Research Article
Enhancing Parkinson's Disease Detection with a GAN-CNN Hybrid Dual-Stream Model
@INPROCEEDINGS{10.4108/eai.28-4-2025.2357835, author={E. Anandaperumal and Azhagiri Mahendiran and M.B. Abhishek and R. Kaavya}, title={Enhancing Parkinson's Disease Detection with a GAN-CNN Hybrid Dual-Stream Model}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part I}, publisher={EAI}, proceedings_a={ICITSM PART I}, year={2025}, month={10}, keywords={parkinson's disease neurodegenerative disorder movement disorder clinical diagnosis mri scan dat scans cnn-gan hybrid model early stage detection yolo ensemble net swin transformer accuracy (85\% 98\%) f1 scor gan-cnn hybrid dual stream}, doi={10.4108/eai.28-4-2025.2357835} }- E. Anandaperumal
Azhagiri Mahendiran
M.B. Abhishek
R. Kaavya
Year: 2025
Enhancing Parkinson's Disease Detection with a GAN-CNN Hybrid Dual-Stream Model
ICITSM PART I
EAI
DOI: 10.4108/eai.28-4-2025.2357835
Abstract
Parkinson's disease is a chronic and progressive neurodegenerative disorder that affects movement. The clinical diagnosis for Parkinson's disease is made through neurological examination and imaging techniques like MRI and a DaT scan. So, the MRI Scan is used in this approach. Traditional diagnostic approaches using MRI data often encounter challenges, including data scarcity, low image quality, and model overfitting. To overcome challenges, this approach uses GAN and CNN. GAN handles generating synthetic images and integrates with real images. The combined data is used as input for the CNN classifier for training, this model is named the GAN-CNN hybrid dual stream model. This model is capable of detecting the early stage of Parkinson’s and is more suitable than the other existing models, such as MobileNet, YOLOv7, and 1D-CNN. By leveraging synthetic data generation and deep learning classification, this model demonstrates improved performance on Parkinson's disease detection and performs well compared to existing systems, and the application of this model extends to real-world diagnostics, potentially enabling early and accurate detection of Parkinson's disease, thereby improving patient outcomes and treatment strategies.
