Investigation of low frequency drift in attention deficit hyperactivity disorder fMRI Signal

Abstract

This paper analyzes the resting state fMRI signal of 21 ADHD subjects and 27 healthy volunteers, and proposes a novel method for extracting an effective feature in frequency domain. Utilizing this feature, the ADHD subjects and the control persons are classified with an accuracy of 95.83% by support vector machine (SVM). Furthermore, using this method, some specific brain regions such as the right amygdaloid nucleus, the left thalamus, cerebellum and vermis, with high classification accuracies, are relative to the pathological mechanism of ADHD which are consistent with the previous research results.