Optimal Spectral Histology of Human Normal Colon by Genetic Algorithm

Abstract : Fourier transform infrared (FTIR) imaging combined with unsupervised clustering method, such as k-means, achieves a real histology of human tissues. This technique has been successfully applied to diagnose different cancers. However, the clustering methods used in spectral histology are local search algorithms, i.e. these methods converge to a local optimum. Metaheuristics are effective methods to overcome this problem and to reach the optimal solution. In this work, we propose a genetic algorithm for the optimal clustering of FTIR images of normal human colon tissues. The obtained results show the efficiency of the proposed genetic algorithm to retrieve more precisely than k-means the structures of normal colon.