Computational Intelligence Algorithms for Bioimpedance-Based Classification of Biological Material

摘要

This paper reports a new particle swarm optimization method for fitting electrical bioimpedance spectra with the Cole-Cole model. It was compared with least squares and genetic algorithm fitting methods, where the proposed method was shown to have some advantages on computational performance, and also superior robustness to spectral noise and distortions. An artificial neural network for bovine tissue classification is also presented, where the use of the fitted model parameters as inputs has shown a better noise robustness than raw data or other fitting methods.