Fast image recovery using plug-and-play multi-scale energy (MuSE) framework

Jyothi Rikhab Chand; Mathews Jacob

doi:10.1109/IEEECONF59524.2023.10476808

Back

Conference proceeding

Fast image recovery using plug-and-play multi-scale energy (MuSE) framework

Jyothi Rikhab Chand and Mathews Jacob

2023 57th Asilomar Conference on Signals, Systems, and Computers, pp.32-35

10/29/2023

DOI: 10.1109/IEEECONF59524.2023.10476808

View Online

Abstract

We introduce a multi-scale energy formulation for plug-and-play (PnP) image recovery. The main highlight of the proposed framework is energy formulation, where the log prior of the distribution is learned by a convolutional neural network (CNN) module. The energy formulation enables us to introduce optimization algorithms with guaranteed convergence, even when the CNN module is not constrained as a contraction. Current PnP methods, which do not often have well-defined energy formulations, require the contraction constraint which restricts their performance in challenging applications. The energy and the corresponding score function are learned from reference data using denoising score matching, where the noise variance serves as a smoothness parameter that controls the shape of the learned energy function. We introduce a multi-scale optimization strategy, where a sequence of smooth approximations of the true prior are used in the optimization process. This approach improves the convergence of the algorithm to the global minimum, which translates to improved performance. The preliminary results in the context of MRI shows that the multi-scale energy PnP frame-work offers comparable performance to unrolled algorithms.

Magnetic Resonance Imaging

Noise Reduction

Approximation algorithms

Computational modeling

Convolutional neural networks

Energy model

Inverse problems

Parallel MRI

Plug-and-play

Score model

Shape

Details

Title: Subtitle: Fast image recovery using plug-and-play multi-scale energy (MuSE) framework
Creators: Jyothi Rikhab Chand - University of Iowa
Mathews Jacob - University of Iowa
Resource Type: Conference proceeding
Publication Details: 2023 57th Asilomar Conference on Signals, Systems, and Computers, pp.32-35
Publisher: IEEE
DOI: 10.1109/IEEECONF59524.2023.10476808
eISSN: 2576-2303
Language: English
Date published: 10/29/2023
Academic Unit: Electrical and Computer Engineering; Radiology; Iowa Technology Institute; Iowa Neuroscience Institute; Radiation Oncology
Record Identifier: 9984621258302771

Metrics

1 Record Views