An effective Alzheimer's disease segmentation and classification using Deep ResUnet and Efficientnet

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) is a neurodegenerative disorder characterized by progressive brain cell impairment and cognitive decline.
• Early diagnosis of AD is critical for effective patient care and treatment, yet conventional methods face limitations.
• Advanced image analysis techniques are needed for accurate and early AD detection.

Purpose of the Study:

• To develop an accurate and efficient system for Alzheimer's disease (AD) identification using magnetic resonance imaging (MRI).
• To leverage deep learning and advanced feature extraction for improved AD diagnosis.
• To categorize the stages of Alzheimer's disease with high precision.

Main Methods:

• Brain MRI images were segmented using a Deep ResUnet-based approach.
• Global and local features were extracted via a Multi-Scale Attention Siamese Network (MASNet).
• The Slime Mould Algorithm was employed for feature selection, followed by classification using EfficientNetB7.

Main Results:

• The proposed method achieved high accuracy in Alzheimer's disease (AD) categorization.
• Tested on Kaggle and ADNI datasets, the system demonstrated accuracies of 99.31% and 99.38%, respectively.
• The results indicate the system's effectiveness for accurate AD staging.

Conclusions:

• The developed AI-driven system offers a highly accurate and efficient approach for Alzheimer's disease (AD) diagnosis.
• This method shows significant potential for enhancing early detection and patient management in clinical settings.
• The study highlights the utility of integrated deep learning techniques for neurodegenerative disease classification.