Analysis of noise effects on DTI-based tractography using the brute-force and multi-ROI approach

Diffusion tensor tractography based on line propagation is a promising and widely used technique, but it is known to be sensitive to noise and the size and location of the seed regions of interest (ROIs). The effects of these parameters on the tractography results were analyzed quantitatively using high-resolution diffusion tensor imaging (DTI) with a high signal-to-noise ratio (SNR) on a fixed mouse brain. The anterior commissure (AC), as judged from a T₂-weighted image, was used as an anatomical reference within which the tracts could be located. Monte Carlo simulation was performed by adding Gaussian noise to the time domain data and repeating the tractography. Deviations of the tracking results were measured as a function of SNR. Such noise effects were evaluated for a simple one-ROI approach and a combined two-ROI and brute-force (BF) approach. The influence of ROI size and location for the two-ROI + BF approach was also analyzed. The results confirmed the hypothesis that one can increase the validity of DTI-based tractography by adopting the BF and multi-ROI approach, with respect to the simple one-ROI approach.