Detection of Alzheimer's Disease Based on Cloud-Based Deep Learning Paradigm

Area of Science:

• Artificial Intelligence
• Medical Imaging Analysis
• Neurology

Background:

• Deep learning excels in complex pattern recognition for medical image analysis.
• Accurate Alzheimer's disease (AD) classification is crucial for timely intervention.
• Traditional algorithms face challenges in classifying intricate neurodegenerative patterns.

Purpose of the Study:

• To develop a local cloud-based deep learning solution for multi-class Alzheimer's disease (AD) classification.
• To leverage transfer learning with a pre-trained GoogLeNet model for enhanced AD diagnosis.
• To create a user-friendly web application for remote AD prediction using MRI scans.

Main Methods:

• Utilized a pre-trained GoogLeNet model for image classification.
• Employed transfer learning to fine-tune the GoogLeNet model for AD classification.
• Developed a local cloud-based web application integrating the fine-tuned model.
• Input modality: Magnetic Resonance Imaging (MRI) scans.
• Classification: Multi-class for four stages of AD.

Main Results:

• Achieved a high classification accuracy of 98%.
• Successfully developed a functional local cloud web application for Alzheimer's prediction.
• Demonstrated the efficacy of transfer learning in improving AD classification accuracy.

Conclusions:

• The proposed deep learning approach, utilizing GoogLeNet and transfer learning, offers a highly accurate method for Alzheimer's disease classification.
• The developed cloud-based application facilitates remote and efficient AD diagnosis for medical professionals.
• This technology has the potential to significantly improve the early detection and management of Alzheimer's disease.