A Hybrid Frequency-domain/Image-domain Deep Network for Magnetic Resonance Image Reconstruction

TL;DR

This paper introduces a hybrid deep learning architecture that operates in both frequency and image domains to improve compressed sensing MRI reconstruction, especially in challenging regions, demonstrating promising results with volumetry accuracy.

Contribution

The study presents a novel hybrid neural network combining k-space and image domain processing for MRI reconstruction, outperforming existing image-only methods in hard-to-reconstruct regions.

Findings

Hybrid approach improves reconstruction quality in difficult regions.

Method achieves good volumetry agreement with fully sampled data.

Second place in quantitative analysis among compared methods.

Abstract

Decreasing magnetic resonance (MR) image acquisition times can potentially reduce procedural cost and make MR examinations more accessible. Compressed sensing (CS)-based image reconstruction methods, for example, decrease MR acquisition time by reconstructing high-quality images from data that were originally sampled at rates inferior to the Nyquist-Shannon sampling theorem. In this work we propose a hybrid architecture that works both in the k-space (or frequency-domain) and the image (or spatial) domains. Our network is composed of a complex-valued residual U-net in the k-space domain, an inverse Fast Fourier Transform (iFFT) operation, and a real-valued U-net in the image domain. Our experiments demonstrated, using MR raw k-space data, that the proposed hybrid approach can potentially improve CS reconstruction compared to deep-learning networks that operate only in the image domain. In this study we compare our method with four previously published deep neural networks and examine their ability to reconstruct images that are subsequently used to generate regional volume estimates. We evaluated undersampling ratios of 75% and 80%. Our technique was ranked second in the quantitative analysis, but qualitative analysis indicated that our reconstruction performed the best in hard to reconstruct regions, such as the cerebellum. All images reconstructed with our method were successfully post-processed, and showed good volumetry agreement compared with the fully sampled reconstruction measures.