Early Prediction of Alzheimer’s Disease Using Low Complexity Techniques for High Data Classification

Abstract

Alzheimer’s disease is one of the most common Neuro Degenerative diseases that commonly affects people above age of 60. The disease is progressive in nature and affects the grey matter of brain cells, causing permanent damage to the thinking capability. This paper proposes a new technique for identification and observing progression of Alzheimer’s disease using contours and clustering technique. The number of MRI used for classification is 900 in which 400 is of Alzheimer’s disease (AD) and 500 is of cognitive normal (CN) and the classification is done using random forest. The feature used for classification is shape and structure of brain and the end result showed a significant improvement in the classification of AD patient and CN patient. This paper compares our method with current methods in the field of Alzheimer’s disease detection using machine learning to show the reliability of our method. The proposed method shows improvement in accuracy when compared with existing machine learning methods. The accuracy of this proposed method calculated using 5-fold cross validation is found to be 91.6%.