A GA-Based Multiresolution Feature Selection for Ultrasonic Liver Tissue Characterization

This work describes the feasibility of multiresolution feature selection and its application to classify ultrasonic liver images. The proposed approach uses genetic algorithm and defines a novel fitness function for medical applications since the diagnosis correctness is the most important consideration. Via the measurement of class seperability, we can uniquely select the feature vector. The effectiveness of the proposed approach is tested by using five different classifiers. From the experimental results, the proposed approach is trustworthy.