中国科学院心理研究所机构知识库

静息态功能磁共振成像(rsfMRI)测量人脑自发神经活动,安全可靠且易于操作而大规模实施,用于揭示认知活动与精神疾病神经影像变化,已成为人口神经科学收集大脑表型信息而倚重工具之一.高信度是rsfMRI测量个体差异的基本要求,但当前rsfMRI测量的信度水平偏低(多低于0.6),不同研究和不同测量指标信度差异较大,脑网络间信度存在较大异质性.当前, rsfMRI信度研究在领域内渐备受关注,如何开展rsfMRI信度研究亟待领域规范和指南.通过描述人口特征学、样本规模和测量安排三个信度研究维度对rsfMRI信度的影响,本文系统介绍rsfMRI信度研究进展,据此提出人口神经科学rsfMRI信度研究指南:基于人口神经科学规范化建模进行rsfMRI测量标准化,提供rsfMRI研究测量指标的信度信息;扩大研究样本量、高效安排扫描间隔和延长扫描时间;优化扫描参数、减少扫描噪声和进行标准化扫描;用标准化程序预处理静息态功能像数据,有效化计算和分析方法.

Population neuroscience is a discipline generated by the intersection of neuroscience (brain), genomics (gene), and epidemiology (environment), providing new perspectives and methods for studying a full picture of human brain structure and function. Resting state functional magnetic resonance imaging (rsfMRI) has become one of the most widely used tool for collecting brain phenotype information in population neuroscience. Its measurement of spontaneous neural activity in the human brain is safe, reliable, and easy to operate and implement in large-scale studies. RsfMRI can reveal the patterns of cognitive activity and psychiatric conditions. Its reliability is a fundamental requirement for measuring individual differences in the spontaneous brain activity. However, the importance of reliability in rsfMRI has not been fully recognized by the field. This article first provides a brief introduction to the definition and calculation of reliability of rsfMRI. By reviewing the literature, we describe the research progresses on the reliability of rsfMRI. Intraclass correlation (ICC) is the most ideal indicator for measuring the reliability of rsfMRI and used in most current research on rsfMRI to calculate the test-retest reliability of rsfMRI measurements. The overall proportion of reliability research on rsfMRI is small, and empirical research on reliability needs to be strengthened. The attention to the reliability of rsfMRI in the field is constantly increasing, and the reliability of rsfMRI is often involved and discussed. The overall measurement reliability level of rsfMRI is mostly below 0.6, mostly around 0.4; The reliability obtained from different studies and measurement indicators varies greatly, and there is also significant heterogeneity in the reliability between brain networks. Secondly, a study on the reliability of rsfMRI was proposed from three dimensions: demographic characteristics, sample size, and measurement arrangement. The impact of various factors in these three dimensions on the reliability of rsfMRI was summarized, including gender, age, sample size, scanner, magnetic field intensity, scanning parameters, physiological noise, head movement, body cognitive changes, open/closed eye status, retest interval, and scanning duration factors such as data preprocessing and functional indicators have varying degrees of impact on the reliability of rsfMRI. Finally, a guideline for the reliability study of population neuroscience with rsfMRI is proposed: (1) Referring to the brain chart of the entire human life cycle and using standardized modeling based on population neuroscience; (2) When conducting rsfMRI, report reliability as a necessary content and provide measurement indicator reliability information; (3) Drawing on a large amount of open access rsfMRI data resources for research, increasing the sample size, avoiding subjects from closing their eyes and sleeping during scanning, and the scanning interval should not be too long, the shorter the better. It is necessary to extend the scanning time each time; (4) Continuously optimizing scanning parameters, reducing scanning noise, and conducting standardized scans to improve the comfort level of participants during scanning, in order to reduce physical movement and cognitive changes, and monitor the scanning process in real time to pay attention to individual physical and mental changes; (5) Adopting standardized procedures to standardize the data rest state function and preprocessing process, selecting calculation indicators and methods to achieve the goal of improving research reliability. To better serve the indepth development of population neuroscience and the widespread use of rsfMRI, it is imperative to comprehensively reveal the reliability and influencing factors of rsfMRI. Population neuroscience still needs to improve research on the impact of demographic characteristics, sample size, and testing arrangements on the reliability of rsfMRI measurements. Soon, population neuroscience should also refer to disciplines such as psychometric to establish a complete set of rsfMRI data collection standards and guidelines, truly achieving stable and reliable measurement of data.

中国博士后科学基金(2022M710432);; 国家基础科学数据中心“中国活体人脑成像共享数据库”(NBSDC-DB-15)资助