Adam teaching learning optimization enabled LeNet for autism spectrum disorder detection using brain MRI

Abstract

The autism spectrum disorder (ASD) is a complicated, lifetime, neuro-developmental circumstance of highly unknown causes. It is greatly more ordinary than formerly believed, frequency only to mental retardation amongst serious developmental disorders. Even though, a heritable element is demonstrated in the etiology of ASD, reputed risk genes are still to be detected. Therefore, Adam Teaching Learning Optimization-LeNet (ATLO-LeNet) is introduced for the detection of ASD utilizing brain Magnetic Resonance Imaging (MRI). Here, input considered is provided for the image pre-processing, wherein extraction of Region of Interest (ROI) is performed and median filter is used to eradicate noises. Then, extraction of the pivotal region is completed by the devised ATLO and hence, output-1 is obtained. On the other side, features are mined from an input MRI image. The extracted features include statistical features as well as texture features. From extracted features, output-2 is attained. By considering the output-1 and output-2, ASD classification is completed by LeNet. The LeNet is trained by ATLO, which is the amalgamation of the Adam algorithm and Teaching–Learning-Based Optimization (TLBO). Moreover, ATLO-LeNet acquired a maximum accuracy of 0.916, Negative Predictive Value (NPV) of 0.916and Positive Predictive Value (PPV) of 0.917 whereas it attained a minimum False Positive Rate (FPR) of 0.132 and False Negative Rate (FNR) of 0.152.