Vision transformer-equipped Convolutional Neural Networks for automated Alzheimer's disease diagnosis using 3D MRI
Zhen Zhao1, Pauline Shan Qing Yeoh2, Xiaowei Zuo3
1Department of Electrical Engineering, Faculty of Engineering, Universiti Malaya, Kuala Lumpur, Malaysia.
Frontiers in Neurology
|December 31, 2024
Summary
A new AI model, VECNN, accurately diagnoses Alzheimer's disease (AD) using 3D MRI scans. This non-invasive method offers a cost-effective and objective approach for early detection and personalized treatment of AD.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Medical Imaging
Background:
- Alzheimer's disease (AD) is a growing global health concern.
- Current diagnostic methods for AD are often subjective and costly.
- Accurate and early diagnosis is crucial for effective Alzheimer's disease management.
Purpose of the Study:
- To develop a novel AI model for improved Alzheimer's disease diagnosis.
- To enhance diagnostic accuracy using 3D magnetic resonance imaging (MRI).
- To provide a non-invasive, cost-effective, and objective diagnostic tool.
Main Methods:
- A Vision Transformer-equipped Convolutional Neural Network (VECNN) was developed.
- The VECNN model was trained and validated using the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset.
- The dataset included 2,248 3D MRI images from patients with Alzheimer's disease, healthy controls (HC), and mild cognitive impairment (MCI).
Main Results:
- The VECNN model achieved high diagnostic performance.
- Accuracy: 92.14%
- Precision: 86.84%, Sensitivity: 93.27%, Specificity: 89.95% in differentiating AD, HC, and MCI.
Conclusions:
- The VECNN model demonstrates significant potential as a valuable clinical tool for Alzheimer's disease diagnosis.
- This AI-driven approach offers a non-invasive, objective, and cost-effective diagnostic alternative.
- The findings support advancements in early Alzheimer's disease detection and personalized therapeutic strategies.
