RETRACTED: Modality feature fusion based Alzheimer’s disease prognosis

Abstract

Alzheimer’s is a dynamic and irreversible cerebrum degenerative issue which slowly affects brain tissues mainly in senile age. Given that there is no specific cure or effective treatment for Alzheimer’s, it has become a leading cause of death affecting an estimate of around 65 million people across the world. The symptoms include memory lapse, abnormal psych emotionalbehaviour owing to the consistent brain shrinkage. Identification and classification of Alzheimer’s from normal brain images captured using medical modalities is an easy and effective idea. Hence researchers are using modalities like MRI, PET and DTI. The objective of this paper is developing feature selection approach forenhancing the accuracy of multimodalities in Alzheimer disease diagnosis. In our work, we define a machine learning based multimodality feature fusion frame work. We extracted the features from each modality individually and then fused them together forming the aggregated feature vector. This feature vector is then fed as input to the training-testing sets for classification into Alzheimer Disease (AD) and Normal Control (NC). The framework is evaluated in terms of accuracy precision score (APS) and the classifier performance is depicted as ROC curve.