Figure 2.
The contribution of each source of signal fluctuation varies between phase and magnitude data, and strongly depends on noise modeling. The pie charts show the fMRI data variance explained [%, mean (standard error) across six subjects of averaged values across voxels in ROIVC] by different noise sources (see legend). We employed as regressor for effects related to the phase of respiratory cycle: (A) Φnoise‐regressor; (B) four RETROICOR respiratory regressors. Note the much larger amount of variance explained by the Φnoise‐regressor in phase data than in magnitude data, and its larger contribution to the data, compared to that of RETROICOR regressors. Note that variance due to drifts over time is accounted before further noise regressor use; therefore, the fraction of the variance explained by drifts over time is identical when using different noise regression techniques. [Color figure can be viewed in the online issue, which is available at http://wileyonlinelibrary.com.]