Parkinson’s disease detection based on 3D Convolutional Neural Network and DaTscan imaging
Year: 2025
Type: conference-paper
Abstract: The diagnosis and the detection of Parkinson’s disease remains a complex challenge, particularly in the early stages where symptoms are subtle. Traditional methods, although valuable, may have limitations in terms of accuracy and time efficiency. To address these challenges, this paper proposes deep learning-based detection model capable of automatically analyzing medical images to assist healthcare professionals in diagnosis. We proposed a 3D-CNN model using SPECT DaTscan images from the PPMI dataset. Following rigorous data preprocessing steps including background intensity removal, patch-wise denoising, and cropping centered on the striatum. Our model outperformed previously published studies and pretrained architectures such as 3D VGG, 3D ResNet, 3D DenseNet, S3D, and MedicalNet, achieving an accuracy of 99.53%, with 99.50% precision, 99.50% recall, and a 99% F1-score.
Source: Procedia Computer Science
Language: English
Subfield: Neurology
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W4416640773