Diagnosis of Liver Fibrosis using RBF Neural Network and Artificial Bee Colony Algorithm

Abstract: Liver turquoise is one of the silent and dangerous diseases. If it can be detected in the early stages, the lives of many affected people can be saved. Providing smart methods to identify and diagnose this disease can save patients' lives in addition to reducing medical costs and overheads. In this research, an innovative method using a three-layer radial basis neural network is proposed as a multi-class method for diagnosing liver fibrosis. To increase the accuracy and efficiency of the pre-processed data, the data are balanced using the SMOT method. Also, feature selection is done with the bee algorithm. In this way, the desired features are first reduced using the bee algorithm. For this purpose, a mapping of features is done using the bee algorithm. Then the data with reduced features are applied to the proposed RBF network. The simulation results show that the proposed method is 5% more accurate than similar methods.

Keywords: Liver Fibrosis Diagnosis - Feature Selection – Artificial Bee Colony - Radial Basis Function (RBF) neural network.