Wilcoxon Signed-rank Test Based Phase Lag Identification Of EEG Signals

In brain functional network connectivity analysis,phase synchronization has been effective in detecting regions demonstrating similar dynamics over time.The previously proposed connectivity indices such as phase locking value(PLV),phase lag index(PLI)and weighted phase lag index(wPLI)are widely used.They are,however,influenced by volume conduction or noise.In addition,appropriate thresholds have to be chosen in order to employ them successfully,which leads to uncertainty.In this paper,a novel connectivity index named phase lag based on the Wilcoxon signedrank test(PLWT)is proposed under the framework of Wilcoxon signed-rank test,which avoids using thresholds to identify effective connections.In order to verify the superiority of PLWT,experiments based one simulated dataset and two real EEG datasets were designed.First,we analyzed and compared PLWT with previous indices under the influence of volume conduction,noise and sample size bias based on simulated dataset.Second,the test-retest reliability was predicted and compared between PLWT and other indices,which shows that PLWT has high test-retest reliability while performing on real EEG data.Third,we tested the usability of PLWT based on the resting state EEG data of mathematically gifted adolescents.The PLWT-based brain functional networks displayed scale-free character.And it revealed reliable and distinct brain topology for mathematically gifted adolescents compared with control group.All the experimental results can reach the conclusion that PLWT can be utilized as a reliable and convincing measure to reveal true connections while effectively diminishing the influence of volume conduction and noise.When processing real data,useful graph measures can be extracted from PLWT-based brain functional networks and used as effective features to distinguish special group of people.