Impact of Segmentation Errors on QSM-based Brain Iron Estimation

Motivation: Studies have proposed iron content as an atrophy-independent metric in MS to study iron dyshomeostatis, however, existing T1-based segmentation tools like FSL FIRST may introduce systematic errors, potentially biasing group differences. Goal(s): To evaluate the effect of segmentation choice on DGM iron content measurements in MS where atrophy is common. Approach: We compared FSL FIRST and an enhanced bi-parametric joint label fusion (B-JLF) segmentation method in 109 people with MS and 55 healthy controls. Results: FIRST's segmentation errors introduced a bias that affected the group differences in iron content, whereas B-JLF yielding more reliable results. Impact: This study highlights that segmentation method choice impacts clinical findings on iron pathology in MS, underscoring the importance of accurate segmentation for unbiased interpretation of neuroimaging data.