The Effect of Auditory Stimuli on the Quantitative Electroencephalogram in Patients with Parkinson's Disease
Parkinson’s Disease (PD) is the second most common neurodegenerative disorder worldwide with increasing incidence and prevalence. It mainly affects the motor system due to a loss of dopaminergic neurons in the substantia nigra and leads to cardinal symptoms including brady-/akinesia, tremor, muscle...
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Beteiligte: | |
| Format: | Dissertation |
| Sprache: | Englisch |
| Veröffentlicht: |
Philipps-Universität Marburg
2023
|
| Schlagworte: | |
| Online-Zugang: | PDF-Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Zusammenfassung: | Parkinson’s Disease (PD) is the second most common neurodegenerative disorder worldwide with increasing incidence and prevalence. It mainly affects the motor system due to a loss of dopaminergic neurons in the substantia nigra and leads to cardinal symptoms including brady-/akinesia, tremor, muscle stiffness and postural instability. After clinical diagnosis, treatment is primarily based on L- dopa, dopamine agonists and MAO-B inhibitors. Even with therapy, PD continues to progress and remains uncurable. In recent years, music therapy has been established as a complementary therapy due to a variety of positive effects, mainly on the motor system. However, it is still insufficiently explained what exactly renders music therapy so effective. Possible explanations range from an increased dopamine release to a better functional connectivity of different brain areas.
The aim of this methodologically innovative study was to find underlying mechanisms for the effectiveness of music therapy based on EEG analysis. The analysis of the EEG was chosen due to its good temporal resolution, fast availability and relatively low costs. The research questions were first, whether it is generally possible to distinguish patients with PD from Healthy Controls (HC) based on their EEG. Second, whether auditory stimuli show an effect on the EEG. Third, which features precisely a differentiation of both groups in the EEG is based on. And fourth, which characteristics render an auditory stimulus effective.
The study was conducted in collaboration between the University of British Columbia (UBC) in Vancouver and the Philipps-Universität Marburg. In 2017 and 2018, 12 patients with PD and 4 age- matched HC were tested at the UBC campus. A total of 5 EEGs (conditions) were recorded from each subject at rest and under auditory stimulation. The three stimuli differed in complexity (Rain vs Spring Walk) and modulation (rhythmic and non-rhythmic). For a more precise interpretation of the results, natural sounds were used as stimuli instead of music. Due to the amount of data, a custom-made pattern recognition algorithm (Support Vector Machine) was used, distinguishing both groups through a hyperplane within a high-dimensional feature space. Redundant data was removed in advance by calculating the mutual information quotient to include only relevant data in the final analysis.
It could be shown that, first, the differentiation of both groups on the basis of the EEG is generally possible, in this case even with a convincing classification accuracy of up to 90 %. Second, the auditory stimuli mainly had an effect on the EEG samples of HC and made the classification more complex: the EEG samples of the HC approached those of the PD patients within the feature space, rendering a common hyperplane for all conditions ineffective. Based on shared features but with a separate hyperplane in each condition classification accuracy of 80-90 % and thus very good discrimination of both groups could be achieved again even under the influence of auditory stimuli.
68
Third, the by far most important features to distinguish both groups were related to the delta frequency band (0.5-4 Hz) including band power, indices of the delta band, and harmonic parameters. The increased importance of delta in PD matches existing literature, most likely due to cognitive decline. This study enhances existing literature on delta by the harmonic parameters, mainly the center frequency and the spectral value thereof. In addition, the delta frequency band is often linked to relaxation and sleep. Thus, the convergence of the EEG samples is most likely explained by stimulus- induced relaxation. Another important feature seems to be the phase lag index. It is also mentioned in the literature as an indicator of mild cognitive impairment and decreases under the influence of the stimuli. A link between the PLI and functional connectivity, as mentioned in the literature, cannot be shown in this study.
Fourth, the convergence of the HC samples towards the PD samples was particularly evident in the rain conditions with misclassifications of up to 80 %. This was the case in both the rhythmic and non- rhythmic variants. Given the importance of rhythm as often shown in literature on music therapy it appears that the intended modulation was not perceived as rhythmic by the subjects. The convergence of samples was less evident in the spring walk condition, where higher frequency bands were relevant too. Auditory stimuli thus seem to need a basic complexity to show an effect on the EEG.
Approaches to further research arise. For example, if the delta band is expected to be important, greater epoch lengths than in this study (3 seconds) could be analyzed to avoid false interpretations due to epochs being too short to capture very slow oscillations. In addition, a general slowing of the EEG is probably not specific for PD. For a more specific analysis, the inclusion of participants with mild cognitive impairment (PD-MCI as well as MCI not caused by PD) would be useful. Testing more complex stimuli such as music, an inclusion of motor functions in the analysis or even a measurement of dopamine levels would also remain of interest. Looking at the study design, a more balanced patient population might be beneficial.
In order to show an effect of music therapy in the EEG, a convergence of PD samples towards HC samples would have been desirable. Due to the relaxation, the opposite was the case. The chosen methodology, however, seems very appropriate. The classification of both groups was possible on a convincingly high level and recommends this approach for further research, due to its variability beyond neurology and even medicine. |
|---|---|
| DOI: | 10.17192/z2023.0574 |