Cell extraction from sputum images for early lung cancer detection

The context of this work is the design of Computer-Aided Design system for the early detection of lung cancer using digital sputum images. This process requires the detection of the sputum cell region in the sputum image characterized by a cluttered background. In this paper we address this problem using two different methods, namely, a Rule-based method, and Bayesian classification. We describe the two methods and we compare their performances in terms of their behaviors with respect to color representation and color quantization.