An Early Detection and Classification of Alzheimer's Disease Framework Based on ResNet-50

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Medical Diagnostics

Background:

• Alzheimer's disease (AD) poses a significant global health challenge.
• Early detection is crucial for effective management and treatment of AD.
• Conventional methods often lack the sensitivity and specificity required for early diagnosis.

Purpose of the Study:

• To develop an effective early detection system for Alzheimer's disease (AD).
• To leverage Deep Residual Network (ResNet) models and advanced image preprocessing techniques for improved AD classification.
• To address limitations of traditional Convolutional Neural Networks (CNNs) in feature extraction for AD detection.

Main Methods:

• Utilized Contrast Limited Adaptive Histogram Equalizer (CLAHE) and Boosted Anisotropic Diffusion Filters (BADF) for image preprocessing (equalization and noise reduction).
• Employed K-means clustering for segmentation of MRI scans.
• Developed a ResNet-50 model with shortcut connections for efficient feature extraction from AD patient MRI data.

Main Results:

• The proposed ResNet-50 based method achieved a high accuracy of 95% with minimal loss (0.12).
• Demonstrated superior performance compared to other models, avoiding overfitting issues.
• The framework proved robust and reliable for Alzheimer's disease categorization.

Conclusions:

• The integrated approach of ResNet-50 and image preprocessing offers an effective early detection system for AD.
• Highlights the potential of deep learning and image processing in creating accurate AD diagnostic tools.
• Represents a promising advancement over existing AD classification methods, paving the way for future research.