Related Experiment Video
Updated: Aug 22, 2025
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
Multi-stage classification of Alzheimer's disease from 18F-FDG-PET images using deep learning techniques
Mahima Thakur1, U Snekhalatha2
1Department of Electronics and Communication Engineering (Specialization in Biomedical Engineering), College of Engineering and Technology, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India.
This study developed a deep learning framework using 18F-FDG PET brain scans to accurately detect dementia stages, including Alzheimer's disease (AD) and Mild Cognitive Impairment (MCI). The model achieved high accuracy, offering a promising tool for early diagnosis.
Area of Science:
- Neuroimaging
- Artificial Intelligence
- Medical Diagnostics
Background:
- Dementia diagnosis relies on clinical assessment and neuroimaging.
- Early detection of Mild Cognitive Impairment (MCI) and Alzheimer's disease (AD) is crucial for timely intervention.
- 18F-FDG PET scans provide metabolic information about brain function.
Purpose of the Study:
- To implement a convolutional neural network (CNN) framework for dementia detection using 18F-FDG PET brain imaging.
- To classify multiple dementia stages: Cognitively Normal (CN), Early Mild Cognitive Impairment (EMCI), Late Mild Cognitive Impairment (LMCI), and Alzheimer's disease (AD).
- To evaluate the framework's performance using various metrics and feature maps.
Main Methods:
- Utilized 18F-FDG PET brain imaging data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) repository.
- Employed the ResNet50V2 model for feature extraction, fine-tuning final convolutional layers for multi-classification.
- Applied multiple metrics and analyzed feature maps for model evaluation.
Main Results:
- The multi-classification model achieved an overall accuracy of 98.44% and an Area Under the Receiver Operating Characteristic Curve (AUC) of 95% on the testing set.
- The framework demonstrated high diagnostic performance with an inclusive sensitivity of 94% and specificity of 95%.
- Feature map analysis provided insights into the model's decision-making process.
Conclusions:
- The developed CNN framework effectively classifies different stages of dementia, including subtypes of MCI and AD, from 18F-FDG PET images.
- This AI-driven approach shows significant potential for accurate and early dementia diagnosis.
- The study highlights the utility of deep learning in analyzing neuroimaging data for neurological disorders.
