Optimization enabled ResNet features with transfer learning for Alzheimer's disease detection.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Neurodegenerative Disease Diagnostics

Background:

• Alzheimer's Disease (AD) affects millions globally, necessitating early detection for effective management.
• Medical imaging, particularly MRI, shows promise for AD diagnosis but faces challenges with image complexity and limited data.
• Accurate and efficient AD detection methods are critical for improving patient outcomes.

Purpose of the Study:

• To propose a novel, optimization-enabled ResNet feature extraction technique for enhanced Alzheimer's Disease detection.
• To leverage transfer learning by combining LeNet and VGG networks for improved diagnostic accuracy.
• To address challenges in medical image analysis for early AD identification.

Main Methods:

• Pre-processing involved image resizing and median filtering.
• Feature extraction utilized a novel Walrus Optimization Algorithm-Residual neural network (WOA-ResNet) for ResNet training.
• Transfer learning was implemented by integrating LeNet and VGG networks.

Main Results:

• The proposed LeNet-VGG method combined with WOA-ResNet achieved high accuracy in Alzheimer's Disease detection.
• The method demonstrated a maximum accuracy of 95.37%, sensitivity of 97.24%, and specificity of 93.73%.
• Optimization-enabled ResNet feature extraction significantly improved diagnostic performance.

Conclusions:

• The developed technique shows significant potential for accurate and early Alzheimer's Disease detection.
• Combining optimization algorithms with deep learning models enhances diagnostic capabilities for neurodegenerative diseases.
• This approach offers a promising tool for clinical application in Alzheimer's Disease diagnostics.