Overlapping Oral Epithelial Cells Segmentation: Voronoi-Based Hybrid Active Contour Model

Oral cancer is a major healthcare problem and accounts for approximately 40% of cancer risk. It is evident that the detection of oral cancer at an early stage can markedly improve the survival rate. Though several invasive ‘gold standard’ methods continue to be practiced clinically, non-invasive detection modalities look promising for early oral cancer risk assessment. Exfoliative cytology is one such method in which the desquamated cells are collected from the oral epithelium and examined under microscope. However, the main challenge in automated quantification arises here on the formation of cellular clump during the smear preparation. In this chapter, the role of computational advances in segmentation of overlapping cells is illustrated. A computational model, based on Voronoi-based hybrid active contour method, is proposed for segmenting the overlapping oral epithelial cells.