Universal Generative Modeling for Calibration-free Parallel Mr Imaging

TL;DR

This paper introduces UGM-PI, an unsupervised deep learning framework that enables calibration-free parallel MRI reconstruction by leveraging wavelet transforms and score-based networks, outperforming existing methods.

Contribution

The paper presents a novel calibration-free deep learning approach for parallel MRI that combines wavelet transforms and score networks, eliminating the need for explicit coil sensitivity profiles.

Findings

Comparable or superior to state-of-the-art CS-PI methods on phantom and in vivo data.

Effective in accelerating MRI without calibration or explicit coil sensitivity estimation.

Demonstrates robustness across different datasets and imaging conditions.

Abstract

The integration of compressed sensing and parallel imaging (CS-PI) provides a robust mechanism for accelerating MRI acquisitions. However, most such strategies require the explicit formation of either coil sensitivity profiles or a cross-coil correlation operator, and as a result reconstruction corresponds to solving a challenging bilinear optimization problem. In this work, we present an unsupervised deep learning framework for calibration-free parallel MRI, coined universal generative modeling for parallel imaging (UGM-PI). More precisely, we make use of the merits of both wavelet transform and the adaptive iteration strategy in a unified framework. We train a powerful noise conditional score network by forming wavelet tensor as the network input at the training phase. Experimental results on both physical phantom and in vivo datasets implied that the proposed method is comparable and even superior to state-of-the-art CS-PI reconstruction approaches.