- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- Quantifying motor vigour in Parkinson’s disease using...
Open Collections
UBC Theses and Dissertations
UBC Theses and Dissertations
Quantifying motor vigour in Parkinson’s disease using EEG: a dual workflow of orthogonal multiple instance learning and ICA-based analysis Mahdizadeh, Ailar
Abstract
This thesis investigates the application of linear and non-linear methods for analyzing Electroencephalogram (EEG) signals to quantify motor vigor, particularly in the context of Parkinson’s disease (PD). Traditional EEG analysis methods often struggle with the complexity and variability of EEG data, leading to challenges in accurately interpreting neural activity. This study explores both linear Independent Component Analysis (ICA) and advanced Multiple Instance Learning (MIL) models to enhance the robustness and generalizability of EEG analysis. In the linear analysis, Independent Component Analysis (ICA) is employed to decompose EEG signals into statistically independent components. This method helps in isolating significant neural patterns related to motor vigor. The study also incorporates Canonical Correlation Analysis (CCA) to identify correlations between EEG embeddings and behavioral metrics. For the non-linear analysis, the research introduces an Orthogonal Spatio- Temporal Attention Network, an advanced MIL model that separately handles spatial and temporal features of EEG data. This model improves upon conventional methods by incorporating dual attention pathways, facilitating a more detailed and accurate capture of neural dynamics. The study utilizes a Catch-22 embedding technique to extract key temporal features for the MIL framework. Empirical evaluations demonstrate the efficacy of both linear and nonlinear models in predicting motor vigor metrics, providing reliable tools for clinical assessment and monitoring of PD. The findings underscore the potential of combining linear ICA and non-linear MIL-based approaches in neurophysiological research, offering significant improvements in the analysis and interpretation of complex EEG data.
Item Metadata
| Title |
Quantifying motor vigour in Parkinson’s disease using EEG: a dual workflow of orthogonal multiple instance learning and ICA-based analysis
|
| Creator | |
| Supervisor | |
| Publisher |
University of British Columbia
|
| Date Issued |
2024
|
| Description |
This thesis investigates the application of linear and non-linear methods
for analyzing Electroencephalogram (EEG) signals to quantify motor vigor,
particularly in the context of Parkinson’s disease (PD). Traditional EEG
analysis methods often struggle with the complexity and variability of EEG
data, leading to challenges in accurately interpreting neural activity. This
study explores both linear Independent Component Analysis (ICA) and advanced
Multiple Instance Learning (MIL) models to enhance the robustness
and generalizability of EEG analysis.
In the linear analysis, Independent Component Analysis (ICA) is employed
to decompose EEG signals into statistically independent components.
This method helps in isolating significant neural patterns related to motor
vigor. The study also incorporates Canonical Correlation Analysis (CCA)
to identify correlations between EEG embeddings and behavioral metrics.
For the non-linear analysis, the research introduces an Orthogonal Spatio-
Temporal Attention Network, an advanced MIL model that separately handles
spatial and temporal features of EEG data. This model improves upon
conventional methods by incorporating dual attention pathways, facilitating
a more detailed and accurate capture of neural dynamics. The study utilizes
a Catch-22 embedding technique to extract key temporal features for
the MIL framework.
Empirical evaluations demonstrate the efficacy of both linear and nonlinear
models in predicting motor vigor metrics, providing reliable tools for
clinical assessment and monitoring of PD. The findings underscore the potential
of combining linear ICA and non-linear MIL-based approaches in
neurophysiological research, offering significant improvements in the analysis
and interpretation of complex EEG data.
|
| Genre | |
| Type | |
| Language |
eng
|
| Date Available |
2024-09-03
|
| Provider |
Vancouver : University of British Columbia Library
|
| Rights |
Attribution-NonCommercial-NoDerivatives 4.0 International
|
| DOI |
10.14288/1.0445274
|
| URI | |
| Degree | |
| Program | |
| Affiliation | |
| Degree Grantor |
University of British Columbia
|
| Graduation Date |
2024-11
|
| Campus | |
| Scholarly Level |
Graduate
|
| Rights URI | |
| Aggregated Source Repository |
DSpace
|
Item Media
Item Citations and Data
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International