Data-Driven Identification of the Regions of Interest for Fiber Tracking in Patients with Brain Tumors

Abstract

We investigated the added value of combining information from direction-encoded color (DEC) maps with high-resolution structural magnetic resonance imaging scans (T1-weighted images [T1WIs]) to improve the identification of regions of interest (ROIs) for fiber tracking during preoperative planning for patients with brain tumors. The dataset included 42 patients with gliomas and 10 healthy subjects from the Human Connectome Project. For identification of the ROIs, we combined the structural information from high-resolution T1WIs and the directional information from DEC maps. To test our hypothesis, we examined the interrater and intrarater agreement. We identified specific ROIs to extract the main white matter bundles. The directional information from the DEC maps combined with the T1WIs (T1WI-DEC maps) had significantly facilitated ROI identification in patients with brain tumors, especially patients in whom the tracts had been displaced by the mass effect of the tumor. Fiber tracking using the combined T1WI-DEC maps showed significantly greater inter- and intrarater agreement compared with using either T1WI or DEC maps alone. Combining the information from diffusion-derived color-encoded maps with high-resolution anatomical details from structural imaging (T1WI-DEC map), especially in patients with brain tumors, could be useful for accurate identification of the ROIs.