PSYCH OpenIR  > 认知与发展心理学研究室
多模态脑成像信号在解释个体差异中的贡献度研究
Alternative TitleThe Contributions of Multimodal Brain Imaging Signals in Explaining Individual Differences
陈宁轩
Contributor严超赣
2021-06
Abstract多模态技术手段已经广泛应用于脑科学研究,但是针对特定的研究问题,仍然缺乏能够指导选择特定模态技术手段的实证研究。脑成像领域中的基础研究主要与个体特征有关,临床应用研究往往和疾病特征与治疗疗效相关。据此,本研究依据时间维度,挑选了四个变量,即性别(进化维度)、年龄(超长时程)、是否患精神疾病(长时程)和治疗效果(中、短时程)作为不同时程的研究对象,比较了不同模态指标及其贡献度在解释相应变量的差异。对于特定的研究问题,我们采用变异分解的方法比较不同模态指标的贡献度。变异分解可以分离某个变量不同来源的方差,得到各自独立贡献和共同贡献,以实现贡献度比较的目的。本研究中,针对特定变量,我们首先通过构建包含两个模态指标的二元一次方程,对其进行模型拟合显著性检验;之后在显著脑区进行变异分解,得到两个模态指标的独立变异,共同变异以及残差。之后,我们使用置换检验去比较两个模态指标贡献度之间的差异。在研究一中,通过使用磁共振结构指标和功能指标,我们发现基于体素的结构指标对于性别的解释作用要优于功能指标,而基于皮层的分析没有得到相应结果,这可能与性别差异的相关脑区主要是皮层下脑结构有关;在研究二中,我们使用类似的方法,发现基于体素的结构指标对于年龄的解释作用要优于功能指标,而基于皮层的结构指标和功能指标的贡献度在不同脑区有不同的表现;在研究三中,我们发现在区分个体是否患有重性抑郁障碍时,基于体素的结构指标的解释作用要优于功能指标;而在研究四中,同样使用磁共振结构指标和功能指标,对于精神分裂症不管是短时程还是中时程治疗,功能指标对精神分裂症疗效的解释作用都要优于结构指标。因此,通过变异分解,我们针对不同的变量可以探索和比较多模态指标的贡献。综上所述,基于目前的研究进展,对于固有生物学特征变量(性别),我们推荐使用结构指标;对于长时程变化变量(年龄),我们可以针对不同的感兴趣区,选择特定的结构指标或功能指标;对于个体是否患有重性抑郁障碍,我们也推荐使用结构指标;而对于解释精神分裂症疗效,功能指标比结构指标贡献度更大,我们更推荐使用功能指标。上述结果为后人针对类似的研究主题选择特定的技术手段提供了指导建议。
Other AbstractMultimodal neuroimaging methods have been widely used in brain research. But for specific brain-related questions, there still lack empirical research that can guide the selection of specific multimodal neuroimaging methods. The fundamental research of brain imaging is mainly related to demographic, and clinical variables. In accordance with the different time scales, this study selected sex (evolutional scale), age (super long-term), diagnose results (long-term), and treatment effect (medium- or short-term) as research variables, and compared the contribution of multimodal indexes to explain the corresponding variables. For specific research questions, we used variance partitioning to compare the contribution of different modal indexes. This method could separate the portion of variances of different multimodal metrics to explain a specific variable. In this study, for specific predictor variables, we conducted a model fit test with two variables (one structural metric and another functional metric) and then performed a variance partitioning that could be explained sufficiently. Permutation tests were applied to compare the contribution difference between each pair of measurements. In study one, we found that voxel-based structural metrics are better than functional metrics in explaining sex, while surface-based analysis did not reveal any findings. As sex differences mainly manifested on the subcortical brain structure, the voxel-based structural metrics might be a better option. In study two, we used similar methods and found that by using voxel-based analysis, structural metrics had better explainable effect than functional metrics for age in most brain areas, while by surface-based analysis, structural metrics and functional metrics had different explaining effects in different brain regions. This indicated that for explaining aging, we should consider to pick structural or functional metrics on certain brain regions of interest. For study three, we found that structural metrics had better explainable effect than functional metrics for discriminating major depression disorder or health control. But in the fourth study, no matter explaining medium-term or short-term treatment, voxel-based functional metrics showed stronger explainable value for treatment. This indicated that relatively short-term alterations like treatment effects may be better revealed by functional metrics. In sum, for constant biological feature (sex), the voxel-based structural metrics may be a better option than voxel-based functional metrics in subcortical areas; for long-term change variable (age), we should carefully pick structural of functional metrics on specific brain areas; for checking whether or not suffering from mental illness (major depression disorder), the voxel-based structural metrics may be a better option too; but for short-term or medium-term change variable (schizophrenia treatment), we should focus more on the functional measurements.
Keyword变异分解 结构指标 功能指标 基于体素的分析 基于皮层的分析
Subtype博士
Language中文
Degree Name理学博士
Degree Discipline认知神经科学
Degree Grantor中国科学院心理研究所
Place of Conferral中国科学院心理研究所
Document Type学位论文
Identifierhttp://ir.psych.ac.cn/handle/311026/39604
Collection认知与发展心理学研究室
Recommended Citation
GB/T 7714
陈宁轩. 多模态脑成像信号在解释个体差异中的贡献度研究[D]. 中国科学院心理研究所. 中国科学院心理研究所,2021.
Files in This Item:
File Name/Size DocType Version Access License
陈宁轩- 博士学位论文.pdf(19536KB)学位论文 限制开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[陈宁轩]'s Articles
Baidu academic
Similar articles in Baidu academic
[陈宁轩]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[陈宁轩]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.