Parcellating an Individual Subject's Cortical and Subcortical Brain Structures Using Snowball Sampling of Resting-State Correlations

• Published in: Cerebral cortex (New York, N.Y. 1991) Vol. 24; no. 8; pp. 2036 - 2054
• Main Authors: Wig, Gagan S., Laumann, Timothy O., Cohen, Alexander L., Power, Jonathan D., Nelson, Steven M., Glasser, Matthew F., Miezin, Francis M., Snyder, Abraham Z., Schlaggar, Bradley L., Petersen, Steven E.
• Format: Journal Article
• Language: English
• Published: United States Oxford University Press 01.08.2014
• Subjects: Adult; Brain - physiology; Brain Mapping - methods; Female; Humans; Magnetic Resonance Imaging - methods; Male; Neural Pathways - physiology; Neuropsychological Tests; Reproducibility of Results; Rest; Signal Processing, Computer-Assisted; Time Factors; Young Adult; snowball sampling; brain area parcellation; brain networks; boundary mapping; resting-state functional correlations; individual differences
• ISSN: 1047-3211; 1460-2199; 1460-2199
• DOI: 10.1093/cercor/bht056

We describe methods for parcellating an individual subject's cortical and subcortical brain structures using resting-state functional correlations (RSFCs). Inspired by approaches from social network analysis, we first describe the application of snowball sampling on RSFC data (RSFC-Snowballing) to identify the centers of cortical areas, subdivisions of subcortical nuclei, and the cerebellum. RSFC-Snowballing parcellation is then compared with parcellation derived from identifying locations where RSFC maps exhibit abrupt transitions (RSFC-Boundary Mapping). RSFC-Snowballing and RSFC-Boundary Mapping largely complement one another, but also provide unique parcellation information; together, the methods identify independent entities with distinct functional correlations across many cortical and subcortical locations in the brain. RSFC parcellation is relatively reliable within a subject scanned across multiple days, and while the locations of many area centers and boundaries appear to exhibit considerable overlap across subjects, there is also cross-subject variability-reinforcing the motivation to parcellate brains at the level of individuals. Finally, examination of a large meta-analysis of task-evoked functional magnetic resonance imaging data reveals that area centers defined by task-evoked activity exhibit correspondence with area centers defined by RSFC-Snowballing. This observation provides important evidence for the ability of RSFC to parcellate broad expanses of an individual's brain into functionally meaningful units.