This is my last week in Kiel. On Tuesday all data, software, and notes related to my project were handed in. Anyone continuing to measure for Steady-State Visually Evoked Potentials and/or Fields in Kiel will be able to pick up from where I left off and continue research that may hopefully lead to a functional BCI using MEG signals.
I will continue work on my Masters Thesis back in the United States and plan to finish within the next few months. I've already written and learned quite a bit on thesis from my time in Kiel and I see very clearly what needs to be done next. Everyone in the Medical Signal Processing lab was wonderful and I will have fond memories of my many discussions with Prof. Dr. Schmidt.
Living in Germany has been an interesting and insightful experience with several people I've met that will stay with me the rest of my life. I wish everyone that I've met in the Medical Signal Processing Lab and the student dorms the best in life and continued success.
This week I continued working on my thesis, cleaning up figures and code, and gave a presentation at the Science Slam on February 20 on my analysis from last week.
EEG data was gathered on February 6 with an epoch length of 750ms and other parameters as follows:
- Trials 1-6 were single epochs saved using a 7.5 Hz trigger signal and the checkerboard pattern flickering 2 frames/trigger period
- Trials 7-12 were single epochs saved using a 15 Hz trigger signal and the checkerboard pattern flickering 1 frame/trigger period
- Trials 13-14 were averaged in real time with a 1.9 Hz trigger signal and the checkerboard pattern flickering 2 frames/trigger period
- Trials 15-16 were averaged in real time with a 10 Hz trigger signal and the checkerboard pattern flickering 2 frames/trigger period
- Trials 17-18 were averaged in real time with a 10 Hz trigger signal and the checkerboard pattern flickering 1 frame/trigger period
When analyzing the data, it was very clear even in the time domain that the 1 frame/trigger period were sinusoidal and similar with each other while the averaging the data in real time with 2 frames/trigger period led to a more random signal.
With trials 1-12, taking epochs and concatenating became better indicators rather than averaging the signals with a total length of only 750 ms. Power spectral density and spectrogram analyses are shown: