Deep attention super-resolution of brain magnetic resonance images acquired under clinical protocols
-
Published:2022-08-25
Issue:
Volume:16
Page:
-
ISSN:1662-5188
-
Container-title:Frontiers in Computational Neuroscience
-
language:
-
Short-container-title:Front. Comput. Neurosci.
Author:
Li Bryan M.,Castorina Leonardo V.,Valdés Hernández Maria del C.,Clancy Una,Wiseman Stewart J.,Sakka Eleni,Storkey Amos J.,Jaime Garcia Daniela,Cheng Yajun,Doubal Fergus,Thrippleton Michael T.,Stringer Michael,Wardlaw Joanna M.
Abstract
Vast quantities of Magnetic Resonance Images (MRI) are routinely acquired in clinical practice but, to speed up acquisition, these scans are typically of a quality that is sufficient for clinical diagnosis but sub-optimal for large-scale precision medicine, computational diagnostics, and large-scale neuroimaging collaborative research. Here, we present a critic-guided framework to upsample low-resolution (often 2D) MRI full scans to help overcome these limitations. We incorporate feature-importance and self-attention methods into our model to improve the interpretability of this study. We evaluate our framework on paired low- and high-resolution brain MRI structural full scans (i.e., T1-, T2-weighted, and FLAIR sequences are simultaneously input) obtained in clinical and research settings from scanners manufactured by Siemens, Phillips, and GE. We show that the upsampled MRIs are qualitatively faithful to the ground-truth high-quality scans (PSNR = 35.39; MAE = 3.78E−3; NMSE = 4.32E−10; SSIM = 0.9852; mean normal-appearing gray/white matter ratio intensity differences ranging from 0.0363 to 0.0784 for FLAIR, from 0.0010 to 0.0138 for T1-weighted and from 0.0156 to 0.074 for T2-weighted sequences). The automatic raw segmentation of tissues and lesions using the super-resolved images has fewer false positives and higher accuracy than those obtained from interpolated images in protocols represented with more than three sets in the training sample, making our approach a strong candidate for practical application in clinical and collaborative research.
Funder
Wellcome Trust
UK Research and Innovation
China Scholarship Council
Mrs Gladys Row Fogo Charitable Trust
Stroke Association
Fondation Leducq
Garfield Weston Foundation
Dunhill Medical Trust
Medical Research Council
Edinburgh and Lothians Health Foundation
Publisher
Frontiers Media SA
Subject
Cellular and Molecular Neuroscience,Neuroscience (miscellaneous)
Reference80 articles.
1. Automatic correction of motion artifacts in magnetic resonance images using an entropy focus criterion;Atkinson;IEEE Trans. Med Imaging,1997
2. “Recycle-gan: unsupervised video retargeting,”;Bansal,2018
3. Random search for hyper-parameter optimization;Bergstra;J. Mach. Learn. Res,2012
4. “A framework for jointly assessing and reducing imaging artefacts automatically using texture analysis and total variation optimisation for improving perivascular spaces quantification in brain magnetic resonance imaging,”;Bernal,2020
5. “Analysis of spatial spectral features of dynamic contrast-enhanced brain magnetic resonance images for studying small vessel disease,”;Bernal,2019
Cited by
3 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献