Back to Top

A Matlab guide to Phase-Amplitude Coupling

Phase-Amplitude Cross-Frequency Coupling algorithm
Phase-Amplitude Cross-Frequency Coupling algorithm

Serving the spirit of reproducible research, we include a Matlab Guide for single-trial and across-trials Phase-Amplitude Coupling estimator.

We are including the main functions (m-files) and scripts developed for the publication:

"A novel biomarker of amnestic MCI based on dynamic Cross-Frequency Coupling patterns during cognitive brain responses", Frontiers in Neuroscience 2015, edited by S.Dimitriadis & N.Laskaris.

along with a sample single-trial dataset, all the ensemble averages and all the derived PAC features based on which we applied our (machine-learning related) methodology for introducing the new biomarker.

There are three script-files , namely memo1-3, which demonstrate the main steps employed in our analysis. -The first one demonstrates the use of matlab rankfeatures command for estimating the discriminatory power of temporal patterning from the ensemble-average waveforms (provided as a mat-dataset).

- The second one exemplifies the use of PAC-estimator (described via eq.3 in our paper). The across-trial estimator is applied to single-trial responses from a normal subject (separately from ''target'' and ''non-target'' condition ) and the corresponding temporal profiles are compared (reproducing Fig.4 from our paper) .

- The third one exploits the PAC-traces that were computed for all subjects (and made availabe in FV_from_TV_PAC_analysis.mat) so as to select the more ''useful'' features for building an SVM-classifier. Feature selection and classifier application (training + testing) appear as distinct parts. In addition the code for reproducing Fig.7 is also provided.
Our code has been based on previous publications on the topic, which are aknowledged within the core m-file.

You may download the full package here

-Hoping it would become useful to similar neuroinformatic-explorations.