Westerhoek, Luuk (2025) Feature extraction methods for 3D FDG-PET brain scans for the diagnosis of neurodegenerative disorders. Bachelor's Thesis, Computing Science.
|
Text
bCS2025WesterhoekLR.pdf Download (9MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
akkoord westerhoek.pdf Restricted to Registered users only Download (188kB) |
Abstract
This thesis presents a comparative analysis of feature extraction methods for neurodegenerative disease diagnosis using 3D fluorodeoxyglucose positron emission tomography (FDG-PET) brain imaging. We compare Principal Component Analysis (PCA) and Region-of-Interest (ROI) aggregation, using 10-fold stratified cross-validation for consistent comparison between the methods. Additionally, we investigate the reconstruction accuracy of both methods and between different atlases. For the ROI method, we also examine the disease-specific regions to determine optimal atlas selection for specific diseases. Results show PCA's great performance in information retention, through reconstruction, and classification performance. For ROI aggregation, despite higher reconstruction errors, shows comparable or even slightly better classification results than PCA. This result indicates that anatomical dimensionality reduction can very well capture metabolic patterns even with significant information loss.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Biehl, M. and Bunte, K. and Lovdal, S.S. |
| Degree programme: | Computing Science |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 27 Aug 2025 07:14 |
| Last Modified: | 27 Aug 2025 07:14 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/36867 |
Actions (login required)
 |
View Item |