Hi! I am working with motor imagery dataset using mne, trying to calculate the ERDS ratio; I have used mne script from the website for ERDS for single participant, however, I am getting outliers (values are too high not sure why), I want to calculate the ratio for multiple participants, so trying to…

Hi, I am not exactly sure I understand what you are trying to achieve but here as some thoughts. Regarding multiple subjects. I am not sure whether you need an across-subjects average baseline. If not, just loop the same code over subjects. If yes, I think you need to do it manually. Loop the co…

@sotpapad Many thanks. This was really helpful. Actually I am trying to find an average ERD/ERS ratio w.r.to all the subjects. I tried this, this is really helpful. I am stuck at the shape. I took Pref (-1,3.0) and Pevent (-1,9.). Now I have two different powers with different shapes. Pref(40,1,20…

Hi @SSB24 , Glad I could help. Since power evolves over time, I think that what most people would do is take the average during the baseline period and subtract it from every time point in the task period. In other words, you need Pref to be of dimensions (40,1). This way subtraction should work (a…

Hi @sotpapad Many thanks. This actually worked well for me. Event had a dimension of (1,5001) and Pref as (1,) . {I took overall average of trials}. Many Thanks. Just to confirm, and be sure, please do correct me if I am wrong, when I took Hilbert transform, I had the data in real and imaginary pa…

The envelope/magnitude would be abs(Pref) which is equivalent to using apply_hilbert(envelope='True'). Pref ** 2 corresponds to power spectral density (aka squared magnitude). And finally your formula gives you ERDS.

@sotpapad Many thanks . I tried it and getting the following result by plotting erd/ers manually: is this normal? I filter 8-32 Hz, remove bad epochs, and apply all that we discussed before and plot it. this motor imagery is average of few subjects at 0-9 sec window, event shown and happening 4 - 7.…

Hi @SSB24 , Is this normal? Difficult to say. Maybe. I’ve data like this before, but that doesn’t really mean anything. It looks like it’s more noisy than what you would expect, but roughly counting some cycles I’d say it’s in the beta band alright. And, at least, there seems to be a ERD-like trend …

@sotpapad I tried removing bad epochs first and then filter and reduced time points from 2 to 8 second, I am taking baseline of 2.0 to 3.5 as Prest now and Pevent is 2.0 to 8.0 . I apply hillbert transform to get erd/ers and used envelope to get average. I get the following graph for channel C3: […

@sotpapad this is for C3 and C4 together for one subject thinking left: [image]

Error in ERD / ERS ratios in numbers and manual plotting

Support & Discussions

SSB24 January 19, 2023, 12:37pm 12

@sotpapad I was expecting a graph like this as per literature and they have used same formula and ERD is in % i.e (Prest - Pevent / Prest) *100

Topic		Replies	Views
ERD/ERS when there is only one event Support & Discussions preprocessing , eeg	10	1209	April 13, 2021
EEG preprocessing pipeline for ERP analysis: am I following all the right steps? Support & Discussions preprocessing , eeg , evoked	1	1393	July 7, 2022
Alpha/Beta ratio (PSD) for single epochs EEG for specific electrodes Support & Discussions eeg , epochs	10	2155	December 19, 2021
Suggestions to Reduce Noise in EEG ERP Analysis Support & Discussions evoked	8	302	December 26, 2024
ERD/ERS map tutorial Mailing List Archive (read-only) list-archive	1	209	May 24, 2020

Error in ERD / ERS ratios in numbers and manual plotting

Related topics