Adaptive network based fuzzy inference system efficientnet for autism spectrum disorder detection with optimization based pivotal region extraction.

Autism Spectrum Disorder (ASD) refers to abnormal neural activity that may increase the risk of premature brain development issues. Many traditional treatment methods are available for ASD, but they require continuous assessment and analysis of patient behavior. However, traditional models are time-consuming and often lead to unclear outcomes, significantly impacting patient's social lives and communication ability. Hence, a novel hybrid approach is developed, namely, Adaptive Network Based Fuzzy Inference System EfficientNet (ANFIS-EffNet), which combines the ability of Adaptive Network Based Fuzzy Inference (ANFIS) and EfficientNet by modifying their layers. At first, the input autism brain image is given to the pre-processing stage, performed using Anisotropic diffusion and Region of Interest (ROI) extraction. Next, functional connectivity-based pivotal region extraction is done by Adam War Strategy Optimization (AWSO), a combination of the Adam optimization algorithm and War Strategy Optimization (WSO) technique. Next, feature extraction is carried out by including Learned Invariant Feature Transform (LIFT) and statistical methods. At last, ANFIS-EffNet is employed to detect autism spectrum disorder. The ANFIS-EffNet model has achieved outstanding performance with an accuracy of 93.219 %, sensitivity of 93.670 % and specificity of 93.840 % respectively.