LD-205：面向喉部肌张力障碍的全脑ROI图谱用于功能MRI分析的开发与技术验证

文献信息

中文摘要

目的： 喉部肌张力障碍（LD）是一种涉及多层级中枢网络的发声运动障碍。现有的LD功能磁共振成像（fMRI）研究使用异质性的感兴趣区（ROI）定义和分区方案，限制了可比性和可重复性。为了给ROI定义和全脑网络水平分析提供更一致的框架，我们旨在开发并技术验证一个标准化的、保留小结构的全脑ROI图谱，用于LD fMRI研究。
方法： 我们整合了自动解剖标记版本1（AAL1）衍生的脑区和选定的皮层下区域、空间无偏幕下模板（SUIT）小脑图谱、Najdenovska丘脑核团图谱以及Brainstem Navigator衍生的间脑和脑干区域。使用最近邻插值和小结构优先合并将各图谱协调至蒙特利尔神经学研究所（MNI）空间，生成1 mm参考图谱（LD-208）和2 mm fMRI就绪图谱（LD-205）。质量控制包括ROI连续性、零体素检查、体素计数、连通分量和来源追踪。LD-205在10名健康参与者的静息态fMRI数据中进行了测试。
结果： LD-208和LD-205分别包含208个和205个ROI。与LD-208相比，LD-205在2 mm网格上未保留三个极小的脑干ROI。两个图谱均具有连续的ROI编号且无零体素ROI。所有LD-205 ROI在每位参与者中均产生有效体素、平均血氧水平依赖（BOLD）时间序列以及完整的205 × 205连接矩阵。
结论： LD-208和LD-205为面向LD的fMRI研究提供了可追溯且经过质量控制的全脑ROI框架。技术验证表明将LD-205应用于常规静息态fMRI数据具有可行性，并支持其作为未来疾病特异性研究的标准化ROI框架。

英文摘要

OBJECTIVES: Laryngeal dystonia (LD) is a voice motor disorder involving multilevel central networks. Existing LD functional magnetic resonance imaging (fMRI) studies use heterogeneous region-of-interest (ROI) definitions and parcellation schemes, limiting comparability and reproducibility. To provide a more consistent framework for ROI definition and whole-brain network-level analysis, we aimed to develop and technically validate a standardized, small-structure-preserving whole-brain ROI atlas for LD fMRI research.
METHODS: We integrated Automated Anatomical Labeling version 1 (AAL1)-derived cerebral and selected subcortical regions, the Spatially Unbiased Infratentorial Template (SUIT) cerebellar atlas, the Najdenovska thalamic nuclei atlas, and Brainstem Navigator-derived diencephalic and brainstem regions. Atlases were harmonized to Montreal Neurological Institute (MNI) space with nearest-neighbor interpolation and small-structure-prioritized merging, generating a 1-mm reference atlas (LD-208) and a 2-mm fMRI-ready atlas (LD-205). Quality control included ROI continuity, zero-voxel checks, voxel counts, connected components, and source tracing. LD-205 was tested in resting-state fMRI data from 10 healthy participants.
RESULTS: LD-208 and LD-205 contained 208 and 205 ROIs, respectively. Compared with LD-208, LD-205 did not retain three extremely small brainstem ROIs on the 2-mm grid. Both atlases had continuous ROI numbering and no zero-voxel ROIs. All LD-205 ROIs yielded valid voxels, mean blood oxygen level-dependent (BOLD) time series, and complete 205 × 205 connectivity matrices in every participant.
CONCLUSIONS: LD-208 and LD-205 provide traceable and quality-controlled whole-brain ROI frameworks for LD-oriented fMRI research. The technical validation demonstrates the feasibility of applying LD-205 to conventional resting-state fMRI data and supports its use as a standardized ROI framework for future disease-specific studies.