Classification of Dementia Using Harmony Search Optimization Technique

Abstract

Soft computing techniques can be used in automated classification of dementia, to help the clinician in dementia diagnosis. This research paper uses Harmony Search optimization technique to classify dementia through MRI images. In literature, Harmony Search algorithm is used extensively for optimization problem, feature selection and training Neural Networks. But using Harmony Search for classification of medical images is ingenious. OASIS cross sectional dataset containing MRI brain images of 30 non-dementia and 30 dementia patients are used in this analysis. After the selection of optimum values for Harmony Memory Considering Rate and Pitch Adjusting Rate, this technique yields Goodness Detection Ratio of 94.73% while Particle Swarm optimization and Artificial Bee Colony with optimum weights yields only 64.15% and 62.7% in dementia classification respectively.