DCAN-Labs/abcd-hcp-pipeline: DCANBOLDProcessing updates

This release makes a number of changes relating to the DCANBOLDProcessing module. Notably, the default behavior of the bandstop motion filter has been modified, so outputs of the DCANBOLDProcessing stage onward are expected to differ between this and prior versions.

For data previously processed on versions v0.1.3 - v0.1.5, it should be possible to rerun the processing on this version starting from DCANBOLDProcessing with --stage DCANBOLDProcessing (and --legacy-motion-filter, if the original motion filter behavior of those versions is desired), but we recommend rerunning the full pipeline with a fresh output directory if possible.

Changelist

FIX: DCANBOLDProcessing bandstop motion filter revised to apply half the number filter repetitions when a bidirectional filter is used (which is the default when using --bandstop). As a result, expect mean filtered FD values to be slightly greater compared to prior versions. (https://github.com/DCAN-Labs/dcan_bold_processing/pull/33) FIX: restore output of DCANBOLDProcessing motion outliers file (https://github.com/DCAN-Labs/dcan_bold_processing/pull/27) FIX: pass brain radius parameter correctly in DCANBOLDProcessing, instead of forcing default 50 mm (https://github.com/DCAN-Labs/dcan_bold_processing/pull/29) ENH: new option --legacy-motion-filter (use the "old" motion filter behavior, for consistency with v0.1.3 - v0.1.5 of this pipeline) ENH: new option --no-gsr (disables global signal regression)

Version bumps:

• external-software 0.0.6 -> 0.0.7
• internal-tools 1.0.11 -> 1.1.0rc2
• dcan_bold_processing 4.0.12 -> 4.1.0rc2
• Matlab MRE 9.1 (R2016b) -> 9.6 (R2019a)