A CAD System for Alzheimer's Disease Classification Using Neuroimaging MRI 2D Slices.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Alzheimer's disease (AD) is a leading cause of dementia, posing significant economic and healthcare challenges.
• Early and accurate diagnosis of AD is crucial for effective patient management and treatment.
• Existing machine learning methods show promise, but advancements in deep learning offer enhanced capabilities for complex pattern recognition in medical data.

Purpose of the Study:

• To develop and evaluate a hybrid deep learning model integrating Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) for Alzheimer's disease classification.
• To leverage the feature extraction strengths of CNNs and the classification power of SVMs for improved diagnostic accuracy.
• To validate the proposed hybrid CNN-SVM model on independent datasets (ADNI and OASIS) for robust performance assessment.

Main Methods:

• Utilized brain Magnetic Resonance Imaging (MRI) scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset.
• Developed a hybrid model combining a CNN for automated feature extraction from MRI scans and an SVM for classification.
• Compared the performance of the hybrid CNN-SVM model against a standalone CNN model.

Main Results:

• The hybrid CNN-SVM model demonstrated superior classification performance compared to the CNN model alone across various diagnostic comparisons (AD vs. CN, CN vs. MCI, AD vs. MCI, and CN vs. MCI vs. AD).
• Relative accuracy improvements ranged from 0.85% to 3.4% on the ADNI testing dataset.
• The model achieved an accuracy of 86.2% when tested on the OASIS dataset, indicating strong generalization capabilities.

Conclusions:

• The hybrid CNN-SVM model represents a promising advancement in computer-aided diagnosis of Alzheimer's disease.
• This integrated approach effectively enhances classification accuracy, offering a valuable tool for early AD detection and differentiation from mild cognitive impairment and normal cognition.
• The findings support the potential of deep learning techniques in improving diagnostic outcomes for neurodegenerative diseases.