Human Connectome Project Data Release and new Pipeline Posted By: Jennifer Elam - Jun 7, 2013Tool/Resource: HCP WU-Minn Consortium On June 11, 2013, HCP will release the Q2 datasets (68 subjects scanned between December, 2012 and February, 2013). All preprocessing of the Q2 data was done using Version 2 of the preprocessing pipelines. Version 2 includes a number of refinements compared to the Version 1 pipelines used for Q1 data released in March (see below for details). To maximize compatibility for analyses going forward, the Q1 datasets (also 68 subjects) have been reprocessed using the Version 2 pipelines. For any analyses that will use both Q1 and Q2 HCP subjects, we recommend using datasets that have all been processed using Version 2 pipelines. Accordingly, beginning June 11, 2013, only Version 2 of the Q1 and Q2 preprocessed datasets will be available via ConnectomeDB and via Connectome-in-a-Box. If you have been working with a subset of the Q1 data downloaded from ConnectomeDB and wish to obtain a complete set of Q1 Version 1 preprocessed data, we will accept "HCP Q1 Version1" Connectome-in-a-Box orders (i.e., legacy datasets) until Friday, June 7th. What's different between version 2 and version 1 preprocessing? Most of the changes are small; the refinements mainly involve simplification of output file management for subsequent analyses, especially for combined surface and volume analyses. The exception to this is the diffusion preprocessing, which is processed in a different coordinate space. See Glasser et al., 2013 (Neuroimage, special issue on The Connectome) for additional details. Specific changes implemented in Version 2 pipelines: o In all pipelines: § updated Connectome Workbench utilities (and removed Caret5 utilities), including full incorporation of CIFTI files into Workbench pipeline outputs § used improved algorithms for all surface resampling steps o In the Post FreeSurfer processing pipeline: § added creation of the 2mm grayordinates space § added normalization of generated myelin maps o In the fMRI pipelines: § incorporated backwards differentiation to get motion parameter derivatives § incorporated a mask to ensure motion correction does not introduce zeros into the data (from the brain having moved outside the field of view) o In the Diffusion processing pipeline: § added transformation of diffusion data, gradient deviation, and gradient directions to 1.25mm structural space Note: In the initial Q1 release, the preprocessed dMRI data was in the coordinate system of the individual diffusion scans. Remember to visit the Connectome-in-a-Box order website at http://humanconnectome.org/data/data-req... by June 7th if you wish to obtain a complete set of Q1 Version 1 preprocessed data. Sincerely, The HCP Informatics Team |
| Latest News |
|