An Evolutionary Algorithm for Enhanced Magnetic Resonance Imaging Classification
T.S. Murunya, Audithan Sivaraman · Research Journal of Applied Sciences Engineering and Technology · 2014
This study presents an image classification method for retrieval of images from a multi-varied MRI database. With the development of sophisticated medical imaging technology which helps doctors in diagnosis, medical image databases contain a huge amount of digital images. Magnetic Resonance Imaging (MRI) is a widely used imaging technique which picks signals from a body's magnetic particles spinning to magnetic tune and through a computer converts scanned data into pictures of internal organs. Image processing techniques are required to analyze medical images and retrieve it from database. The proposed framework extracts features using Moment Invariants (MI) and Wavelet Packet Tree (WPT). Extracted features are reduced using Correlation based Feature Selection (CFS) and a CFS with cuckoo search algorithm is proposed. Naïve Bayes and K-Nearest Neighbor (KNN) classify the selected features. National Biomedical Imaging Archive (NBIA) dataset including colon, brain and chest is used to evaluate the framework.