Bagging and Fusion of Multiple Feature Extraction Models for Early Diagnosis of Alzheimer's disease

Abstract

In this work, a new algorithm is proposed using a combination of different point detection feature extraction methods like SURF, FAST, BRISK, Harris and Min Eigen for early diagnosis and prediction of various stages of Alzheimer's disease. This new method is integrated with different classifiers like Decision Tree and Naive Bayes for the classification of different stages of Alzheimer's disease. The classification accuracy and performance parameters are determined and analyzed for both classifiers. The new proposed method provides an accuracy rate of 98.13% when Decision tree classifier is used and gives an accuracy rate of 97.31% when Naïve Bayes classifier is used. The accuracy rate and performance of the proposed method is found to be very high when compared to the existing feature extraction methods. From the analysis results it's been observed that the new proposed method is found to be more accurate and sensitive than the existing algorithms since it uses multiple combined feature extraction techniques, when compared to the techniques which uses single feature extraction method.