A quality control pipeline for probabilistic reconstruction of white-matter pathways.

BACKGROUND

One of the most well-validated tools for DTI data analysis is TRACULA, part of the FreeSurfer software. TRACULA automatically segments 18 major white matter (WM) tracts. Occasionally, tracts may be only partially reconstructed, thus requiring intervention to avoid biasing analyses. A majority of studies have not reported any quality control procedures and those that have tend to discard partially reconstructed tracts from group analyses if they cannot be salvaged during TRACULA reinitialization.

NEW METHOD

We propose a semi-automated method to improve the detection and recovery of incomplete WM tracts. We detail several steps to maximize the quality of preprocessed DTI data. The steps include: (1) a visual inspection of eddy current corrected diffusion weighted images and (2) an automated evaluation of color- encoded FA images; (3) assessment of the volume of each tract saved in the TRACULA output file; (4) re-processing of tracts with a volume smaller than a specified threshold; (5) minimal manual editing of the control points for tracts that remained partially reconstructed; and (6) final re-initiation of TRACULA.

RESULTS

Our method can speed and improve quality control relative to tract-by-tract visual inspection and can recover data that otherwise would need to be excluded from analyses due to incomplete reconstruction.

COMPARISON WITH EXISTING METHODS

To our knowledge, there are no publications proposing alternative methods for quality control and recovering of partially reconstructed tracts in the TRACULA environment.

CONCLUSIONS

Our method helps TRACULA users automatically access the quality of reconstructed WM tracts and semi-automatically recover those in-complete WM tracts.