nipreps/fmriprep 20.1.0

on GitHub

The second minor release series of 2020 is finally here!

Release Notes

fMRIPrep 20.1.0 introduces a magnitude of new features and improvements. Originally nominated to become the first LTS (long-term support) version of fMRIPrep, this release has been supercharged with many new features and bug-fixes. To ensure long-term stability, we have postponed the LTS nomination to 20.2 to allow us unlocking the 20.1 earlier, and a more extensive stress testing of the series before jumping into a longer support commitment. Some key additions in this release include:

• A centralized configuration module keeping track of fMRIPrep's many options and run-time and environmental circumstances and settings. The new config module, which has been also propagated to other NiPreps dMRIPrep, MRIQC), comes to robustify the run-to-run replicability of fMRIPrep (e.g., tracking random seeds), make the option handling more modular but consistent (e.g., setting the ground for a command-line interface built off of the config module), and ease troubleshooting and telemetry.
• The anatomical preprocessing fast-track: a new experimental command-line option (--anat-derivatives <PATH>) checks that all necessary anatomical derivatives required by fMRIPrep are present under <PATH>, and skips the anatomical processing in full if fMRIPrep's expectations are met. Because now functional processing of many sessions and runs can be efficiently split into more digestible computational units (i.e., cluster job) while guaranteeing the exact same anatomical results are being used, this can significantly speed up longitudinal study preprocessing, and it is a fundamental optimization to process databases of densely scanned individuals such as My Connectome. This option is not recommended for single-session processing.
• A change in output CIFTI2 subcortical volume orientation to be compatible with HCP Pipeline tools and data.

Thanks

With thanks to Basile Pinsard, Joe B. Wexler, Noah Benson, and Marc Bue for contributions.

Caution!

As with all minor version increments, working directories from previous versions should not be reused.

Thank you for using fMRIPrep!

If you encounter any issues with this release, please let us know by posting an issue on our GitHub page!

CHANGES

• FIX: MultiLabel interpolations should not use float=True (#2147)
• FIX: Generate proper LTA transform prior BOLD sampling on surfaces (#2146)
• FIX: Temporary config file in work directory gets clobbered in parallel jobs (#2138)
• FIX: Dismiss echo entity on several derivatives and figures outputs (#2133)
• FIX: Correct summary report when using previously run recon-all (#2124)
• FIX: Ensure correct WM and CSF masks are picked in confounds workflow (#2128)
• FIX: Explicitly add default native resolution to volumetric outputs (nipreps/niworkflows#494)
• ENH: Finish the upstreaming of NiTransforms interfaces to NiWorkflows (#2132)
• ENH: Enable filtering for ANY or NONE in --bids-filter-file (#2123)
• ENH: Use new DerivativesDataSink from NiWorkflows 1.2.0 (#2114)
• ENH: Config module (#2018)
• ENH: Add option to ignore T2w / FLAIR images (#2015)
• ENH: Ensure subcortical volume in CIFTI is in LAS orientation (nipreps/niworkflows#484)
• ENH: Add option to skip brain extraction (#2039)
• ENH: Use CIFTI sampling for carpetplot when available (#2055)
• MAINT: Stop printing full boilerplate, black fmriprep/cli (#2119)
• MAINT: Ensure YAML loader is specified (#2125)
• MAINT: PIN tedana version (#2117)
• MAINT: Bump minimum Python to 3.7 (#2017)
• MAINT: Remove unused console scripts (#2048)
• MAINT: Reduce the overall size of outputs (nipreps/niworkflows#492)
• DOC: Update parallel subject neurostars link in FAQ (#2104)
• DOC: Add FAQ about reusing work directory (#2045)