GSMorph: Gradient Surgery for cine-MRI Cardiac Deformable Registration

TL;DR

GSMorph introduces a hyperparameter-free gradient surgery approach for deep learning-based cine-MRI cardiac registration, improving accuracy and smoothness without additional tuning or computational cost.

Contribution

The paper presents GSMorph, a novel model-agnostic method that reformulates the optimization process using gradient projection to balance registration accuracy and smoothness without hyperparameters.

Findings

Outperforms five state-of-the-art registration models.

Surpasses traditional methods like SyN and Demons.

Achieves superior registration accuracy and smoothness.

Abstract

Deep learning-based deformable registration methods have been widely investigated in diverse medical applications. Learning-based deformable registration relies on weighted objective functions trading off registration accuracy and smoothness of the deformation field. Therefore, they inevitably require tuning the hyperparameter for optimal registration performance. Tuning the hyperparameters is highly computationally expensive and introduces undesired dependencies on domain knowledge. In this study, we construct a registration model based on the gradient surgery mechanism, named GSMorph, to achieve a hyperparameter-free balance on multiple losses. In GSMorph, we reformulate the optimization procedure by projecting the gradient of similarity loss orthogonally to the plane associated with the smoothness constraint, rather than additionally introducing a hyperparameter to balance these two competing terms. Furthermore, our method is model-agnostic and can be merged into any deep registration network without introducing extra parameters or slowing down inference. In this study, We compared our method with state-of-the-art (SOTA) deformable registration approaches over two publicly available cardiac MRI datasets. GSMorph proves superior to five SOTA learning-based registration models and two conventional registration techniques, SyN and Demons, on both registration accuracy and smoothness.