Hello Dear all
I have some data from the auditory oddball paradigm, that I have passed some stages like removing noisy components, epoching to trials, and averaging. I would like to know your opinion about these data if I have cleaned them enough. I eagerly accept your suggestions.
figure 1. standard (upper) and deviant ( bottom) averaged trials
figure 2. difference of means ( deviant events - standard events)
It is difficult to address this question as we don't know what the data looked like in the beginning, what was the output at each stage of cleaning, what is the actual SNR of the effects you're expecting to study, what is your expected data quality, etc. And most of all, what "clean EEG signal" should look like depends a lot between labs: some scrub a lot the data so that no artifact can cause a dramatic bias in the ERP, others leave it as unprocessed as possible not to risk removing any of the data of interest.
The ERPs shown here could look OK to me, but I can't tell if they are OK in the context of your experiment and analysis.
From our point of view, our job is accomplished: you know how to fully pre-process and visualize your EEG data with Brainstorm. Our goal is to provide easy to use tools so that you can play a lot with your data (repeat many times the different cleaning steps with slightly different parameters) and interact closely with the signals (many visualization tools that be accessed with a single click).
Now you need to spend some time manipulating your data, getting the feeling for what is the best way to clean it in order to observe better your signal components of interest, iteratively with a lot of trial and errors, at the individual level and at the group level. I don't think any generic advice or fully automated cleaning procedure can replace your personal expertise.
I do appreciate you a lot, Francois. Your point of view was wonderful for me.
