Non-Invasive Prospective Approach for the Detection of Alzheimer’s disease

Abstract

We proposed an automated novel method to detect Alzheimer disease (AD). The methodology involves the analysis of normal and AD MRI (Magnetic Resonance Imaging) brain scans, we extracted some specific portions of brain which changes in case of diseased subjects such as Hippocampus, Septum Pellucidum, fornix and some portion of thalamus. We measured the area of brain parts lost due to AD and compared these measurements with the same aged normal subjects. In this research work various pattern recognition techniques were used that separates the AD brain scans form the brain scans of healthy controlled subjects. These pattern recognition techniques includes segmentation of brain images, wavelet based texture features extraction for the classification of brain scans. We used two different classifiers ANN (Artificial Neural Network) and SVM (Support Vector Machine) and which showed the comparable accuracy, execution time than other classifiers reported so far.