[FieldTrip] Interpretation of Granger spectrum and link with directed coherence

Simon Van Eyndhoven Simon.VanEyndhoven at esat.kuleuven.be
Fri May 19 13:51:10 CEST 2017

Dear FieldTrip community,

As a novice in EEG connectivity analysis, I am trying to reproduce and 
understand the connectivity tutorial of the website 

I was hoping that someone could shed light on the computation of the 
Granger frequency spectrum:

cfg           = [];
cfg.method    = 'granger';
granger       = ft_connectivityanalysis(cfg, mfreq);

More specifically, I would like to know which computations are done to 
obtain the values in granger.grangerspctrm. (Unfortunately, inspecting 
the paper by Brovelli et al. PNAS (2004), as indicated in 
ft_connectivity_granger did not clear this up...)

Moreover, I wonder what the relationship is between the Granger spectrum 
and other metrics such as (partial) directed coherence (DC / PDC) and 
directed transfer function (DTF), which also provide connectivity 
information in the form channels x channels x frequencies?

In the appendix in Van Mierlo et al., "Functional brain connectivity 
from EEG in epilepsy: Seizure prediction and epileptogenic focus 
localization" and in Baccala and Sameshima, "Partial directed coherence: 
a new concept in neural structure determination", it is explained that 
both DC and DTF are computed using the transfer function matrix H(f), 
while PDC is computed using the A(f) matrix. Is this the reason why only 
PDC is termed 'partial', and the others not (even though they are also 
computed in a multivariate framework, i.e. taking all available 
variables into account)?

Although I could not find an answer to my questions in the mailing 
archives, I apologize if this question has been asked before.

Thank you in advance for your help!
Best regards,


More information about the fieldtrip mailing list