Hello! I am a medicine student in the Czech Republic that is working towards being a neurologist. But connecting my work with being a part-time neuroscientist interests me also. Which is something big hospitals appreciate. And my university offers a lot of opportunities for students to get into science. Plus doing a student research is a very interesting way how to get the doctors from the hospital to know you. This can help when you apply for a job there. So I would love to take up the challenge and start working on something next year or maximally in 2 years. In the meantime, I want to prepare as best as I can.
I need to prepare from scratch. The list of tasks I need to complete before I start is long. Starting with learning about how to actually write a research; refreshing my statistics knowledge and skills; learning how to work with EEG data and which tools would benefit my research; and last but not least research the topics I could study and master the theory of the problem I choose to work on.
But why I am asking for your help here. I’ve been trying multiple tools to work with EEG. However, I found MNE-Python to be the best tool among the ones I tried. I have basic skills in programming, but I am a fast learner. I would like to ask you, if you could recommend me skills to master in MNE-python and python in general, which I could use in my research. I downloaded some EEG samples in .csv format and I learned to view them in the MNE-Python viewer, label them with the right electrode names and filter them. It isn’t much, but that’s a solid start, I guess. If you could recommend me useful skills to learn that would be appropriate for a student research, I would be very grateful!
Thank you!
I would recommend starting with the work of Mike X Cohen, who teaches statistics and EEG analysis using Python programming. Not specifically using MNE Python, but the concepts that he presents are applicable in general: https://www.youtube.com/@mikexcohen1
Going through the tutorials in the MNE-python website. Use first the data that are given to you by default in the tutorials, and then apply the same analysis to your own data. Play with the parameters. Try to visualize everything that you compute to get a deeper understanding.
For questions that you have, search in this forum. If you don’t find the answer, feel free to ask. You can also discuss your questions during the MNE Online Office Hours
Thank you also! That’s what I am doing slowly. But I was thinking about if anyone can give me tips of what the workflow of a potential student project could contain. Because there is so much to learn, that I feel overwhelmed right now, since I study EEG basics at the same time. It’s like jumping into the water without knowing how to swim. And I feel like most of the stuff I would use mainly if I get deeply into neurosciences, which is far away from me right now, I am trying to start with an adequate challange.
I understand it might be too overwhelming. To help you narrow things down, you can start with the typical EEG workflow to compute an ERP, which based on my knowledge is:
Load raw data file and channel location
Rereferencing (if necessary)
Line noise removal, Filter data (if necessary)
Interpolating bad channels (if necessary)
Removing bad data segments (if necessary)
Removing eye blinks, muscle artifacts - ICA is a typical algorithm you can use (if necessary)
Define trials (even though I usually do this step right after loading the data)
Baseline correct data (if necessary)
Average across trials for each channel to commute ERPs
Plot ERPs in as a topography or as a timeseries
Notice that I add a lot of ‘‘if necessary’’ comments. In general each dataset is different so I suggest understanding each step and not applying it ‘‘blindly’’. There is an interesting paper stating that EEG data is better left alone, without too much preprocessing (see (1))
Thank you so much, that is the best advice I have ever gotten since I started doing research about my plans. And since I want my future research almost certainly to be about analyzing biomarkers in eegs, the most important aspects will be being able to analyze Power Spectral Density and Event Related Potencials. So I will work on learning how to work with these. Thank you!
