Skip to main content
. Author manuscript; available in PMC: 2016 Jul 1.
Published in final edited form as: Neuroimage. 2015 Apr 7;114:379–385. doi: 10.1016/j.neuroimage.2015.03.074

Figure 2. Performance Metric Distribution for Different Pipelines.

Figure 2

Panel A displays the boxplots for performance measures when running the pipeline with a different fractional intensity (FI), using smoothed data (top) or unsmoothed data (bottom). Panel B presents the smoothed data only, rescaled to show discrimination between the different FI. Overall, FI of 0.01 and 0.1 perform better than 0.35 in all categories other than specificity. Using smoothed data improves performance in all performance metrics, markedly when an FI of 0.35 is used. Panel B demonstrates that using an FI of 0.01 on smoothed data has high performance on all measures.