On spatial selectivity and prediction across conditions with fMRI

TL;DR

This paper introduces two machine learning methods, transfer learning and selection transfer, to identify task-related brain regions in fMRI data, demonstrating that selection transfer provides more context-specific ROIs.

Contribution

It compares transfer learning and selection transfer for defining ROIs in fMRI, showing selection transfer's advantage in context-specific spatial localization.

Findings

Selection transfer yields more specific ROIs than transfer learning.

Selection transfer identifies well-known regions like the Visual Word Form Area.

Preliminary quantification of similarities between brain activation maps.

Abstract

Researchers in functional neuroimaging mostly use activation coordinates to formulate their hypotheses. Instead, we propose to use the full statistical images to define regions of interest (ROIs). This paper presents two machine learning approaches, transfer learning and selection transfer, that are compared upon their ability to identify the common patterns between brain activation maps related to two functional tasks. We provide some preliminary quantification of these similarities, and show that selection transfer makes it possible to set a spatial scale yielding ROIs that are more specific to the context of interest than with transfer learning. In particular, selection transfer outlines well known regions such as the Visual Word Form Area when discriminating between different visual tasks.