Intelligence Ensemble-Based Feature Selection (Iefs) Algorithm and Fuzzy Convolutional Neural Network (Fcnn) for Hepatocellular Carcinoma (Hcc) in Liver Disease System

Abstract

Predicting HCC (Hepato Cellular Carcinoma) risks is an important pre-disposing condition for patients with affected with fatty liver diseases who are non-alcoholic and patients recovering from hepatitis C viral infections.  Though HCC can be diagnosed early, missing values and innumerable features in patient’s data/datasets makes it a complicated issue. Missing values can be neutralized by ignoring or removing or imputing them where ignoring or removing information reduces processing information while accuracy while analyzing them. This research work imputes missing values using an Improved FCM (Fuzzy C Means) clustering and thus enhances analysis accuracy. Improved FCM is used as data can belong to multiple groups based on their membership values. The imputed features are then reduced by selecting required feature using an intelligent ensemble based feature selection. Multiple ensemble methods are used in the study for selecting the optimal feature subset which includes SAFSA (Score based Artificial Fish Swarm Algorithm), MSBOA (Mutation Score Butterfly Optimization Algorithm) and SMFO (Score Moth-Flame Optimization). These ensemble methods result in enhancing prediction accuracy in classifications executed using FCNNs (Fuzzy Convolution Neural Networks). The proposed scheme when tested on MATLAB (Matrix Laboratory) could classify better than most other methods as it had better values in terms of precision, recall, F-measure and accuracy.