A Self-Attention Model for Robust Rigid Slice-to-Volume Registration of Functional MRI

A self-attention based deep learning model for aligning 2D functional MRI slices with a 3D reference volume, improving robustness against input data variations and uncertainties.