Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Multi-task batteries for individual brain mapping: Optimal battery selection for brain parcellation and connectivity modeling

Bassel Arafat¹, Caroline Nettekoven¹, Jorn Diedrichsen¹; ¹University of Western Ontario

Presenter: Bassel Arafat

Group-level atlases are commonly used in neuroimaging to define regions-of-interest (ROIs) - however, they ignore the substantial inter-individual variability in brain organization. While resting state data can be used to derive individual functional maps, recent work has shown that maps obtained with a broad task battery generalize better to new mental states. With limited scanning time an important question becomes which tasks to choose for an optimal task battery. Here we propose to base this selection on the empirical activity maps themselves, and evaluate two selection strategies: One that seeks to maximize the imaging contrast in the region of interest (activation strength) and one that seeks to maximize the independence of different subregions (representational spread). Using simulations and real fMRI data, we show that representational spread consistently yielded better performance for brain parcellations and connectivity models. In simulations, representational spread outperforms activation strength and random selection for batteries from 3-16 tasks. We confirm these findings for real fMRI data, for the cases of cerebellar and cortical parcellations, and a cortico-cerebellar connectivity model. Our study therefore offers an automated method for optimizing task battery selection for different brain areas and demonstrates the value of principled task selection for individual brain mapping.

Topic Area: Brain Networks & Neural Dynamics

Extended Abstract: Full Text PDF