For subsequent pages in the processing stream suggestions and instructions, it is important to be at least somewhat familiar with the various formats supported and created by NeuroElf during the course of acquiring data. This pages gives an overview and contains links to further details about each of the formats.
NeuroElf has been growing for a few years and now supports a relatively wide range of formats used in neuro-imaging software packages, but began mostly as a toolbox to facilitate the use of BrainVoyager QX data formats in Matlab (i.e. to inspect results and import/manipulate files in BrainVoyager's diverse formats).
While BrainVoyager's formats, to this day, lack some of the features offered by other packages such as SPM (for instance allowing non-isometric reslicing of data), their main advantage (in particular when compared to formats used by SPM2 and, in part, SPM5) is the gain in performance, as the internal organization of data is almost always optimized w.r.t. improved data in- and output rates.
For the time being, the processing stream (once preprocessing is complete) as presented here re-uses BrainVoyager's file formats in the following instances (which might change in future versions):
In the table presented below, preferred formats are printed in bold.
|Processing stage||File format(s)||Comments|
|Data acquisition||DICOM, PAR/REC, Analyze, proprietary||while these files can be read with NeuroElf functions, they are not supported for data analysis directly; instead, all but Analyze files must first be converted (either into Analyze/NIftI or BrainVoyager format)|
|Data preprocessing||Analyze/NIftI, FMR, VMR||coregistration of FMR and VMR objects has not yet been implemented, as the preprocessing currently relies on a working SPM installation (either SPM5 or SPM8 for now)|
|After preprocessing||VTC||after the last step of classical preprocessing (slice-timing, realignment, coregistration, MNI-space non-linear warping, smoothing), further analysis in NeuroElf (e.g. regression) requires the data to be converted (loss-less) into VTC format|
|Regression||MDM, PRT or RTC/SDM, VTC → GLM||during first-level regression, the time courses from the VTC files are regressed against per-run design matrices (either in SDM format or, on-the-fly created design from protocol information in PRT formats)|
|Contrasts and correlations||GLM → VMP||as the final result (additional visualizations not withstanding), VMPs (maps) are created|
In addition to the main formats required to follow the entire processing stream (Analyze/NIftI during preprocessing, VTC, GLM, and VMP files), these formats can be incorporated during or after the analysis:
'rp_*.txt') can be incorporated during 1st level regression (as confounds)