 This paper proposes a new method called multivariate non-parametric dynamical Granger causality, MNDGC, for analyzing time-varying directed electroencephalogram, e.g., networks. The MNDGC method is based on a data-driven approach rather than a model-based one, allowing it to be more robust against noise and to capture the dynamic changes in the network. This makes it suitable for analyzing the dynamics of brain activity during motor imagery tasks. The results show that MNDGC outperforms the Adaptive Direct Transfer Function, ADTF, method in terms of noise resistance and ability to reveal the differences between left and right hand motor imagery tasks. This article was authored by Chan-Lin Yi, Yuan Shou, Wang Jin-Chen, and others.