Brain Connectivity Network Analysis and Classifications from Diffusion Tensor Imaging

In this paper an automated method was introduced for classifying Autism Spectrum Disorder (ASD) and Typically Developed (TD) brain based on graph theory-based features and classification using Diffusion Tensor Imaging (DTI). DTI is a promising method for characterizing microstructural changes in brain. This technique detects how water travels along the white matter tracks in brain. It is an MRI based neuroimaging technique. Using the created connectivity matrices developed from DTI, a graph theory-based analysis is performed and using that analysis some features have been extracted. Using these features, a classification work has been done between ASD and TD brains based on Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA). Principle Component Analysis (PCA) is an approach of reducing noisy features for better accuracy. Using SVM, classification accuracy of 75.00%, 62.50% & 56.25% are achieved from 2, 7 & 10 PCA features. Accuracy of 64.58%, 60.42% & 58.33% are achieved using LDA. Using less features is also showing better accuracy comparing same test samples.