Lucid Ant Colony Optimization based Denoiser for Effective Autism Spectrum Disorder Classification

Abstract

Autism Spectrum Disorder (ASD) indicates developmental disorders in neuro. Diagnosis of ASD involves analyzing clinical symptoms at the early manifest and imprecise communication with society. It arises a necessity to deeply analyze the resting-state functional magnetic resonance imaging (rsfMRI) for detecting the ASD. Presence of noise in rsfMRI reduces the detection rate of ASD. The main objective of this paper is to propose an optimization-based denoiser for efficient classification of ASD. Kalman filter is enhanced with ant colony optimization to detect the noises in rsfMRI more effectively. Lastly, convolutional neural network used to classify rsfMRI for ASD. The results with enhanced classification accuracy indicate that the proposed denoiser has better performance towards removing noises and helps classifier in detecting ASD.