FireANTs: Adaptive Riemannian Optimization for Multi-Scale Diffeomorphic Matching

TL;DR

FireANTs introduces a GPU-accelerated, training-free multi-scale Riemannian optimization method for dense diffeomorphic image matching, achieving faster performance, lower memory usage, and broad robustness without domain-specific training.

Contribution

The paper presents FireANTs, a novel training-free, multi-scale Riemannian optimization algorithm that significantly improves speed and robustness in dense diffeomorphic image matching.

Findings

FireANTs runs 2.5x faster than ANTs on CPU.

FireANTs is up to 1200x faster on GPU.

FireANTs uses up to 10x less memory than deep learning methods.

Abstract

The paper proposes FireANTs, a multi-scale Adaptive Riemannian Optimization algorithm for dense diffeomorphic image matching. Existing state-of-the-art methods for diffeomorphic image matching are slow due to inefficient implementations and slow convergence due to the ill-conditioned nature of the optimization problem. Deep learning methods offer fast inference but require extensive training time, substantial inference memory, and fail to generalize across long-tailed distributions or diverse image modalities, necessitating costly retraining. We address these challenges by proposing a training-free, GPU-accelerated multi-scale Adaptive Riemannian Optimization algorithm for fast and accurate dense diffeomorphic image matching. FireANTs runs about 2.5x faster than ANTs on a CPU, and upto 1200x faster on a GPU. On a single GPU, FireANTs performs competitively with deep learning methods on inference runtime while consuming upto 10x less memory. FireANTs shows remarkable robustness to a wide variety of matching problems across modalities, species, and organs without any domain-specific training or tuning. Our framework allows hyperparameter grid search studies with significantly less resources and time compared to traditional and deep learning registration algorithms alike.