Role of Sex in Structural MRI Prediction of Cognition Severity in Autism

Document Type

Conference Proceeding

Source of Publication

Proceedings International Symposium on Biomedical Imaging

Publication Date

4-8-2026

Abstract

Autism and structural MRI (sMRI) are closely linked through measurable brain morphology differences that reflect disorder severity, which is measured by the Social Responsiveness Scale (SRS) Cognition subscore. This study investigates whether adding sex information improves sMRI-based prediction of autism severity. Using T1-weighted sMRI data from the ABIDE-II cohort, cortical and subcortical morphometric features (e.g., area, volume, thickness, and curvature) are extracted via FreeSurfer to classify typically developing (TD) versus mild, moderate, and severe autistic groups. Four experimental variants were evaluated both with and without sex information inclusion and Recursive Feature Elimination and Cross-Validation (RFECV). Ridge-based RFECV with a SMOTE Ridge classifier achieved the highest overall performance. Prediction performance was mostly enhanced for mild cases when sex information was added (F 1=77.0 ± 16.8%, AUC=94.8 ± 8.6%). Morphology alone had a little higher results for moderate (F1 =72.7 ± 10.0%, AUC=92.3 ± 5.9%) and severe (F1 =72.0 ± 11.5%, AUC =90.2 ± 7.0%) levels, although morphology alone decreased the F1 score of mild level by >8.5%. These findings suggest that sex contributes additional discriminative value mainly at early autism severity stages, whereas structural morphology dominates at higher levels.

ISBN

[9798331577636]

ISSN

1945-7928

Publisher

IEEE

Volume

2026-April

Disciplines

Computer Sciences | Medicine and Health Sciences

Keywords

ASD, Cognition, Morphological, Sex, sMRI

Scopus ID

105041635149

Indexed in Scopus

yes

Open Access

no

Link to Full Text

Share

COinS