Overview¶
DELTA-SVD is a containerised diffusion-MRI pipeline that turns preprocessed diffusion-weighted images into diffusion metrics designed for longitudinal tracking of change in cerebral small vessel disease (cSVD). It fits the diffusion (bi)-tensor, projects the resulting maps onto a white matter skeleton (TBSS-style), applies a custom mask and reports summary endpoints per timepoint and per region.
For a longitudinal run it builds a within-subject template so that the skeleton and the skeleton-derveid metrics are directly comparable across timepoints, rather than re-deriving them independently at each visit.
Pipeline at a glance¶
Preprocessed DWI (per timepoint) → tensor fit → white-matter skeleton projection → endpoints table + QC report
- Input — one preprocessed 4D DWI per timepoint, with its gradient table and a brain mask.
- Processing — diffusion-tensor fitting, then skeletonisation and masking; longitudinal runs first build a within-subject template so timepoints share one skeleton.
- Output — a metrics table (
delta-svd_results.csv) and an HTML quality-control report.
Typical input¶
Each timepoint is a preprocessed 4D DWI plus its b-values, b-vectors, and a DWI-space brain mask. DELTA-SVD does not preprocess the data itself. See Data requirements for acquisition guidance and the expected preprocessing.
Usage in a nutshell¶
DELTA-SVD runs as a container; you pass the pipeline's arguments after the image name:
Pass a single DWI for a cross-sectional run, or several (one per timepoint) to trigger longitudinal processing. Full options, mask handling, and cross-subject aggregation are in Usage.
Outputs¶
delta-svd_results.csv— the metrics table, reporting three validated endpoints per timepoint and per region:- MSMD — mean skeletonised mean diffusivity; the recommended primary endpoint for most datasets.
- PSMD — peak width of skeletonised mean diffusivity; an established marker of white-matter damage in cSVD.
- MSFW — mean skeletonised free water.
delta-svd_qc.html— a quality-control report with the skeleton and masks overlaid on the data.
For guidance on which endpoint to report, see the FAQ; for the full output detail and QC levels, see Usage.