Identification of brain activity by fractal scaling analysis of functional MRI data

Abstract

Functional magnetic resonance imaging (fMRI) is a powerful tool for studying brain function, especially related to disease and aging. One of the major tasks of fMRI data analysis is to find a few specific regions involved in certain functionality by studying huge but noisy 3-dimensional spatial plus 1-dimensional temporal data. Therefore, developing simple and reliable signal/image processing algorithms for fMRI data analysis is very important. In this paper, we systematically study how fractal scaling analysis can help us reliably detect brain activity through fMRI data analysis. We examine two types of fractal analysis, the fluctuation analysis (FA) and detrended fluctuation analysis (DFA). We show that while FA is able to readily distinguish active brain regions from inactive ones, it fails to robustly recognize which active regions in the brain are truly involved in a certain task. On the other hand, we show that DFA is very effective for this task.