Multimodal Neuroimaging Based Alzheimer's Disease Diagnosis Using Evolutionary RVFL Classifier

Area of Science:

• Neuroscience
• Medical Imaging
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) is a leading cause of dementia, characterized by cognitive decline.
• Early detection, often through identifying mild cognitive impairment (MCI), is crucial for managing AD.
• Standard diagnostic biomarkers include structural changes (MRI) and metabolic abnormalities (PET).

Purpose of the Study:

• To propose a novel method for early Alzheimer's disease detection.
• To integrate structural and metabolic information from MRI and PET scans.
• To enhance diagnostic accuracy for neurodegenerative diseases.

Main Methods:

• Wavelet transform-based multimodality fusion of MRI and PET scans.
• Feature extraction using the ResNet-50 deep learning model.
• Classification of features using an optimized Random Vector Functional Link (RVFL) network with an evolutionary algorithm.

Main Results:

• The proposed method demonstrates efficacy in early Alzheimer's disease detection.
• Multimodality fusion effectively incorporates structural and metabolic information.
• Optimized RVFL network achieves high accuracy in classification tasks.

Conclusions:

• The developed technique shows promise for the early diagnosis of Alzheimer's disease.
• Combining MRI and PET data through wavelet fusion enhances diagnostic capabilities.
• Deep learning and evolutionary algorithms offer powerful tools for neurodegenerative disease research.