Classification of Liver Tissue Based on Histological and Imaging features

Abstract

Computer-aided diagnostic(CAD) system for characterization of different datasets for different diseases has the potential to assist the radiologists and clinicians in deducing whether the patient is suffering from the disease or not through histological processing by different machine learning algorithms which have to be applied on the given liver disorder datasets. In the absence of real time database, the current work of histological processing is based on a benchmark BUPA database created by University of California, Irvine. The current work aims to provide a means of economic assistance to the clinicians for the effective diagnosis of the liver diseases. For this purpose, two types of classification techniques are used. First, Histological classification and second, classification through imaging features. In histological feature classification, five classifiers namely K Nearest Neighbor (KNN), Probabilistic Neural Network (PNN), Support Vector Machine (SVM), Neural Networks Toolbox (NN) and Smooth Support Vector Machines (SSVM) are used. In imaging classification, first step is feature extraction, then feature selection and ultimately, classification. Here, Ultrasound images are used to maintain the costeffectiveness of the tool. Further, the work uses histological and imaging features collectively to further improve the overall efficiency of the designed tool