A Feature-Fusion Technique-Based Alzheimer's Disease Classification Using Magnetic Resonance Imaging

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Neuroscience

Background:

• Early Alzheimer's disease (AD) detection is crucial for effective management.
• Deep learning (DL) models like CNNs and ViTs show promise for AD diagnosis but often lack interpretability and require significant computational resources.
• Hybrid Vision Transformers (ViTs) offer improved feature visualization and interpretability with reduced computational demands.

Purpose of the Study:

• To present an innovative, resource-efficient model for classifying Alzheimer's disease (AD) using MRI images.
• To address the limitations of existing deep learning models in terms of interpretability and computational cost.

Main Methods:

• Modified existing Vision Transformers (ViTs) to enhance Alzheimer's disease (AD) feature extraction.
• Employed a CatBoost-based classifier for multi-class classification of extracted features.
• Validated the model's generalization capabilities on the OASIS dataset.

Main Results:

• The proposed model achieved a high classification accuracy of 98.8% on the OASIS dataset.
• The model demonstrated a minimal loss value of 0.12, indicating strong performance.
• The approach proved effective in classifying Alzheimer's disease (AD) with high precision.

Conclusions:

• The developed model presents an interpretable and computationally efficient solution for Alzheimer's disease (AD) classification in healthcare settings.
• Future research should incorporate genetic and clinical data to further enhance model robustness and applicability.
• This work highlights the potential of hybrid ViTs in medical diagnostics.