One-stop Fine-grained Brain Iron Quantification Based on Mutual Transformer

Motivation: Brain iron imbalance contributes to neurological disorders, and precise quantification across regions is essential. Traditional methods suffer from registration errors that reduce measurement accuracy. Goal(s): This study aims to develop a one-stop fine-grained brain iron quantification model covering 115 brain regions, improving precision through multi-modal integration. Approach: A mutual Transformer model (mT-QSM) integrates QSM and T1-weighted data to perform simultaneous QSM reconstruction, brain segmentation, and iron quantification, reducing reliance on complex registration steps. Results: The mT-QSM achieved state-of-the-art reconstruction with fewer artifacts, and iron quantification across 115 regions showed high consistency with the gold standard. Impact: The proposed mutual Transformer and research framework can also be applied to other data fusion research, such as combining ASL or BOLD for one-stop analysis of cerebral blood flow or blood oxygenation, offering broad clinical application prospects.