[FieldTrip] What exactly does 'minnbchan' refer just to spatial neighbors?

David Groppe dgroppe at cogsci.ucsd.edu
Mon Feb 7 16:48:40 CET 2011


Hi fieldtrippers,
   I'm a little bit confused about the meaning of the cluster-based
permutation test parameter 'minnbchan'.  I understand the rationale
for the parameter, but based on the tutorial description:

"The value of cfg.minnbchan  is a tuning parameter that determines the
way the clusters are formed. More specifically, we use cfg.minnbchan
to specify the minimum number of neighborhood channels that is
required for a selected sample (i.e., a sample who's T-value exceeds
the threshold) to be included in the clustering algorithm. With
cfg.minnbchan = 0 (the default), it sometimes happens that two
clusters are spatially connected via a narrow bridge of samples.
Because they are connected, these two clusters are considered as a
single cluster. If clusters are interpreted as reflecting spatially
distinct sources, such a combined cluster does not make much sense. To
suppress this type of combined clusters, one can choose to ignore all
selected samples (on the basis of their T-values) if they have less
than some minimum number of neighbors that were also selected. This
minimum number is assigned to cfg.minnbchan. This number must be
chosen independently of the data."

I can't tell if "minimum number of neighbors" means just spatial
neighbors (i.e., nearby sensors) or if includes temporal and frequency
neighbors as well.  Could someone clarify?

Also, just to be clear, if I want to be able to detect an effect that
I think is at only a single electrode, I should set minnbchan to 0,
yes?  If understand the parameter correctly, any greater value of
minnbchan will ignore clusters that are restricted to a single
electrode.
    thanks for your help,
        -David


-- 
David Groppe, Ph.D.
dgroppe at cogsci.ucsd.edu
http://www.cogsci.ucsd.edu/~dgroppe/



More information about the fieldtrip mailing list