A systematic review of structural MRI biomarkers in autism spectrum disorder: A machine learning perspective. Issue 71 (December 2018)

• Record Type: Journal Article
• Authors: Pagnozzi, Alex M.; Conti, Eugenia; Calderoni, Sara; Fripp, Jurgen; Rose, Stephen E.
• Is Part Of: International journal of developmental neuroscience. Issue 71(2018:Dec.)
• Page Start: 68
• Page End: 82
• Publication Date: 2018-12
• Subjects: Autism spectrum disorder -- Structural magnetic resonance imaging -- Biomarkers -- Machine learning; Developmental neurobiology -- Periodicals; Neurology -- Periodicals; Neurologie du développement -- Périodiques; Developmental neurobiology; Periodicals; 612.8
• DOI: 10.1016/j.ijdevneu.2018.08.010
• Languages: English
• ISSNs: 0736-5748

Abstract

Graphical abstract: Highlights: Reviews 123 studies investigating structural brain changes associated with ASD. Identifies biomarkers of brain structure for ASD diagnosis and treatment selection. Outlines directions for aggregating image data to elucidate inconsistent findings. Provides a discussion on the advantages and utility of machine learning methods. Abstract: Autism Spectrum Disorder (ASD) affects approximately 1% of the population and leads to impairments in social interaction, communication and restricted, repetitive behaviours. Establishing robust neuroimaging biomarkers of ASD using structural magnetic resonance imaging (MRI) is an important step for diagnosing and tailoring treatment, particularly early in life when interventions can have the greatest effect. However currently, there is mixed findings on the structural brain changes associated with autism. Therefore in this systematic review, recent (post-2007), high-resolution (3 T) MRI studies investigating brain morphology associated with ASD have been collated to identify robust neuroimaging biomarkers of ASD. A systematic search was conducted on three databases; PubMed, Web of Science and Scopus, resulting in 123 reviewed articles. Patients with ASD were observed to have increased whole brain volume, particularly under 6 years of age. Other consistent changes observed in ASD patients include increased volume in the frontal and temporal lobes, increased cortical thickness in the frontal lobe, increased surface area and cortical gyrification, and increased cerebrospinal fluid volume, as well as reduced cerebellum volume and reduced corpus callosum volume, compared to typically developing controls. Findings were inconsistent regarding the developmental trajectory of brain volume and cortical thinning with age in ASD, as well as potential volume differences in the white matter, hippocampus, amygdala, thalamus and basal ganglia. To elucidate these inconsistencies, future studies should look towards aggregating MRI data from multiple sites or available repositories to avoid underpowered studies, as well as utilising methods which quantify larger-scale image features to reduce the number of statistical tests performed, and hence risk of false positive findings. Additionally, studies should look to perform a thorough validation strategy, to ensure generalisability of study findings, as well as look to leverage the improved image resolution of 3 T scanning to identify subtle brain changes related to ASD.