EcoSta 2018: Registration
View Submission - EcoSta2018
Title: Extended likelihood approach to brain connectivity analysis Authors:  Donghwan Lee - Ewha Womans University (Korea, South) [presenting]
Youngjo Lee - Seoul National University (Korea, South)
Abstract: Conventional multiple testing procedures are commonly used for testing of brain connectivity. However, they are often based on assumptions of independence, so can distort conclusions in brain connectivity analysis. We introduce a hierarchical random effect model for brain connectivity analysis by incorporating a proper correlation structure of test statistics. Based on the extended likelihood approach, we show that the proposed method can provide an accurate estimation of the false discovery rate numerically, and outperforms the other existing methods in terms of validity of error control and power. A real neuroimaging data example for comparing connectivity in two groups are illustrated. We found that an appropriate model is important for the efficiency of connectivity tests.