EEG analysis of Parkinson's disease patients treated with 50 Hz repetitive transcranial magnetic stimulation
Kaarna, Esa (2015)
2015
Master's Degree Programme in Biomedical Engineering
Tieto- ja sähkötekniikan tiedekunta - Faculty of Computing and Electrical Engineering
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2015-09-09
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201509151580
https://urn.fi/URN:NBN:fi:tty-201509151580
Tiivistelmä
Repetitive trancranial magnetic stimulation (rTMS) has shown promising results in treating Parkinson's disease (PD), but the best frequencies and parameters have not yet been established. The objective of this study was to determine whether 50 Hz rTMS treatment had an effect on PD patients studying the recorded EEG data before and after the treatment. A novel automatic artifact rejection (AAR) algorithm was also tested with the data analysis. The hypothesis was that the power spectral density of EEG beta band increases after the treatment.
The amount of patients eventually narrowed down to eight (N=8). The complete dataset was around 20 minutes of bipolar 21-channel EEG signal (pre and post) from each patient. EEG was recorded only for safety purposes in the clinical study, so the data was quite filled with artifacts. The AAR algorithm tried was a novel F-wICA algorithm which automatically identifies EEG containing EOG or EMG artifacts based on fractal dimensions. After the artifact rejection the data was divided into power spectral densities of different EEG bands and channels using FFT. The results were then combined and studied with Wilcoxon signed rank test.
Before the actual results, the use of AAR algorithm was assessed and it was found out not to be ready and could not be used for the actual analysis. The results of statistical analysis with only manually rejected data showed that with null hypothesis only one of 12 possible changes of interest (2 Beta Bands and 6 EEG channels) had a significant change (< 0.05) into the expected direction. That alone and the fact that the EEG data was filled with artifacts led to the conclusion that there was no notable difference of EEG activity after the rTMS treatment.
Further research should be made by focusing more on the EEG recording. It has been shown to indicate the affect of the TMS treatment better the behavioral studies. Results are also quantitative so the comparison would be easier in the future.
The amount of patients eventually narrowed down to eight (N=8). The complete dataset was around 20 minutes of bipolar 21-channel EEG signal (pre and post) from each patient. EEG was recorded only for safety purposes in the clinical study, so the data was quite filled with artifacts. The AAR algorithm tried was a novel F-wICA algorithm which automatically identifies EEG containing EOG or EMG artifacts based on fractal dimensions. After the artifact rejection the data was divided into power spectral densities of different EEG bands and channels using FFT. The results were then combined and studied with Wilcoxon signed rank test.
Before the actual results, the use of AAR algorithm was assessed and it was found out not to be ready and could not be used for the actual analysis. The results of statistical analysis with only manually rejected data showed that with null hypothesis only one of 12 possible changes of interest (2 Beta Bands and 6 EEG channels) had a significant change (< 0.05) into the expected direction. That alone and the fact that the EEG data was filled with artifacts led to the conclusion that there was no notable difference of EEG activity after the rTMS treatment.
Further research should be made by focusing more on the EEG recording. It has been shown to indicate the affect of the TMS treatment better the behavioral studies. Results are also quantitative so the comparison would be easier in the future.