Transfer Learning from Pre‐trained 3D Models for Enhanced Parkinson's Disease Detection
Our transfer learning approach demonstrated excellent performance in PD detection, achieving 94.03% accuracy and 100% sensitivity. The significant improvement over the baseline model (75% accuracy) supports the effectiveness of utilizing pre-trained weights from diverse medical imaging tasks. These results establish a promising framework for developing reliable computer-aided diagnosis systems of PD and neurodegenerative diseases using MRI data.
Authors
- K. -M. Ng
- Tanvi Verma
- Lok Yee Tweety Au
- Simon Kang Seng Ting (ORCID: https://orcid.org/0000-0003-4852-9790)
- Adeline Su Lyn Ng
- Shahul Hameed
- Eng King Tan
- Fei Gao
- Siow Mong Rick Goh
- Yong Liu
- Mingrui Tan
- Louis CS Tan
- Jia Huang
- Yuting Song
Institutions
- Agency for Science, Technology and Research (SG)
- Institute of High Performance Computing (SG)
- National Neuroscience Institute (SG)
Publication Details
- Journal
- Alzheimer s & Dementia
- Published
- 2025-12-01
- DOI
- https://doi.org/10.1002/alz70856_101856
- Primary Topic
- Parkinson's Disease Mechanisms and Treatments
- Type
- article
- Field-Weighted Citation Impact
- 0.00