Note: Most Internet Explorer 8 users encounter issues playing the presentation videos. Please update your browser or use a different one if available.

Gender Differences in the Neuroanatomy of Young Children with Autism Spectrum Disorders Detected by Machine Learning Techniques

Saturday, 4 May 2013: 11:30
Meeting Room 1-2 (Kursaal Centre)
10:30
S. Calderoni1, A. Retico2, A. Giuliano2, L. Biagi3, M. Tosetti3 and F. Muratori4, (1)Magnetic Resonance Laboratory, Division of Child Neurology and Psychiatry University of Pisa; Stella Maris Scientific Institute, Pisa, Italy, (2)Istituto Nazionale di Fisica Nucleare; Sezione di Pisa, Pisa, Italy, (3)Magnetic Resonance Laboratory, Stella Maris Scientific Institute, Pisa, Italy, (4)University of Pisa – Stella Maris Scientific Institute, Pisa, Italy
Background: Genetic, hormonal, and environmental factors contribute since infancy to sexual dimorphism in regional brain structures of typical development subjects. However, the neuroanatomical differences between male and female children with autism spectrum disorders (ASD) are an intriguing and still poorly investigated issue.

Objectives: To evaluate whether the brain regions of children with ASD exhibit sex-related structural differences.

Methods: A total of 152 structural MRI scans were selected. Specifically, 76 ASD children (2-7 years of age; mean=53 months; SD=17) were evaluated employing a support vector machine (SVM) approach to the gray matter (GM) segmented with the SPM8 preprocessing algorithm, based on the diffeomorphic anatomical registration through exponentiated lie algebra (DARTEL) procedure. The leave-pair-out cross-validation protocol has been adopted to evaluate the classifier performance. The recursive feature elimination (RFE) procedure has been implemented both to reduce the large number of features in the classification problem and to enhance the classifier performance. The SVM-RFE allows also to localize the most discriminant voxels and to visualize them in a discrimination map. Group comparisons consisted of 76 age, gender and non-verbal IQ matched children with typical development or idiopathic developmental delay without autism. In addition, data from 38 male children with ASD and 38 female children with ASD were separately compared with the data of the corresponding age, gender and non-verbal IQ matched control subjects. 

Results: SVMs applied to GM scans correctly discriminate ASD male and female individuals with respect to controls with an area under the ROC curve (AUC) above the 87% with a fraction of retained voxels in the 0.4-29% range. By choosing as operative point of the system that corresponding to the lower amount of significant voxels (0.4% of the total number of voxel) we obtained a sensitivity of 0.82 and a specificity of 0.80. The discrimination maps obtained at that operative point showed the following main significant regions where the GM of ASD subjects (males and females) is greater than that of the matched control group: Left (L) and Right (R) Superior Frontal Gyrus (BA 10); L and R Precuneus (BA 31); R Temporo-Parietal Junction (BA 39); L Superior Temporal Gyrus (BA 22); R Superior Temporal Gyrus (BA 41). The separate analysis of the male and female subgroups revealed gender differences in the following regions where an excess of GM is found in the ASD subjects with respect to controls: L and R Precuneus dominates the male ASD group; L and R Superior Frontal Gyrus characterizes both males and females, whereas the Middle Frontal Gyrus prevails in the female group.

Conclusions: Multivariate approach based on the SVM could contribute not only to distinguish ASD from control children, but also to disentangle the gender specificity of ASD brain alterations. Regional neural differences between male and female ASD children could, in its turn, be related on sex-based differences in the phenotypic expression of ASD disorder. Future studies are therefore warranted to specifically investigate this issue.

See more of: Brain Imaging
See more of: Brain Imaging
See more of: Brain Structure & Function
| More