Classification of Spatio-Temporal fMRI Data in the Spiking Neural Network

Abstract

Deep learning machine that employs Spiking Neural Network (SNN) is currently one of the main techniques in computational intelligence to discover knowledge from various fields. It has been applied in many application areas include health, engineering, finances, environment, and others. This paper addresses a classification problem based on a functional Magnetic Resonance Image (fMRI) brain data experiment involving a subject who reads a sentence or looks at a picture. In the experiment, Signal to Noise Ratio (SNR) is used to select the most relevant features (voxels) before they were propagated in an SNN-based learning architecture. The spatiotemporal relationships between Spatio Temporal Brain Data (STBD) are learned and classified accordingly. All the brain regions are taken from data with label star plus-04847-v7.mat. The overall results of this experiment show that the SNR method helps to get the most relevant features from the data to produced higher accuracy for Reading a Sentence instead of Looking a Picture.