Alzheimer’s Disease Detection using Particle Swarm optimization

Abstract

Alzheimer’s disease is a chronic, progressive, and presently incurable neurologic disorder that leads to the demise of a patient. Alzheimer’s Disease should be detected at an early stage so that an effective treatment may be provided to the patient in time. In the proposed method Dual tree complex wavelet transform is used for feature extraction. The Principal Component Analysis is used to transform the raw features into new low-dimensional feature space. Principal Component Analysis elects the features based on their eigen values. All high eigenvalue features may not provide assurance for high classification accuracy and sensitivity. The best solution to this problem is the feature optimization approach. Thus, Particle swarm optimization is used for the selection of optimum subset of transformed features. These selected features are given to four different classifiers for performance evaluation. The proposed method is also executed with Discrete Wavelet Transform to evaluate its performance and is compared with three models viz. Dual Tree Complex Wavelet Transform, Dual Tree Complex Wavelet Transform +Principal Component Analysis and Dual Tree Complex Wavelet Transform +Particle swarm optimization. The proposed technique yielded 95.5% accuracy, 97.6% sensitivity, and 93.4% specificity on the Open Access Series of Imaging Studies dataset with k-Nearest Neighbour classifier. The obtained results outperform the other existing methods for AD prediction.