DD-CISENet: Dual-Domain Cross-Iteration Squeeze and Excitation Network for Accelerated MRI Reconstruction

TL;DR

This paper introduces DD-CISENet, a novel deep learning model that efficiently fuses k-space and image data iteratively for accelerated MRI reconstruction, significantly reducing errors compared to existing methods.

Contribution

The study proposes a dual-domain network with cross-iteration residual connections for improved sparse MRI reconstruction, outperforming prior deep learning approaches.

Findings

Reconstruction error of 2.28%, outperforming existing methods.

Outperforms image-domain prediction, k-space synthesis, and dual-domain fusion.

Validated on 720 multi-coil brain MRI cases from fastMRI dataset.

Abstract

Magnetic resonance imaging (MRI) is widely employed for diagnostic tests in neurology. However, the utility of MRI is largely limited by its long acquisition time. Acquiring fewer k-space data in a sparse manner is a potential solution to reducing the acquisition time, but it can lead to severe aliasing reconstruction artifacts. In this paper, we present a novel Dual-Domain Cross-Iteration Squeeze and Excitation Network (DD-CISENet) for accelerated sparse MRI reconstruction. The information of k-spaces and MRI images can be iteratively fused and maintained using the Cross-Iteration Residual connection (CIR) structures. This study included 720 multi-coil brain MRI cases adopted from the open-source fastMRI Dataset. Results showed that the average reconstruction error by DD-CISENet was 2.28 $\pm$ 0.57%, which outperformed existing deep learning methods including image-domain prediction (6.03 $\pm$ 1.31, p < 0.001), k-space synthesis (6.12 $\pm$ 1.66, p < 0.001), and dual-domain feature fusion approaches (4.05 $\pm$ 0.88, p < 0.001).