MRIQC - Singularity Images associated with paper https://doi.org/10.1371/journal.pone.0184661

<a href="https://embed.stanford.edu/iframe/?url=https%3A%2F%2Fpurl.stanford.edu%2Ffr894kt7780" class="su-underline">Show Content</a>

Abstract/Contents

Abstract: The quality control of MRI is essential for excluding problematic acquisitions and avoiding bias in subsequent image processing and analysis. Visual inspection is subjective and inapplicable in large scale datasets. Although automated quality assessments have been demonstrated on single-site datasets, those solutions do not generalize to unseen data acquired at new sites. Here, we evaluate the MRI Quality Control tool –MRIQC– in an experimental setup that addresses the generalization problem. MRIQC calculates a set of quality measures from each image and uses them as features in a binary (accept/exclude) classifier. The classifier is trained on a publicly available and multi-site dataset (17 sites, N=1102). We perform model selection and estimate an accuracy of 76%±13% on new sites using nested cross-validation, with a leave one-site-out split strategy. We confirm that result on a held-out dataset (2 sites, N=265) obtaining a 76% accuracy. We show how the classifier is blind to idiosyncratic artifacts and site effects as a result of over-fitting the training dataset. MRIQC performs with high accuracy in intra-site prediction, but improving the performance on new sites will require more labeled data and new approaches to overcome the between-site variability of the quality features. Breaking those limitations is crucial for a more objective quality assessment of neuroimaging, and to enable the analysis of extremely large and multi-site samples.

Type of resource	software, multimedia
Date created	January 15, 2016 - [ca. May 31, 2017]

Subject	MRIQC
Subject	singularity
Subject	Department of Psychology
Subject	Center for Reproducible Neuroscience

Related Publication	Esteban O, Birman D, Schaer M, Koyejo OO, Poldrack RA, Gorgolewski, KJ. (2017) MRIQC: Advancing the automatic prediction of image quality in MRI from unseen sites. PLOS ONE 12(9): e0184661. https://doi.org/10.1371/journal.pone.0184661
Location	https://purl.stanford.edu/fr894kt7780

Use and reproduction: User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor.
License: This work is licensed under a Creative Commons Attribution 3.0 Unported license (CC BY).

Preferred Citation: Esteban O, Birman D, Schaer M, Koyejo OO, Poldrack RA, et al. (2017) MRIQC: Advancing the automatic prediction of image quality in MRI from unseen sites. PLOS ONE 12(9): e0184661. https://doi.org/10.1371/journal.pone.0184661

Stanford Research Data

Loading usage metrics...