Per-slice variation: this image shows, 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". It is computed as follows:

1. For each voxel remove the mean and detrend across time.
2. Calculate the absolute value of the z-score across time for each voxel.
3. For each slice at each time point, compute the average of this absolute z-score over all voxels in the single slice, producing a Z*T matrix AAZ.
4. For each slice at each time point, calculate the absolute value of the "jackknife" z-score of AAZ across all slices at that time point, producing a new Z*T matrix JKZ, which is the per-slice variation. (To compute a "jackknife" z-score, use all slices except the current slice to calculate mean and standard deviation. The jackknife has the effect of amplifying outlier slices.)

Douglas N. Greve, Nathan S. White, Syam Gadde, FIRST-BIRN. "Automatic Spike Detection for fMRI." Poster. Organization for Human Brain Mapping Annual Meeting, Florence IT 2006.

• Mean: this is the volume composed of the mean of each voxel across time.

• Standard deviation:this is the volume composed of the standard deviation of each voxel across time. The colorbar goes from 0 to 0.3 * the maximum intensity range of the mean volume.

• Signal-to-Fluctuation Noise Ratio (SFNR):this is a signal-to-noise (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.

  Friedman L, Glover GH, The FBIRN Consortium. "Reducing interscanner variability of activation in a multicenter fMRI study: Controlling for signal-to-fluctuation-noise-ratio (SFNR) differences." Neuroimage. September 2, 2006.

• Mask: this mask is generated by creating a histogram of voxel intensities, fitting a curve to the histogram, and choosing the first local minimum of the curve as a threshhold. The assumption of this algorithm is that the data will exhibit two "humps" in the histogram, the first being noise, and the second being actual brain signal.