Detection and Classification of Alzheimer's Disease Using Deep and Machine Learning

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Imaging

Background:

• Alzheimer's disease (AD) is the primary cause of dementia, presenting significant challenges in early diagnosis and management.
• Traditional diagnostic methods for AD often lack sensitivity, especially in early disease stages.
• A robust diagnostic approach is critical for improving patient outcomes and reducing socioeconomic impact.

Purpose of the Study:

• To develop and validate a dual-modal framework integrating clinical data and MRI for enhanced Alzheimer's disease staging.
• To leverage machine learning (ML) and deep learning (DL) models, augmented by explainable AI (XAI), for improved diagnostic accuracy.
• To provide clinicians with a reliable tool for timely AD diagnosis and personalized treatment planning.

Main Methods:

• Four ML classifiers (KNN, SVM, DT, RF) were trained on demographic and clinical features.
• Five DL models (CNN, EfficientNetB3, DenseNet-121, ResNet-50, MobileNetV2) were applied to MRI scans for stage-wise classification.
• SHAP and Grad-CAM techniques were employed for model interpretability and identification of key diagnostic features.

Main Results:

• The Random Forest model achieved 97% accuracy using clinical data.
• The CNN model demonstrated the highest performance (94%) in MRI-based AD staging.
• Explainable AI identified significant indicators such as hippocampal atrophy and ventricular enlargement.

Conclusions:

• Integrating clinical data, MRI, and interpretable AI significantly enhances the accuracy and reliability of Alzheimer's disease staging.
• The proposed framework offers a clear and validated diagnostic pathway for clinicians.
• This approach supports timely diagnosis and facilitates the adjustment of individualized treatment strategies for AD patients.