Self-supervised feature learning for cardiac Cine MR image reconstruction

TL;DR

This paper introduces a self-supervised learning framework for cardiac MRI reconstruction that leverages undersampled data to improve image quality without requiring fully-sampled images, outperforming existing methods.

Contribution

The authors develop a self-supervised feature learning approach that enhances MRI reconstruction from undersampled data, reducing reliance on fully-sampled datasets and improving generalization.

Findings

Outperforms existing self-supervised MRI reconstruction methods.

Achieves comparable or better results than supervised methods up to 16x undersampling.

Effectively extracts global features that improve artifact removal and generalization.

Abstract

We propose a self-supervised feature learning assisted reconstruction (SSFL-Recon) framework for MRI reconstruction to address the limitation of existing supervised learning methods. Although recent deep learning-based methods have shown promising performance in MRI reconstruction, most require fully-sampled images for supervised learning, which is challenging in practice considering long acquisition times under respiratory or organ motion. Moreover, nearly all fully-sampled datasets are obtained from conventional reconstruction of mildly accelerated datasets, thus potentially biasing the achievable performance. The numerous undersampled datasets with different accelerations in clinical practice, hence, remain underutilized. To address these issues, we first train a self-supervised feature extractor on undersampled images to learn sampling-insensitive features. The pre-learned features are subsequently embedded in the self-supervised reconstruction network to assist in removing artifacts. Experiments were conducted retrospectively on an in-house 2D cardiac Cine dataset, including 91 cardiovascular patients and 38 healthy subjects. The results demonstrate that the proposed SSFL-Recon framework outperforms existing self-supervised MRI reconstruction methods and even exhibits comparable or better performance to supervised learning up to $16\times$ retrospective undersampling. The feature learning strategy can effectively extract global representations, which have proven beneficial in removing artifacts and increasing generalization ability during reconstruction.