Automated quality assessment of structural magnetic resonance images in children: Comparison with visual inspection and surface‐based reconstruction

White, Tonya; Jansen, Philip R.; Muetzel, Ryan L.; Sudre, Gustavo; Marroun, Hanan El; Tiemeier, Henning; Qiu, Anqi; Shaw, Philip; Michael, Andrew M.; Verhulst, Frank C.

doi:10.1002/hbm.23911

Cited by 58 publications

(76 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition to the need for objective exclusion criteria, the current neuroimaging data deluge makes the manual QC of every magnetic resonance imaging (MRI) scan impractical. For these reasons, there has been great interest in automated QC [5][6][7][8] , which is progressively gaining attention [9][10][11][12][13][14][15][16] with the convergence of machine learning solutions. Early approaches [5][6][7][8] to objectively estimate image quality have employed "image quality metrics" (IQMs) that quantify variably interpretable aspects of image quality [8][9][10][11][12][13] (e.g., summary statistics of image intensities, signal-to-noise ratio, coefficient of joint variation, Euler angle, etc.).…”

Section: Background and Summarymentioning

confidence: 99%

Crowdsourced MRI quality metrics and expert quality annotations for training of humans and machines

Estéban

Blair

Nielson

et al. 2018

Preprint

View full text Add to dashboard Cite

The neuroimaging community is steering towards increasingly large sample sizes, which are highly heterogeneous because they can only be acquired by multi-site consortia. The visual assessment of every imaging scan is a necessary quality control step, yet arduous and time-consuming. A sizeable body of evidence shows that images of low quality are a source of variability that may be comparable to the effect size under study. We present the MRIQC Web-API, an open crowdsourced database that collects image quality metrics extracted from MR images and corresponding manual assessments by experts. The database is rapidly growing, and currently contains over 100,000 records of image quality metrics of functional and anatomical MRIs of the human brain, and over 200 expert ratings. The resource is designed for researchers to share image quality metrics and annotations that can readily be reused in training human experts and machine learning algorithms. The ultimate goal of the database is to allow the development of fully automated quality control tools that outperform expert ratings in identifying subpar images.

show abstract

Section: Background and Summarymentioning

confidence: 99%

Crowdsourced MRI quality metrics and expert quality annotations for training of humans and machines

Estéban

Blair

Nielson

et al. 2018

Preprint

View full text Add to dashboard Cite

show abstract

“…showing that younger participants and male participants are more likely to have lower scan quality (Ducharme et al, 2016;Pardoe et al, 2016;White et al, 2018). These findings once again underline the importance of taking scan quality into account in developmental studies, as scan quality might confound findings attributed to age effects.…”

Section: Exploratory Analyses Of Age and Sex Effects Confirmed Previomentioning

confidence: 56%

“…A recent study found that image quality metrics (e.g., such as background voxels located outside the brain) derived from raw T1-weighted scans indeed affect FreeSurfer derived cortical thickness and surface area (White et al, 2018). However, raw image quality metrics typically use information from noise outside the head (Esteban et al, 2017;Pizarro et al, 2016;White et al, 2018). By using FreeSurfer derived measures from inside the brain we increase the likelihood of detecting artifacts that might remain unnoticed when they do not alter characteristics of the air around the head (Pizarro et al, 2016).…”

Section: Discussionmentioning

confidence: 99%

“…Our aim was to test which supervisedlearning model would best distinguish good quality FreeSurfer-processed scans from scans of insufficient quality by comparing three often used supervised-learning algorithms (see Methods for details). Although several studies have described automatic tools to assess the quality of (raw) T1-weighted MRI data (Alfaro-Almagro et al, 2018;Esteban et al, 2017;Pizarro et al, 2016;White et al, 2018;Woodard & Carley-Spencer, 2006) these procedures did not focus on automatic segmented data (e.g. as processed in FreeSurfer) but were performed early in the processing stream.…”

Section: Introductionmentioning

confidence: 99%

“…We propose that a data-driven supervised-learning model in combination with manual QC on a subset of the FreeSurfer-processed data could overcome many of the issues related to manual QC. The procedure introduced in the present paper moves beyond earlier studies on automated quality assessment for a number of reasons: 1) We are one of the first to apply automated QC procedures on FreeSurfer processed data in a developmental dataset (see also Rosen et al, 2017 for QC of T1-weighted and segmented scans in a developmental sample; White et al, 2018 for QC of T1-weighted scans in a developmental sample). Developmental samples would benefit particularly from a thorough QC, given their increased susceptibility to artifacts including motion (e.g., Power et al, 2012).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Qoala-T: A supervised-learning tool for quality control of automatic segmented MRI data

Klapwijk

Kamp

Meulen

et al. 2018

Preprint

View full text Add to dashboard Cite

Performing quality control to detect image artifacts and data-processing errors is crucial in structural magnetic resonance imaging, especially in developmental studies.Currently, many studies rely on visual inspection by trained raters for quality control.The subjectivity of these manual procedures lessens comparability between studies, and with growing study sizes quality control is increasingly time consuming. In addition, both inter-rater as well as intra-rater variability of manual quality control is high and may lead to inclusion of poor quality scans and exclusion of scans of usable quality. In the current study we present the Qoala-T tool, which is an easy and free to use supervised-learning model to reduce rater bias and misclassification in manual quality control procedures. First, we manually rated quality of N = 784 FreeSurferprocessed T1-weighted scans. Different supervised-learning models were then compared to predict manual quality ratings. Results show that the Qoala-T tool using random forests is able to predict scan quality with both high sensitivity and specificity (mean area under the curve (AUC) = 0.98). In addition, the Qoala-T tool was also able to adequately predict the quality of a novel unseen dataset (N = 112; mean AUC = 0.95). These outcomes indicate that using Qoala-T in other datasets could greatly reduce the time needed for quality control. More importantly, this procedure could further help to reduce variability related to manual quality control, thereby benefiting the comparability of data quality between studies.

show abstract

Gender Diversity and Brain Morphology Among Adolescents

et al. 2023

Self Cite

View full text Add to dashboard Cite

ImportanceGender-diverse youths have higher rates of mental health problems compared with the general population, as shown in both clinical and nonclinical populations. Brain correlates of gender diversity, however, have been reported only among youths with gender dysphoria or in transgender individuals.ObjectiveTo examine brain morphologic correlates of gender diversity among adolescents from a general pediatric population who were assigned male or female at birth, separately.Design, Setting, and ParticipantsThis cross-sectional study was embedded in Generation R, a multiethnic population-based study conducted in Rotterdam, the Netherlands. Adolescents who were born between April 1, 2002, and January 31, 2006, and had information on self-reported or parent-reported gender diversity and structural neuroimaging at ages 13 to 15 years were included. Data analysis was performed from April 1 to July 31, 2022.ExposuresGender-diverse experiences among adolescents were measured with selected items from the Achenbach System of Empirically Based Assessment forms and the Gender Identity/Gender Dysphoria Questionnaire for Adolescents and Adults, as reported by adolescents and/or their parents.Main Outcomes and MeasuresHigh-resolution structural neuroimaging data were collected using a 3-T magnetic resonance imaging scanner (at a single site). We used linear regression models to examine differences in global brain volumetric measures between adolescents who reported gender diversity and those who did not.ResultsThis study included 2165 participants, with a mean (SD) age of 13.8 (0.6) years at scanning. A total of 1159 participants (53.5%) were assigned female at birth and 1006 (46.5%) were assigned male at birth. With regard to maternal country of origin, 1217 mothers (57.6%) were from the Netherlands and 896 (42.4%) were from outside the Netherlands. Adolescents who reported gender diversity did not differ in global brain volumetric measures from adolescents who did not report gender diversity. In whole-brain, vertexwise analyses among adolescents assigned male at birth, thicker cortices in the left inferior temporal gyrus were observed among youths who reported gender diversity compared with those who did not. No associations were observed between gender diversity and surface area in vertexwise analyses.Conclusions and RelevanceThe findings of this cross-sectional study suggest that global brain volumetric measures did not differ between adolescents who reported gender diversity and those who did not. However, these findings further suggest that gender diversity in the general population correlates with specific brain morphologic features in the inferior temporal gyrus among youths who are assigned male at birth. Replication of these findings is necessary to elucidate the potential neurobiological basis of gender diversity in the general population. Future longitudinal studies should also investigate the directionality of these associations.

show abstract

Automated quality assessment of structural magnetic resonance images in children: Comparison with visual inspection and surface‐based reconstruction

Cited by 58 publications

References 41 publications

Crowdsourced MRI quality metrics and expert quality annotations for training of humans and machines

Crowdsourced MRI quality metrics and expert quality annotations for training of humans and machines

Qoala-T: A supervised-learning tool for quality control of automatic segmented MRI data

Gender Diversity and Brain Morphology Among Adolescents

Contact Info

Product

Resources

About