Application of the Fuzzy C-Means Clustering Method on the Analysis of non Pre- processed FTIR Data for Cancer Diagnosis

Fourier-transform infrared spectroscopy (FTIR) is an efficient, sensitive and computer operated technique that can detect changes in cellular composition that may reflect the onset of a disease. As such, it is being investigated as a method for automatic early detection of pre-cancerous changes. In previous work, FTIR spectral data was first empirically pre-processed and then classified using various data clustering techniques in order to compare to manually obtained classifications. It was found that accurate clustering could only be achieved by manually applying pre-processing techniques that varied according to the particular sample characteristics. In this paper, two data clustering techniques, Hierarchical Cluster Analysis (HCA) and Fuzzy C-Means (FCM) clustering, are used to classify sets of oral cancer cell data without a pre-processing procedure. The performances of these two techniques are compared and their differences are discussed. The FCM method was found to perform significantly better.