Classification of Autism Spectrum Disorder Using Multi-modal Data Fusion

Abstract

Autism spectrum disorder (ASD) is a neurodevelopmental disorder that causes problem in social behavior and interactions. It can adversely affect the social functioning of the individuals. Each autistic individual is said to have, sort of, unique behavioral pattern. The spectrum contains sub categories as Autism, Asperger, and PDDs- NOS. The term spectrum indicates that it possess a large variety in terms of symptoms and severity. It is further observed that practitioners need to investigate variety of symptoms for accurate diagnosis of ASD. Symptoms may be observed from variety of brain scans and phenotypic data. These aspects present a multifold challenge for computer aided ASD diagnosis. In literature of automated diagnosis of autism, the autism spectrum is ignored rather only autistic/ control categories are considered. Similarly normally symptoms are taken from a single source such as just functional MRI or only clinical data etc. These challenges and gaps has been translated into motivation to present a method that covers the variety exhibited in the autism spectrum while considering the dire need of acquiring symptoms from variety of data sources. In the paper, all these challenges has taken into consideration and it proposes a method that has taken into account the concept of autism spectrum instead of just dealing with autistic and control only. Secondly the method has also considered the multi-modal data for considering variety of symptoms from different sources including phenotypic and neuroimaging. The methods has shown very encouraging output and the achieved results are evident of the efficacy of the proposed method.