v0.1.3

Several quality of life updates.

• Added an option for qc_inference_pipeline to run on cpu only.
• Added an option to skip_masks listing in list_and_anon_bids.py (Relevant for SR)
• Refactored Docker building to have fewer layers.
• Refactored the inference.py to always run classifiaction and regression by default.
• Fixed a bug in LR report generation (iteration of svg_files was not correct in report.html)
• Fixed a bug in brain_extraction where failed mask extraction would make the script crash.
• Created a dedicated file for the parsers of run_inference and run_reports and refactored run_docker to use this interface.
• Updated readme

v0.1.2

Refactoring

• Removed unused decorators in metrics.py and directly passed **kwargs

Bugfix

• Fix problematic outputs when no segmentation or no good masks are available-

v0.1.1

Features

• Added the ability to generate visual reports for super-resolution reconstructed data.

Enhancement

• Computation of IQMs done by default on FetMRQC20
• Change the way conda_prefix is set in compute_segmentation to make sure that it can run from source as well
• Re-added ROC_AUC, F1W as metrics computed on the regression examples.
• Clarify the inputs of inference w.r.t classification/regression

Bugfixes

• Fixed the issue where segmentation maps would not be computed when giving the FetMRQC20 inputs to compute_iqms.py
• Added statsmodel as a requirement. This prevented the segmentation statistics to be computed in the dockerized version of the code.

v0.1.0

Initial release