Best Window Width Determination and Glioma Analysis Application of Dynamic Brain Network Measure on Resting-State Functional Magnetic Resonance Imaging

Traditional Resting-State functional Magnetic Resonance imaging (RS-fMRI) analysis takes the whole time sequence as an input to measure brain functional network, which inevitably neglects dynamic modification of brain functional connections. In order to observe the instantaneous change, a sliding window resampling method was proposed to divide entire signal sequence into several sub-sequences before extract network from the signal. However, verification of the method was not completed and a reasonable way to determine window width has not been presented. To confirm the legitimacy, we took brain small-world character as criteria, which is widely agreed to be a critical organizational character of brain functional network and determined a reasonable window width range. The entire signal sequence was first resampled by sliding windows with different widths and brain networks were extracted from individual sub-sequences. An exponential truncated power-law function was then applied to fit the node degree distribution of these networks to evaluate the small-world character as well as the legitimacy of the corresponding window width. Further application of the method showed major discrepancies on glioma patient brain network in different brain regions and dynamic evolution on regional Hub network, compared to those of normal subjects. These discoveries, which physiologically conform to the impact of glioma to normal brain, extensively proved the legitimacy of the resampling method with the window width we determined. This method retains small-world character, discloses instantaneous modification and enables dynamic measure of brain network.