Change in activation overlap for individual subject optimization.

<p>Points represent average pairwise activation overlap between independently optimized pipelines, comparing standard conservative preprocessing (CONS) against individually optimized pipelines using the D(P, R) metric (IND-D). Results are shown for (a) within-subject between-session overlap, and (b) between-subject, within-session overlap. Overlap is measured by Jaccard index between SPMs at a False Discovery Rate = 0.05 threshold, for each task and analysis model. The ±1 Standard Deviation ellipses are also plotted (enclosing ~68% of data points). To see individual subject overlap values, see <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0131520#pone.0131520.s002" target="_blank">S2 Fig</a>.