Table 1. QA metrics examined in the BXH/XCEDE Tools.

Metrics	Notes *
volmean	The mean intensity of each volume (time point) in the data.
masked_volmean	volmean of the masked and detrended data. Sensitive to spikes.
mean_difference	For a volume vol and a mean volume meanvol, this metric is the mean intensity of (vol—meanvol). Sensitive to slow drifts.
masked_tdiff_volmean	Masked, detrended running difference ('velocity'): this metric subtracts the volmean of a volume from the volmean of its subsequent volume.
cmassx,y and z	Center of mass in x, y, and z directions. Sensitive to head motions.
masked_outlier_percent	The percentage of outlier voxels in each masked volume.
masked_fwhmx,y,z	Full-width half-maximum in x, y and z directions. Used as a measure of data smoothness.
spectrummax	Maximum power for each frequency across all voxels.
per-slice variation	For each slice at each time point in the data, a measure of spikiness at slice granularity that is insensitive to artifacts that affect all slices (e.g., head motion). Higher numbers indicate a spike.
mean	A volume composed of the mean of each voxel across time.
standard deviation	A volume composed of the standard deviation of each voxel across time.
sfnr	Signal-to-Fluctuation Noise Ratio22. This is a SNR measure calculated for each brain voxel in the middle slice. It is essentially the average across time divided by standard deviation of detrended signal across time.

*These notes come from the resultant reports using fmriqa_generate.pl in BXH/XCEDE.