Classification of Liver Abnormality in Ultrasonic Images Using Hilbert Transform Based Feature

The diagnosis of liver abnormalities from ultrasonic images is examined in this paper using transform Based features and classifiers. The liver ultrasonic images are acquired from cancer imaging archive database and Hilbert Transform based features are extracted from the liver ultrasonic images. The five classifiers Softmax Discriminant Classifier (SDC), Detrend Fluctuation Analysis, Naïve Bayesian Classifier (NBC), Harmonic search, Artificial Algae optimization (AAO) are utilized to identify abnormalities in liver ultrasound images. The classifiers performances are analyzed using standard parameters such as sensitivity, Specificity, Accuracy, Error Rate, Mathew Correlation Coefficient (MCC), and Jaccard Metric (JM). The Artificial Algae optimization (AAO) classifier outperforms with other classifiers with highest classification accuracy of 98.92% and an error rate of 1.08%.