ds004504/sub-009 (Scott)

sccn / NEMAR-pipeline

5 stars 1 forks source link

ds004504/sub-009 (Scott) #73

Closed arnodelorme closed 5 months ago

arnodelorme commented 6 months ago

here in the histograms, though 17 of 19 channels were retained and decomposed, the (null) bar heights are 16 or 15.5 !?!?
- also, note that although ICLabel finds 2 two ICs (IC2 and ?IC8?) - the histogram shows 15.5 non-eye ICs and 0 Eye ICs !?!? Please fix these things. Also, make the xticklabels in these histogram plots 0 20 40 60 80 100
- Also, increase the ticklabel and axislabel (and legend label) fontsize.

arnodelorme commented 6 months ago

You have to set the labels manually.

set(gca,'XScale','log')

Also look at the tftopo.m code that can plot in log scale.

dungscout96 commented 5 months ago

Replotted in log scale. @arnodelorme EEG.icawinv has dimension 17x16. I'm not sure why's the case. Arno can you help?

arnodelorme commented 5 months ago

@dungscout96 EEG.icawinv can have 16 components (17 channels) when the data is rank deficient (due to average reference for example). My guess is that the function detected that the data was rank deficient.

dungscout96 commented 5 months ago

After the rerun the data seems to no longer be rank deficient. Maybe it before it still had the avg reref step but now it doesn't. ICLabel histogram shows 17 components now