A Hybrid Approach to detect and characterize Alzheimer’s Disease using Robust PCA and Random Forest Algorithm

Alzheimer's disease is a neurological degenerative disorder that impacts the memory potency and entire function of the human brain that leads to Dementia. It is commonly regarded as the seventh leading cause of mortality in the globe. Despite this, there is still no solid test to diagnose this condition. The focus of this thesis is to imply a new hybrid model for better detection and categorization Alzheimer's disease. The integration of brain MRI (Magnetic Resonance Imaging) data in Computerized Algorithms helps with the rapid recognition and segmentation of Alzheimer's disease. Data preparation, Data Augmentation, Feature extraction, and Classification are the four different phases outlined in the approach. We applied PCA(Principal Component Analysis) method for feature extraction and Random Forest Algorithm as a classification technique to cluster MRI images to achieve flawless results. Those image pixels are clustered into four leverages using the proposed approach. MildDemented, VeryMildDemented, NonDemented and ModerateDemeneted are among the four Alzheimer categories for which the algorithm has been implemented. The proposed system's efficiency is determined using the Kaggle Alzheimer's dataset. The results suggest that this approach has a high sensitivity for detecting lesion location in the brain with an overall accuracy level of 99.95%.