Unsupervised Parcellation of Precentral Gyrus Using Resting-State FMRI

Ali-Mohammad Golestani¹, Mariana Lazar¹

Anatomical landmarks are commonly used to identify regions of interest (ROIs) for brain imaging studies that might not reflect functional specialization. Previous studies showed that some of the anatomically defined ROIs can be subdivided into sub-regions based on their connectivity. Most of the parcellation approaches use clustering algorithms in which the number of clusters should be known in advance. We used a new clustering method that estimates the number and size of sub-regions, and parcellate precentral gyrus into posterior and inferior parts. This study highlights the importance of using a parcellation method that can automatically detect the number of clusters.