The content discusses a novel end-to-end slice-to-volume registration (SVR) model for aligning 2D functional MRI (fMRI) slices with a 3D reference volume. The key highlights are:
Functional MRI is vital for neuroscience research and clinical practice, but head motion during scans can lead to distortion and biased analyses. Retrospective slice-level motion correction through SVR is crucial.
Previous deep learning-based SVR methods overlooked the uncertainty stemming from the input stack of slices and did not assign weighting or scoring to each slice.
The proposed Self-Attention SVR (SA-SVR) model incorporates a self-attention mechanism to enhance robustness against input data variations and uncertainties. It uses independent slice and volume encoders, and a self-attention module to assign pixel-wise scores for each slice.
Experiments on synthetic rigid motion generated from the Healthy Brain Network dataset show that SA-SVR achieves competitive performance in alignment accuracy compared to state-of-the-art methods, and significantly faster registration speed compared to conventional iterative methods.
The end-to-end SVR model facilitates real-time head motion tracking during fMRI acquisition, ensuring reliability and robustness against uncertainties in inputs.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Samah Khawal... at arxiv.org 04-09-2024
https://arxiv.org/pdf/2404.04546.pdfDeeper Inquiries