A comprehensive hybrid model: Combining bioinspired optimization and deep learning for Alzheimer's disease...

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Intelligence

Background:

• Alzheimer's disease (AD) is a progressive neurodegenerative disorder causing cognitive decline.
• Accurate diagnosis and monitoring of AD are crucial for effective patient care and treatment decisions.
• Current diagnostic methods can be enhanced by advanced image segmentation and classification techniques.

Purpose of the Study:

• To develop and evaluate a novel bioinspired hybrid optimization technique for segmenting brain subregions relevant to AD diagnosis.
• To improve the accuracy of identifying Alzheimer's disease biomarkers through enhanced image segmentation.
• To integrate segmentation with deep learning for accurate classification of AD patients and normal controls.

Main Methods:

• A hybrid approach combining Gray Wolf Optimization (GWO) and Harris Hawk Optimization (HHO) for brain region segmentation.
• Concurrent subpopulation position updates within the GWO and HHO frameworks to refine segmentation.
• Application of deep learning (DL) techniques for classifying normal controls (NC) and AD patients post-segmentation.
• Validation of segmentation accuracy using statistical measures against ground truth (GT).

Main Results:

• The hybrid GWO-HHO technique achieved a segmentation accuracy of 92%.
• The proposed hybrid method demonstrated superior performance compared to HHO alone.
• Deep learning classification achieved 90% accuracy in distinguishing between NC and AD.
• Clinical validation using the Mini Mental State Examination (MMSE) supported disease progression monitoring.

Conclusions:

• The hybrid GWO-HHO technique offers a significant advancement in segmenting brain regions for AD diagnosis.
• The integrated approach of segmentation and deep learning classification shows high accuracy for AD detection.
• This method provides valuable support for clinicians in monitoring AD progression and making treatment decisions.