Enabling Stable Cortical Reconstruction in the HCP Pipeline at Standard Resolution Using FastSurfer Integration with Hybrid T2w- and T1w-based Masking

Abstract

This study aimed to enable stable cortical reconstruction within the Human Connectome Project (HCP) pipeline under standard-resolution (1 mm3) conditions by developing an integrated framework combining FastSurfer with a hybrid T2w- and T1w-based masking strategy. We implemented a FastSurfer-integrated HCP pipeline incorporating a hybrid masking approach (t2log-hybrid) to stabilize cortical reconstruction, where SynthStrip-derived masks were refined using log-transformed T2-weighted and squared T1-weighted images. An anterior commissure-anchored spatial switching mechanism selectively replaced artifact-prone T2w regions with T1w-derived information in orbitofrontal areas. Performance was evaluated against FreeSurfer v6 (FS6), v7 (FS7), and FastSurfer using geometric agreement, regional consistency, and vertex-wise thickness-myelin analyses. The proposed framework reduced total processing time by over 50% compared with FS6 and maintained high geometric agreement with modern pipelines (vertex-wise r = 0.970 vs FastSurfer; r = 0.929 vs FS7). Agreement with FS6 was lower (r = 0.842), reflecting systematic differences in boundary definition rather than random error. The hybrid masking approach showed higher regional consistency and lower variability in T1w/T2w-derived myelin estimates (CV = 16.50% vs 40.56% in FastSurfer), with more spatially consistent thickness-myelin interaction patterns in artifact-prone ventral regions. Integrating FastSurfer into the HCP pipeline with hybrid T2w- and T1w-based masking enables stable cortical reconstruction by constraining signal-driven instability while preserving algorithm-dependent spatial organization. These findings highlight the importance of masking as a constraint on instability in cortical reconstruction and support the applicability of HCP-style analysis to clinical and legacy MRI datasets.