Page 28 - AIH-2-4

Artificial Intelligence in Health

ORIGINAL RESEARCH ARTICLE
Accurate early detection of Parkinson’s disease from

single photon emission computed tomography
imaging through convolutional neural networks

R. Prashanth*

Independent Researcher, Bengaluru, Karnataka, India

Abstract

Early and accurate detection of Parkinson’s disease (PD) remains a crucial diagnostic
challenge with substantial clinical implications, particularly for ensuring effective
treatment and patient management. For instance, a group of subjects with scans
without evidence of dopaminergic deficit (SWEDD) who are initially diagnosed
as PD but exhibit normal single photon emission computed tomography (SPECT)
scans. Over time, follow-up assessments often lead to a revised diagnosis of
non-PD. In the meantime, these subjects may receive PD-specific medications
that can cause more harm than benefit. In this paper, a case study is presented in
which machine learning models are developed and trained on SPECT images to
distinguish early PD from healthy controls, as well as to differentiate SWEDD cases
*Corresponding author: from early PD. The case study utilizes a well-known, publicly available dataset and
R. Prashanth explores several machine learning classifiers, including support vector machines,
(prashanth.r.iitd@gmail.com)
logistic regression, feed forward neural networks, and convolutional neural
Citation: Prashanth R. Accurate networks (CNNs). The CNN model gave the best performance in differentiating PD
early detection of Parkinson’s
disease from single photon from healthy subjects. All these models demonstrated strong potential for early
emission computed tomography differentiation of SWEDD cases from PD. These results suggest that the proposed
imaging through convolutional approach could support clinicians in making more accurate and timely diagnostic
neural networks. Artif Intell Health.
2025;2(4):22-32. decisions.
doi: 10.36922/AIH025040005
Received: January 21, 2025 Keywords: Computer-aided diagnosis; Machine learning; Deep learning; Parkinson’s
1st revised: May 13, 2025 disease; Medical imaging
2nd revised: May 22, 2025
Accepted: May 30, 2025
Published online: June 17, 2025 1. Introduction
Copyright: © 2025 Author(s).
This is an Open-Access article Parkinson’s disease (PD) is a progressive neurodegenerative disorder affecting millions
distributed under the terms of the of people worldwide and is characterized by the loss of dopaminergic neurons in the
Creative Commons Attribution substantia nigra. Its prevalence increases with age, impacting approximately 1%
1,2
License, permitting distribution, 3
and reproduction in any medium, of individuals over 60 years. The clinical diagnosis of PD is challenging as there are
provided the original work is no definitive diagnostic tests and the diagnosis is based on the presence of cardinal
properly cited. symptoms, such as tremor at rest, rigidity, and bradykinesia, along with a subject’s
1
Publisher’s Note: AccScience response to PD medications. However, these symptoms appear in the later stages of the
Publishing remains neutral with disease and by the time the patient manifests these symptoms, the patient might have
regard to jurisdictional claims in 4
published maps and institutional already crossed the early stage of the disease. Early detection of PD is important because
5
affiliations. appropriate targeted therapies could be initiated before any major deterioration occur.
Volume 2 Issue 4 (2025) 22 doi: 10.36922/AIH025040005