Mass-Univariate and Pattern Classification Analysis on Structural MRI In Children with Autism Spectrum Disorders: a Focus on Females

Background:  The several but inconclusive studies on structural MRI in children with autism spectrum disorders (ASD) have mainly focused on samples entirely or prevailingly consisting of males. Sex differences in brain structure are observable since infancy and therefore caution is required in transferring to females the results obtained for males. The neuroanatomical phenotype of female children with ASD represents indeed a neglected area of research.

Objectives:  The purpose of this study is to investigate anatomical brain structures in a sample entirely composed of ASD female children, using mass-univariate and pattern classification approaches.

Methods:  Thirty-eight ASD females (ASDf) between 25 and 88 months (2-7 years of age; mean=53 months; SD=18) were compared to 38 female control children (19 idiopathic developmentally delayed, 19 typically developing) between 22 and 89 months (mean=53 months; SD=19) selected to carefully match by age and non-verbal IQ (NVIQ). Regional grey matter (GM) volumes were compared between the two groups using voxel-based morphometry (VBM) with the Diffeomorphic Anatomical Registration using Exponentiated Lie algebra (DARTEL) procedure.

Results:  

The subjects in the ASDf group showed significantly more GM in the superior frontal gyrus (SFG; MNI coordinates: -19 44 23), compared with the age and IQ matched controls. A pattern classification approach has also been implemented to discriminate between ASDf and control subjects. The GM segments obtained in the VBM-DARTEL preprocessing have been classified with a support vector machine (SVM). The recursive feature elimination (SVM-RFE) approach allows for the identification of the most discriminating voxels in the GM segments. This information is shown in the discrimination maps obtained for different threshold values on the area under the receiver operating characteristic curve (AUC), which is implemented as a figure of merit to estimate the classifier performance.

The regional cortical volume alteration revealed in the SVM-RFE map obtained with the most exiguous set of voxels (0.01% of the total GM voxels) with high discrimination power (AUC within the 2% of relative difference from AUC_max=0.80) is extremely consistent with the significant area in the superior frontal gyrus identified by the VBM analysis. Furthermore, the SVM-RFE map obtained with the most discriminating set of voxels (the 0.08% of the total GM voxels) corresponding to AUC_max, reveals a more complex circuitry of increased cortical volume in ASDf, involving bilaterally the superior frontal gyrus (MNI coordinates: -26 44 20; 26 50 10) and the right temporo-parietal junction (TPJ; MNI coordinates: 45 -55 26).

Conclusions:  These results suggest that structural anomalies of the brain cortical development are present in ASDf children and that these alterations in the SFG and TPJ might have a crucial role in the pathophysiology of ASD in female children.