Benchmarking 3D multi-coil NC-PDNet MRI reconstruction

TL;DR

This paper extends a state-of-the-art neural network for 3D multi-coil MRI reconstruction, benchmarking different undersampling patterns and training configurations, achieving high image quality with efficient inference suitable for clinical use.

Contribution

It introduces a 3D multi-coil extension of NC-PDNet, evaluates training strategies and undersampling patterns, and benchmarks performance on a public dataset.

Findings

Achieved an average PSNR of 42.98 dB for 3D brain MRI reconstruction.

Demonstrated inference time of 4.95 seconds and GPU memory usage of 5.49 GB.

Showed that training on compressed data with varying channels improves performance.

Abstract

Deep learning has shown great promise for MRI reconstruction from undersampled data, yet there is a lack of research on validating its performance in 3D parallel imaging acquisitions with non-Cartesian undersampling. In addition, the artifacts and the resulting image quality depend on the under-sampling pattern. To address this uncharted territory, we extend the Non-Cartesian Primal-Dual Network (NC-PDNet), a state-of-the-art unrolled neural network, to a 3D multi-coil setting. We evaluated the impact of channel-specific versus channel-agnostic training configurations and examined the effect of coil compression. Finally, we benchmark four distinct non-Cartesian undersampling patterns, with an acceleration factor of six, using the publicly available Calgary-Campinas dataset. Our results show that NC-PDNet trained on compressed data with varying input channel numbers achieves an average PSNR of 42.98 dB for 1 mm isotropic 32 channel whole-brain 3D reconstruction. With an inference time of 4.95sec and a GPU memory usage of 5.49 GB, our approach demonstrates significant potential for clinical research application.