Single Image Compressed Sensing MRI via a Self-Supervised Deep Denoising Approach

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Lorenzana, Marlon Bran, Liu, Feng, Chandra, Shekhar S.
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866914929221566464
author Lorenzana, Marlon Bran
Liu, Feng
Chandra, Shekhar S.
author_facet Lorenzana, Marlon Bran
Liu, Feng
Chandra, Shekhar S.
contents Popular methods in compressed sensing (CS) are dependent on deep learning (DL), where large amounts of data are used to train non-linear reconstruction models. However, ensuring generalisability over and access to multiple datasets is challenging to realise for real-world applications. To address these concerns, this paper proposes a single image, self-supervised (SS) CS-MRI framework that enables a joint deep and sparse regularisation of CS artefacts. The approach effectively dampens structured CS artefacts, which can be difficult to remove assuming sparse reconstruction, or relying solely on the inductive biases of CNN to produce noise-free images. Image quality is thereby improved compared to either approach alone. Metrics are evaluated using Cartesian 1D masks on a brain and knee dataset, with PSNR improving by 2-4dB on average.
format Preprint
id arxiv_https___arxiv_org_abs_2311_13144
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Single Image Compressed Sensing MRI via a Self-Supervised Deep Denoising Approach
Lorenzana, Marlon Bran
Liu, Feng
Chandra, Shekhar S.
Image and Video Processing
Computer Vision and Pattern Recognition
Popular methods in compressed sensing (CS) are dependent on deep learning (DL), where large amounts of data are used to train non-linear reconstruction models. However, ensuring generalisability over and access to multiple datasets is challenging to realise for real-world applications. To address these concerns, this paper proposes a single image, self-supervised (SS) CS-MRI framework that enables a joint deep and sparse regularisation of CS artefacts. The approach effectively dampens structured CS artefacts, which can be difficult to remove assuming sparse reconstruction, or relying solely on the inductive biases of CNN to produce noise-free images. Image quality is thereby improved compared to either approach alone. Metrics are evaluated using Cartesian 1D masks on a brain and knee dataset, with PSNR improving by 2-4dB on average.
title Single Image Compressed Sensing MRI via a Self-Supervised Deep Denoising Approach
topic Image and Video Processing
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2311.13144