Self-supervised multi-contrast MRI denoising

Magnetic Resonance Imaging (MRI) is pivotal in medical diagnostics, offering essential multi-contrast imaging capabilities. However, MRI quality is often compromised by inherent noise, which can hinder both image clarity and analytical accuracy. This thesis presents the "Corruption2Self" (C2S) framework, a self-supervised method for multi-contrast MRI denoising. C2S utilizes self-generated pseudo-labels from noisy data to enhance contrast fusion and Signal-to-Noise Ratio (SNR), providing a robust solution that facilitates shorter scanning times or improved spatial resolution—critical factors in enhancing patient experience and diagnostic precision. Comparative tests on the M4Raw dataset show that C2S substantially surpasses traditional methods like BM3D and Noise2Self in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM). These results underscore the potential of self-supervised learning to improve multi-contrast MRI image quality.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical & Computer Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2024

Author dc:creator

• Tu, Jiachen

Contributors dc:contributor

• Lam, Fan

• Mri
• Denoising
• Self-supervised Learning
• Multi-contrast Mri

Statement dc:rights

• Copyright 2024 Jiachen Tu

Language dc:language

eng, en

Handle dc:identifier

https://hdl.handle.net/2142/124727

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/124727