reliability of these findings, this study further employed Mendelian randomization to analyze the causal relationship between anosmia and cortical structural features. MR analysis revealed a significant causal association between loss of smell and 11 structural features of the cerebral cortex, such as the thickness of the superior frontal gyrus, the thickness of the caudal part of the middle frontal gyrus, and the surface area of ​​the orbital gyrus. These results not only support the findings based on imaging data but also further suggest that loss of smell may directly lead to structural and morphological changes in these brain regions. Through MR analysis, this study provides causal evidence for brain region changes caused by loss of smell, enhancing the credibility of the research results.

[0037] 2. SA and TH-related genes revealed by PLSR and SMR

[0038] This study investigated the molecular mechanisms of altered salivary syndrome (SA) and neurotransmitter (TH) in patients with anosmia using PLS and SMR methods. PLS analysis revealed a set of genes associated with SA and TH changes, with 3460 genes associated with SA changes and 1888 genes associated with TH changes. For example, overexpression of genes such as GTF3A and SYT12 was positively correlated with increased SA and TH, respectively, while low expression of genes such as KLHL25 and IGSF22 was negatively correlated with decreased SA and TH, respectively. GTF3A, as a transcription factor of RNA polymerase III, regulates ribosomal RNA synthesis, and its overexpression may function by promoting neuronal RNA metabolism or enhancing the translation of synaptic-associated proteins. SYT12, a member of the synaptic binding protein family, participates in regulating synaptic vesicle release and neurotransmitter transmission, playing an important role in synaptic function and neuronal plasticity. IGSF22, along with other genes (such as SERPINF1, PRR6, TRPA1, BOC, and JUN), was found to be associated with neuronal projection development. These genes are involved in key biological processes such as neuronal development, synaptic plasticity, and cellular metabolism, providing a molecular-level explanation for the structural and functional changes in brain regions after olfactory loss. Further combined with SMR analysis, this study identified 176 risk genes associated with olfactory loss, of which 14 genes overlapped with SAPLS1 background genes, and 7 genes overlapped with THPLS1 background genes. For example, genes such as NUP85, P2RX7, and PNPLA77 showed a significant negative correlation with SA. P2RX7 is a purinergic ATP-sensitive calcium channel widely expressed in neurons and other cell types, triggering physiological responses such as inflammation, immune responses, and neurotransmission by sensing extracellular ATP activation. The results of PLSR and SMR corroborate each other, jointly revealing that olfactory loss may lead to compensatory changes in the structure and function of related brain regions by affecting neuronal and synaptic plasticity.

[0039] 3. Analysis of gene pathways, synaptic function, and cell types

[0040] To further understand the function of genes associated with changes in SA and TH, this study performed gene set enrichment analysis (GSEA), synaptic functional annotation, and cell type analysis. GSEA results showed that SA and TH-related genes were significantly enriched in synaptic pathways, such as the postsynaptic membrane, synaptic membrane, and regulation of chemosynaptic transmission. These pathways play crucial roles in information transmission between neurons and synaptic plasticity; their dysfunction may lead to structural and functional changes in olfactory-related brain regions. Synaptic functional annotation (SynGO) further revealed the important roles of SA and TH-related genes in synapse formation and function. For example, 155 SA-related genes and 141 TH-related genes are involved in synapse formation, with many genes significantly enriched in the presynaptic, postsynaptic, and synaptic membranes. These genes may affect the efficiency of information transmission in olfactory-related brain regions by regulating synaptic structure and function. Cell type analysis showed that SA and TH-related genes were significantly enriched in specific neuronal types. For example, SA-related genes are significantly enriched in GABAergic and glutamatergic neurons, while TH-related genes are significantly enriched in glutamatergic neurons in the prefrontal cortex and pyramidal neurons in the CA region of the hippocampus. These neuronal types play an important role in the processing and integration of olfactory information, and their dysfunction may lead to changes in brain regions in patients with anosmia.

[0041] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for analyzing the impact of loss of smell on the structure of the cerebral cortex, characterized by: It includes the following steps: (1) High-resolution structural magnetic resonance imaging of patients with loss of smell was obtained by 3.0T magnetic resonance scanning, and the surface area and thickness of the cerebral cortex were extracted by FreeSurfer software. (2) Genetic variations related to loss of smell were obtained from the FinnGen database. Combined with the genetic association data of cortical surface area and cortical thickness from the ENIGMA database, Mendelian randomization analysis was used to evaluate the causal effects of loss of smell on cortical structure. (3) Based on the Allen brain map, partial least squares regression was used to identify genes related to changes in cerebral cortex surface area and cortex thickness, and the molecular mechanisms were revealed through enrichment analysis and cell type specific analysis.

2. The analytical method for the impact of loss of smell on the structure of the cerebral cortex according to claim 1, characterized in that: In step (2), the 3D-T1WI sequence is standardized and post-processed using FreeSurfer software, including: head motion correction, volume segmentation, spherical expansion, surface reconstruction, spherical registration, smoothing, visual inspection, and manual correction; the cortex is partitioned using DKT maps to generate regions of interest (ROIs), with each hemisphere containing 34 cortical ROIs, and SA and TH data of the 34 left hemisphere ROIs are extracted; age and gender are covariate-corrected using a general linear model, with the significance level set at P<0.

05.

3. The analytical method for the impact of loss of smell on the structure of the cerebral cortex according to claim 2, characterized in that: In step (2), the Mendelian randomization analysis method uses SA and TH as outcome variables, selects single nucleotide polymorphisms (SNPs) that are significantly associated with loss of smell as instrumental variables, and analyzes the causal relationship between exposure factors and brain structural characteristics. The effectiveness of the instrumental variables is evaluated by heterogeneity test, the influence of potential confounding factors is excluded by sensitivity analysis, and pleiotropic analysis is used to detect whether there is horizontal pleiotropic bias.

4. The analytical method for the impact of loss of smell on the structure of the cerebral cortex according to claim 3, characterized in that: In step (2), to identify genetic variations significantly associated with the exposure factor, the genome-wide significance threshold is set to P < 5 × 10⁻⁶. -8 To eliminate the effects of cascading imbalances, set r 2 Screening criteria: <0.001 and physical distance greater than 10kb; extract association information related only to the exposure factor from the summary data of the outcome variable, and exclude incompatible SNPs and genetic variations with palindromic structures during data matching. through MR-Steiger filtering removes SNPs with incorrect causal relationships. Calculate the F-value for each SNP. The formula for estimating the F-value is F = β* / SE. 2 The F-value measures the strength of the instrumental variable in explaining the risk factors of exposure. F < 10 indicates a weak instrumental variable, which is then removed.

5. The analytical method for the effect of loss of smell on the structure of the cerebral cortex according to claim 4, characterized in that: In step (2), in Mendelian randomization analysis, the inverse variance weighted method is used as the core analysis method. By calculating the effect value of each genetic variant and integrating it with inverse variance weight, the overall effect estimate is obtained. This method assumes that all instrumental variables are effective instruments. Multiple sensitivity analysis methods are used: First, the effect of a single SNP on the overall result is assessed by leaving one out; second, the potential bias and heterogeneity of the instrumental variables are detected by funnel plot; finally, the effect size of each SNP on the cerebral cortex structure and olfactory dysfunction is displayed intuitively by forest plot. The heterogeneity analysis is quantitatively assessed by Cochran's Q test. When the P value of the Q statistic is less than 0.05, it indicates that there is significant heterogeneity. At this time, the random effects IVW model is used for re-estimation. In order to detect level pleiotropy, both MR-PRESSO and MR-Egger methods are used. When the detection result P < 0.05, the robustness of the result is re-evaluated after removing abnormal SNPs. The statistical significance level is set at P < 0.

05.

6. The analytical method for the effect of loss of smell on the structure of the cerebral cortex according to claim 5, characterized in that: In step (3), brain region gene microarray expression data from the Allen Human Brain Atlas (AHBA) are obtained. The AHBA is processed using the Abagen toolbox, and the transcription data is mapped to 83 brain regions based on the DKT map. Gene probes are re-annotated, and probes whose expression intensity is lower than the background noise level in more than 50% of the samples are removed. The tissue samples are spatially located using the corrected MNI coordinates, and mapped to the nearest brain region with a radius of 2 mm. To reduce the misallocation rate, the matching process of samples to brain regions follows the anatomical constraints of cerebral hemisphere division and cortical / subcortical partitioning. For samples that are not matched to a specific brain region through the above standard process, the nearest centroid method is used for supplementary allocation to ensure that all brain regions obtain the corresponding gene expression data. When multiple samples are assigned to the same brain region, the average expression value of each donor is calculated, and the robust sigmoid function is applied to normalize the brain region expression spectrum of each donor and convert it to a unit interval. The expression data of all donors are integrated to construct a gene expression matrix with brain regions as rows and 15,633 genes as columns.

7. The analytical method for the effect of loss of smell on the structure of the cerebral cortex according to claim 6, characterized in that: In step (3), PLS regression analysis is used to explore the relationship between the transcriptome of healthy brain and changes in cortical surface area and thickness in olfactory dysfunction. The PLS algorithm uses gene expression as the predictor variable and changes in SA and TH as the response variables. The first PLS component, PLS1, represents the maximum covariance between gene expression patterns and changes in SA / TH. The relative explained variance of 1-15 components is calculated, and significance is assessed by 1000 permutation tests. The standard error of gene PLS1 weights is estimated using bootstrap sampling. Z-scores are calculated to rank gene contributions. Replacement sampling is performed on 34 brain regions. With 5% FDR as the threshold, genes are divided into positive weight PLS+ and negative weight PLS- according to the positive or negative Z-score to obtain a list of genes related to changes in SA / TH.

8. The analytical method for the effect of loss of smell on the structure of the cerebral cortex according to claim 7, characterized in that: In step (3), Mendelian randomization based on aggregated data is used as an extension of MR to integrate GWAS data and gene expression data, providing conditions for identifying the causal effects of gene expression; the heterogeneity-dependent tool test method is used to distinguish between true causal relationships and genome linkage disequilibrium effects. By integrating genome-wide association study data and quantitative trait locus data of olfactory dysfunction, the SMR method was used to identify risk genes associated with olfactory dysfunction. In the SMR analysis, SNPs associated with loss of smell were used as instrumental variables to investigate the causal effect of exposure factors on the outcome, using default parameters and P-values. SMR <0.05 was set as the significance level; to test the heterogeneity of IVs, a HEIDI test was performed to filter out P-values. HEIDI To explore the relationship between genes related to loss of smell and genes related to SA and TH in SMR analysis, a list of genes related to loss of smell and overlapping genes with SA and TH were obtained, and the expression association of these overlapping genes was then analyzed.

9. The analytical method for the effect of loss of smell on the structure of the cerebral cortex according to claim 8, characterized in that: In step (3), a systems biology analysis was performed on the significant genes in SA-related genes and TH-related genes. The clusterProfiler software package was used to perform gene ontology functional annotation on the PLS1 negative weighted genes. The Metascape online platform was used to perform pathway enrichment analysis on the screened significantly differentially expressed genes. Key signaling pathways were visualized using the QuickGO tool; SA-related and TH-related genes were submitted to the Synapse gene ontology database SynGO for functional annotation, with cellular component information visualized using a sunburst plot; a protein-protein interaction network was constructed based on the STRING database, excluding text mining sources during analysis, and a high confidence threshold of 0.9 was used to screen for reliable interactions; to identify key regulatory nodes, network topology analysis was performed using the CytoHubba plugin on the Cytoscape platform; weighted cell type enrichment analysis was used to determine whether subsets of SA-related and TH-related genes were enriched in specific brain cell types, and this was validated using two independent human-derived cell type-specific gene datasets. First, single-cell transcription data from AHBA-seq were used, followed by repeated analysis using the DroNc-seq dataset to ensure that the results did not depend on a single dataset.

10. An analytical device for analyzing the impact of loss of smell on the structure of the cerebral cortex, characterized in that: It includes: The extraction module is configured to acquire high-resolution structural magnetic resonance imaging of patients with loss of smell using 3.0T magnetic resonance scanning and to extract the surface area and thickness of the cerebral cortex using FreeSurfer software. The assessment module is configured to obtain olfactory loss-related genetic variations based on the FinnGen database, combine them with genetic association data on cortical surface area and cortical thickness from the ENIGMA database, and use Mendelian randomization analysis to assess the causal effects of olfactory loss on cortical structure. The analysis module, configured based on the Allen brain atlas, uses partial least squares regression to identify genes associated with changes in cerebral cortical surface area and cortical thickness, and reveals their molecular mechanisms through enrichment analysis and cell type-specific analysis.

Images