Graph Convolutional Model to Diagnose Autism Spectrum Disorder Using Rs-Fmri Data

摘要

Human brain is a vital organ in the human body which has more than 100 billion nerves. They communicate through synapses. Few important things governed by brain are memory, emotions, creativity, intelligence etc. The human mind has three fundamental functions like figure out, react and relish. These functions of the mind are guided by one native egocentrism or potential rational capacities. Egocentric inclination happens involuntarily. ASD better termed as developmental delays. It can cause challenges in social interaction, behaviour and communication. Diagnosis of Autism includes both psychological and neurological examination. Early detection of the neurological markers with the help of MRI can help in to more accurate diagnosis. A combination of S-MRI (Structural Magnetic Resonance Imaging), fMRI (Functional Magnetic Resonance Imaging) and Rs-fMRI (Resting State- Functional Magnetic Resonance Imaging) can help in learning all characteristics of brain. Full brain connectivity and signal intensities of Region of Interest can be acquired using various atlases. ASD destroys the structure of the functional connectivity between the multiple brain regions. Deep learning methods proves to be successful in classification of imaging data. Provided with multi modal details provided by different MRI procedures, using a deep learning network can provide accurate classification results. Graph Convolutional Neural Networks can help in classifying whether the person has ASD or not. Given a FC matrix of the brain it is really huge to differentiate the significant features. Thus, it is requiring for dimensionality reduction. So, the purpose of using deep learning model using MRI is to identify Autism in early stage and to start with therapies.