Investigation of automated brain region segmentation based on proton density maps

Motivation: Brain segmentation accuracy often relies on clear tissue contrast, yet standard T1-weighted images are susceptible to magnetic field inhomogeneities, and their contrast varies with acquisition parameters. We investigated proton density maps as a more robust alternative. Goal(s): To test the use of proton density maps obtained from phase-cycled bSSFP data as input for automated brain region segmentation. Approach: Proton density maps derived from phase-cycled bSSFP data were used as input for brain segmentation and compared with MP2RAGE-based segmentation. Results: Proton-density-based segmentation agrees with the reference up to a dice coefficient of 0.93, which is promising but further optimization is needed. Impact: Phase-cycled bSSFP-derived PD maps can be used for brain segmentation. This approach is independent of magnetic field strength, enabling more consistent contrast images for automatic brain segmentation tools.