A Hypothesis Testing for Large Weighted Networks With Applications to Functional Neuroimaging Data

Neuroimaging techniques have been routinely applied in various studies in neuroscience, which contribute to providing novel insights into brain functions. One of the most important and challenging questions related to data collected from such studies is hypothesis testing for the differences between two samples of networks of brain regions. This is due to the fact that networks constructed from neuroimaging studies, which can be weighted and large, are very complex. Focusing on this problem, a novel hypothesis testing procedure is proposed under a general framework for large weighted networks. The asymptotic null distribution is derived and the power guarantee is also provided theoretically. Simulation experiments and practical application to the real brain networks are carried out to demonstrate the effectiveness of the proposed testing method.