Indian Statistical Institute, Calcutta

Keywords Information Theory, Neuroscience

Role Summer Intern

Timeline May 2020 - Aug 2020

Compared the consistency and effectiveness of Normalised Transfer Entropy (NTE), Transfer Entropy (TE), Granger Causality (GC), Partial Directed Coherence (PDC) metrics
Graph Edit Distance (GED) between NTE and baseline (TE, GC) graphs are 0.45 and 0.30 respectively, whereas for PDC they are 0.54 and 0.45. This implies that NTE shows more correlation with baselines than other metrics
Information flow graphs for left and right motor tasks using NTE showed the least deviation of 0.05 across the subjects, when compared to TE, GC, PDC with deviations of 0.18, 0.1} 0.06 respectively
Showed that NTE is the most reliable metric to identify effective connectivity with least deviation across subjects
The information flow graphs showed significant activity near the primary motor area of the brain. We also observed predominant activity in the right side of the brain for left motor tasks and vice-versa as expected