Skip to content

@poquirion poquirion released this Sep 5, 2017 · 16 commits to master since this release

  • Fix log level bug in niak_cmd
Assets 4

@poquirion poquirion released this Aug 30, 2017 · 17 commits to master since this release

  • Bug fix with the 64 bit array containers
  • More option to select subjects in bids datasets
Assets 4

@poquirion poquirion released this Jul 8, 2017 · 39 commits to master since this release

New in this release

  • Includes a singularity image that support 64 bit array indexes in Octave.
Assets 4
Jul 8, 2017


Merge branch '_TMP_RELEASE_BRANCH_' into niak-cog

@poquirion poquirion released this May 10, 2017 · 51 commits to master since this release

  • Bug fix in the network list of the connectome report
Assets 4

@poquirion poquirion released this May 8, 2017 · 55 commits to master since this release

This is NIAK 1.0, including three production-ready pipelines: preprocessing of fMRI data (including attached structural MRI scan); generation of functional parcellations at the individual and group levels using a bootstrap analysis of stable clusters (BASC); Generation of connectomes and associated graph properties, as well as functional connectivity maps. Changes from the latest stable release (0.18.1) are as follows:

  • Estimation of motion parameters now uses the same spatial interpolation as for the final resampling of 4D data (instead of being fixed to linear interpolation).
  • Major bug fix in the BOLD-T1 registration, which was introduced in 0.17. The bug manifested itself by high failure rate on certain datasets, but had no real impact on others.
  • Jupyter notebook now fully functional for running NIAK with an octave kernel.
  • A new dashboard for brain maps.
  • A new dashboard for the connectome pipeline, including pipeline parameters and an interactive visualization of functional connectivity maps.
  • niak_glm now generates an estimate of effect size (Cohen's F2) for multivariate regression models.
  • the subtype (private) pipeline now supports multiple contrast estimation, uses network labels to name the outputs, and has numerically stable tests for linear regression.
  • niak_grab_folder has an improved black_list mechanism.
  • niak_brick_montage is ready for production of brain montage.
Assets 4
You can’t perform that action at this time.