fMRI Imaging Based Human Brain Parcellation Methods: A review

Neuroimaging analysis methods.

“…For the atlas-based approach, the whole brain is parcellated into spatially discrete regions based on either the sulcal and gyral landmarks (e.g., automated anatomical labeling; Tzourio-Mazoyer et al, 2002) or probabilistic maps, and each region is taken as a node in the graph. In recent years, the boundary between the atlas-based and datadriven approaches has become blurry due to more advanced methodologies employed in the resting-state fMRI analysis to parcellate the brain into finer divisions of cortical areas (Cohen et al, 2008;Gordon et al, 2016;Schaefer et al, 2018;Wig et al, 2014;Xu et al, 2016;Yeo et al, 2011; and see the review of Abdedayem et al, 2020). These recent maps yield more reliable and neurobiologically meaningful parcellations of cortical areas.…”

A novel 5D brain parcellation approach based on spatio-temporal encoding of resting fMRI data from deep residual learning

“…Atlas-based approaches are widely used and typically segment the brain by applying a predefined anatomical template on the input image and generating a set of regions over associated template. Despite the popularity of these methods, there are drawbacks [14,39]. Obviously, these methods rely heavily on the given template and do not take natural variance of the brain into account for different subjects beyond the geometric warping of the fixed atlas into a standardized space using either voxelbased or surface-based approaches.…”