Advancing ASL Kidney Image Registration: A Tailored Pipeline with Groupwise VoxelMorph

Anne Oyarzun-Domeño, Izaskun Cia, Rebeca Echeverria-Chasco, María A. Fernández-Seara, Paloma L. Martin-Moreno, Nuria Garcia-Fernandez, Gorka Bastarrika, Javier Navallas, Arantxa Villanueva

Keywords: Arterial spin labeling, deep learning-based image registration, cortical signal

Abstract: Arterial spin labeling (ASL) provides a non-invasive assessment of renal blood flow, but it faces difficulties due to motion artifacts and the effects of blood inflow. This work introduces GVox, a deep learning-based motion correction (MoCo) framework tailored for ASL imaging. GVox extends VoxelMorph, incorporating cortical signal enhancement as metric to optimize and groupwise inference as main contribution. Proposed GVox demonstrates superior performance compared to the baseline Elastix, with significantly improved image similarity and computational efficiency.

Submission Number: 10