Self-Supervised Isotropic Superresolution Fetal Brain MRI

TL;DR

This paper introduces SAIR, a self-supervised framework for superresolution of fetal brain MRI that performs well without requiring high-resolution training data, potentially improving clinical imaging workflows.

Contribution

SAIR is the first self-supervised single-volume superresolution method for fetal brain MRI, avoiding the need for high-resolution training data and handling motion artifacts.

Findings

SAIR performs comparably to multi-volume methods in simulated environments.

SAIR is effective across different noise levels and resolution ratios.

Qualitative results on clinical data show potential for integration into existing pipelines.

Abstract

Superresolution T2-weighted fetal-brain magnetic-resonance imaging (FBMRI) traditionally relies on the availability of several orthogonal low-resolution series of 2-dimensional thick slices (volumes). In practice, only a few low-resolution volumes are acquired. Thus, optimization-based image-reconstruction methods require strong regularization using hand-crafted regularizers (e.g., TV). Yet, due to in utero fetal motion and the rapidly changing fetal brain anatomy, the acquisition of the high-resolution images that are required to train supervised learning methods is difficult. In this paper, we sidestep this difficulty by providing a proof of concept of a self-supervised single-volume superresolution framework for T2-weighted FBMRI (SAIR). We validate SAIR quantitatively in a motion-free simulated environment. Our results for different noise levels and resolution ratios suggest that SAIR is comparable to multiple-volume superresolution reconstruction methods. We also evaluate SAIR qualitatively on clinical FBMRI data. The results suggest SAIR could be incorporated into current reconstruction pipelines.