Classification of Task Evoked fMRI Signals Using Temporal Characteristics of Brain Regions

Activation and inhibition in different regions of the brain depends on the activities of an individual subject. Such temporal and spatial characteristics, while performing predefined functional tasks, are useful to study the variations in healthy population due to aging, as well as disease conditions. In this paper, we propose a methodology to capture activation and inhibition characteristics in order to classify the brain responses corresponding to different tasks using functional Magnetic Resonance Imaging (fMRI) signals. Four different tasks namely, visual word and objects, visual working memory, Simon task and tone counting are used for the current analysis. Anatomical parcellation of the brain regions is done using Hammersmith Atlas. We present a novel feature named Temporal Consistency in Activation Level (TCAL) to capture how consistent a pair of regions are, in maintaining the same activation levels. Such a feature is able to outperform few of the standard approaches, in classifying the tasks, which are based on similarity/dissimilarity measures for activation between the pair of brain regions namely, Euclidean distance, Cross correlation, City block distance and Wasserstein distance. Results further indicate that for each of the tasks, a few brain regions identified by TCAL is consistent with previous studies and is able to provide mechanistic reasoning.