Pivotal region and optimized deep neuro fuzzy network for autism spectrum disorder detection. (May 2023)

• Record Type: Journal Article
• Authors: RethikumariAmma, Kavitha Nair; Ranjana, Ponraj
• Is Part Of: Biomedical signal processing and control. Volume 83(2023)
• Publication Date: 2023-05
• Subjects: Autism spectrum disorder -- Deep neuro fuzzy network -- Pivotal region -- T2FCS -- Functional connectivity; 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.2023.104634
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: A new technique is devised for detecting the ASD from autism brain images. Functional connectivity with box neighborhood search generates pivotal regions. A modified Deep Neuro Fuzzy Network (DNFN) is applied for ASD. The DNFN training is executed using Feedback-Henry Gas Optimization (FHGO). FHGO-DNFN provided accuracy of 93.3%, sensitivity of 94.7% specificity of 91.4%. Abstract: Autism spectrum disorder (ASD) represents neuron chaos which may lead to premature death. Classical treatments of ASD are performed by observing the behaviors of patients. However, these methods take more time and can be ambiguous. The earlier treatment can assist the health providers to take necessary precautions and offer essential diagnosis to maintain patient lives. The aim is to learn the ASD on the basis of neuroimaging functional images that helps clinicians to diagnose ASD at an earlier stage. This model employs a deep model to discover if patients suffer from ASD and mine robust features from the image. It interprets the efficiency of pre-processed images for classifying the patterns of neurons. An earlier ASD diagnosis can help to achieve long-term goals. The functional connectivity based on the box neighborhood search algorithm is done to obtain pivotal regions. A Deep Neuro Fuzzy Network (DNFN) is applied for ASD wherein DNFN training is executed using Feedback-Henry Gas Optimization (FHGO). The proposed FHGO-DNFN performed exceptionally well, achieving the highest levels of 93.3% accuracy, 94.7% sensitivity, and 91.4% specificity.