266 Int. J. Data Mining and Bioinformatics, Vol. 5, No. 3, 2011 Image segmentation of biofilm structures using optimal multi-level thresholding Darío Rojas Department of Computer Science, University of Atacama, 485 Copayapu Ave., Copiapó 1532296, Chile E-mail: dario.rojas@uda.cl Luis Rueda* and Alioune Ngom School of Computer Science, University of Windsor, 401 Sunset Ave., Windsor, ON N9B 3P4, Canada E-mail: lrueda@uwindsor.ca E-mail: angom@uwindsor.ca *Corresponding author Homero Hurrutia and Gerardo Cárcamo Center for Biotechnology and Faculty of Biological Sciences, University of Concepción, 4070386, Chile E-mail: hurrutia@udec.cl E-mail: gecarcamo@uder.cl Abstract: The appreciation of biofilm structures in digital images can be subjective to the observer, and hence it is necessary to analyse the underlying images in useful parameters by means of quantification that is, ideally, free of errors. This paper proposes a combination of techniques for segmentation of biofilm images through an optimal multi-level thresholding algorithm and a set of clustering validity indices, including the determination of the best number of thresholds. The results, which are validated through Rand Index and a quantification process performed in a laboratory, are similar to the quantification and segmentation done by an expert. Keywords: clustering; biofilm image processing; image segmentation; multi-level thresholding; biofilm quantification; biofilm structures. Reference to this paper should be made as follows: Rojas, D., Rueda, L., Ngom, A., Urrutia, H. and Cárcamo, G. (2011) ‘Image segmentation of biofilm structures using optimal multi-level thresholding’, Int. J. Data Mining and Bioinformatics, Vol. 5, No. 3, pp.266–286. Copyright © 2011 Inderscience Enterprises Ltd.