Deep learning to differentiate parkinsonian disorders separately using single midsagittal MR imaging: a proof of concept study
Related Experiment Video
Updated: Jan 22, 2026
Author Spotlight: Enhancing PSC-to-Functional Cell Differentiation Using ML Models Based on Live-Cell Bright-Field Imaging
Published on: October 4, 2024
Shigeru Kiryu1, Koichiro Yasaka2, Hiroyuki Akai2
1Department of Radiology, International University of Health and Welfare Hospital, 537-3, Iguchi, Nasushiobara, Tochigi, 329-2763, Japan. kiryu-tky@umin.ac.jp.
Objectives:
To evaluate the diagnostic performance of deep learning with the convolutional neural networks (CNN) to distinguish each representative parkinsonian disorder using MRI.
Methods:
This clinical retrospective study was approved by the institutional review board, and the requirement for written informed consent was waived. Midsagittal T1-weighted MRI of a total of 419 subjects (125 Parkinson's disease (PD), 98 progressive supranuclear palsy (PSP), and 54 multiple system atrophy with predominant parkinsonian features (MSA-P) patients, and 142 normal subjects) between January 2012 and April 2016 was retrospectively assessed. To deal with the overfitting problem of deep learning, all subjects were randomly divided into training (85%) and validation (15%) data sets with the same proportions of each disease and normal subjects. We trained the CNN to distinguish each parkinsonian disorder using single midsagittal T1-weighted MRI with a training group to minimize the differences between predicted output probabilities and the clinical diagnoses; then, we adopted the trained CNN to the validation data set. Subjects were classified into each parkinsonian disorder or normal condition according to the final diagnosis of the trained CNN, and we assessed the diagnostic performance of the CNN.
Results:
The accuracies of diagnostic performances regarding PD, PSP, MSA-P, and normal subjects were 96.8, 93.7, 95.2, and 98.4%, respectively. The areas under the receiver operating characteristic curves for distinguishing each condition from others (PD, PSP, MSA-P, and normal subjects) were 0.995, 0.982, 0.990, and 1.000, respectively.
Conclusion:
Deep learning with CNN enables highly accurate discrimination of parkinsonian disorders using MRI.
Key Points:
• Deep learning convolution neural network achieves differential diagnosis of PD, PSP, MSA-P, and normal controls with an accuracy of 96.8, 93.7, 95.2, and 98.4%, respectively. • The areas under the curves for distinguishing between PD, PSP, MSA-P, and normality were 0.995, 0.982, 0.990, and 1.000, respectively. • CNN may learn important features that humans not notice, and has a possibility to perform previously impossible diagnoses.
Insights
Deep learning using convolutional neural networks (CNNs) accurately distinguishes Parkinson's disease (PD), progressive supranuclear palsy (PSP), and multiple system atrophy (MSA-P) from normal controls using MRI scans. This AI approach shows high diagnostic performance, aiding in the differential diagnosis of parkinsonian disorders.
Area of Science:
- Neurology
- Artificial Intelligence
- Medical Imaging
Background:
- Parkinsonian disorders, including Parkinson's disease (PD), progressive supranuclear palsy (PSP), and multiple system atrophy with predominant parkinsonian features (MSA-P), present overlapping clinical symptoms.
- Accurate differential diagnosis is crucial for appropriate patient management and treatment.
Purpose of the Study:
- To evaluate the diagnostic performance of deep learning, specifically convolutional neural networks (CNNs), in differentiating between PD, PSP, MSA-P, and normal subjects using MRI.
- To assess the potential of CNNs to identify subtle imaging features not readily apparent to human observers.
Main Methods:
- A retrospective analysis of midsagittal T1-weighted MRI scans from 419 subjects (125 PD, 98 PSP, 54 MSA-P, 142 controls) was performed.
- Subjects were randomly divided into training (85%) and validation (15%) datasets to train a CNN model for distinguishing between the conditions.
- The diagnostic performance of the trained CNN was assessed on the validation dataset.
Main Results:
- The CNN achieved high diagnostic accuracies: 96.8% for PD, 93.7% for PSP, 95.2% for MSA-P, and 98.4% for normal subjects.
- Areas under the receiver operating characteristic curves (AUCs) were excellent, ranging from 0.982 to 1.000, indicating strong discriminative power.
- The model demonstrated robust performance in classifying subjects into their respective diagnostic categories.
Conclusions:
- Deep learning with CNNs provides a highly accurate method for the differential diagnosis of parkinsonian disorders using MRI.
- CNNs can potentially identify critical imaging biomarkers, enabling diagnoses that may be challenging for human interpretation.
- This AI-driven approach holds promise for improving the diagnostic accuracy and efficiency in neurodegenerative disease assessment.
Related Concept Videos
00:52Social Proof
01:19Self-Concept
Infancy and Emerging Recognition
During infancy, self-concept is virtually nonexistent. Babies do not distinguish themselves as separate entities and often mistake their...
01:24Concepts and Prototypes
The brain organizes this information using concepts, which are mental categories grouping linguistic data,...
02:56Formula Mass and Mole Concepts of Compounds
02:18Intrinsically Disordered Proteins
01:20Understanding Self-Concept