FOAL: Fast Online Adaptive Learning for Cardiac Motion Estimation

TL;DR

FOAL introduces a fast online adaptive learning framework for cardiac MRI motion estimation, significantly improving accuracy and robustness in clinical settings by adapting to dataset variations in real-time.

Contribution

The paper presents a novel meta-learning based online adaptive framework that enhances deep learning models for cardiac motion estimation in clinical environments.

Findings

FOAL outperforms offline-trained methods in accuracy.

Online optimization takes only 0.4 seconds per video.

Demonstrates robustness across two public clinical datasets.

Abstract

Motion estimation of cardiac MRI videos is crucial for the evaluation of human heart anatomy and function. Recent researches show promising results with deep learning-based methods. In clinical deployment, however, they suffer dramatic performance drops due to mismatched distributions between training and testing datasets, commonly encountered in the clinical environment. On the other hand, it is arguably impossible to collect all representative datasets and to train a universal tracker before deployment. In this context, we proposed a novel fast online adaptive learning (FOAL) framework: an online gradient descent based optimizer that is optimized by a meta-learner. The meta-learner enables the online optimizer to perform a fast and robust adaptation. We evaluated our method through extensive experiments on two public clinical datasets. The results showed the superior performance of FOAL in accuracy compared to the offline-trained tracking method. On average, the FOAL took only $0.4$ second per video for online optimization.