Modified Grey Wolf Randomized Optimization in Dementia Classification Using MRI Images
N. Bharanidharan, Harikumar Rajaguru · IETE Journal of Research · 2020
In this research work, a novel Modified Grey Wolf Randomized Optimization technique is used to classify the brain MRI images into two classes i.e. Non-Dementia or Dementia class. The swarm intelligence algorithms like Grey Wolf Optimization technique is predominantly used in solving optimization and feature selection problems. But using it to categorize brain MRI images for dementia classification will be incisive. Cross-sectional MRI of 65 Non-Dementia and 52 Dementia subjects collected from OASIS are used in the work. The original grey wolf optimization offers very poor performance due to its randomness and so randomness is removed and control parameters are inserted in modified grey wolf optimization. But the accuracy of this technique is not satisfactory due to the local optima problem. Hence the novel modified grey wolf randomized optimization with controlled randomness at an appropriate position of the original algorithm. Further Principal Component Analysis, Detrended fluctuation analysis, Hilbert Transform, K-Means clustering are added to the modified Grey Wolf Optimization versions for accuracy improvement and evaluated with and without statistical features. The highest accuracy of 93.16% is provided by Hilbert Transform-Modified Grey Wolf Randomized Optimization technique without statistical features while 51.28% and 52.99% accuracy is achieved by the original Grey Wolf Optimization with and without statistical features respectively.