Ensemble of vision transformer architectures for efficient Alzheimer's Disease classification
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.This study introduces an ensemble of Vision Transformers (VTs) for efficient Alzheimer's Disease (AD) classification. The proposed VT framework significantly improves accuracy over traditional ML and CNN models, offering a promising tool for early AD detection.
Area Of Science
- Computer Vision
- Artificial Intelligence
- Neuroscience
Background
- Vision Transformers (VTs) have emerged as state-of-the-art in computer vision, excelling at capturing dependencies and handling class imbalance.
- Alzheimer's Disease (AD) classification remains a challenge, particularly under imbalanced and data-scarce conditions.
Purpose Of The Study
- To propose and evaluate an ensemble framework of VTs for efficient and accurate classification of Alzheimer's Disease (AD).
- To assess the model's performance on imbalanced and data-scarce AD datasets.
Main Methods
- An ensemble framework comprising four vanilla Vision Transformers (VTs) utilizing hard and soft-voting approaches.
- Model testing and validation on the OASIS and ADNI Alzheimer's Disease datasets.
- Comparative analysis against state-of-the-art Convolutional Neural Network (CNN) and Machine Learning (ML) models.
Main Results
- The ensemble VT model achieved a 2% improvement over individual VT models.
- Demonstrated superior performance with accuracy gains of 4.14% over ML models and 4.72% over CNN models.
- Effective performance under imbalanced and data-scarce conditions using the ADNI dataset.
Conclusions
- The proposed ensemble VT framework offers a significant advancement in Alzheimer's Disease classification accuracy.
- The approach shows promise for efficient AD detection, especially in challenging data scenarios.
- Identified limitations and future research directions are discussed, with code made publicly available.
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
