An age-dependent Connectivity-based computer aided diagnosis system for Autism Spectrum Disorder using Resting-state fMRI. (January 2022)

• Record Type: Journal Article
• Authors: Haghighat, Hossein; Mirzarezaee, Mitra; Nadjar Araabi, Babak; Khadem, Ali
• Is Part Of: Biomedical signal processing and control. Volume 71(2022)Part A
• Publication Date: 2022-01
• Subjects: Autism Spectrum Disorder (ASD) -- Computer Aided Diagnosis System (CADS) -- Resting-state functional magnetic resonance imaging (rs-fMRI) -- Brain functional development -- Functional and Effective Connectivity (FC & EC) -- Independent Component Analysis (ICA); Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2021.103108
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Our proposed CADS was used functional and effective connectivity features between RSNs. Our proposed CADS achieved with 95.23% classification accuracy in the children group. Our results indicate the positive effect of considering age groups in CADS for ASD. Abstract: Autism spectrum disorder (ASD) is characterized by repetitive behaviors and social interactions. Due to the problems of diagnosing ASD using behavioral symptoms by experts, it seems necessary to propose accurate computer aided diagnosis systems (CADS) for ASD. Recent studies have reported brain connectivity as an important biomarker of ASD. Several studies have also suggested the role of age as an important factor in the brain connectivity disorders of individuals with ASD. In this study, we intend to present an age-dependent connectivity-based CADS for ASD using resting-state fMRI (rs-fMRI). First, the preprocessing was performed on the rs-fMRI data. Second, group independent component analysis (GICA) was applied to obtain the resting-state networks (RSNs). This was followed by obtaining individualized components of RSNs for each subject using dual-regression. Then, full and partial correlation measures were used to extract functional connectivity features and bivariate granger causality was used to extract effective connectivity features between RSNs. To consider the role of age in the classification process, three age groups of children, adolescents and adults were taken into account, and feature selection was performed for each age group separately by using an embedded approach in which all classifiers of WEKA were used simultaneously. Finally, classification accuracy, sensitivity and specificity were obtained for each age group. The proposed CADS was able to operate with 95.23% classification accuracy in the children group using classification via clustering classifier. Furthermore, discriminative biomarkers of functional connectivity were obtained in this age group which might play an important role in diagnosing ASD.