A fast and reliable approach to cell nuclei segmentation in PAP stained cervical smears

Abstract

Fast and reliable segmentation of cervical cell nuclei is one of the crucial steps of an automated screening system that aims early detection of cervical cancer. In this paper, we propose an edge based approach using customized Laplacian of Gaussian (LoG) filter to segment free lying cell nuclei in bright-field microscope images of Pap smear. The LoG is generally employed as a second order edge detector in image processing. The images may have the challenges of inconsistent staining, overlapping and folded cells. Experimenting proposed method over all types of cervical images including sufficient number of high grade lesions of cervical cancer shows that our method performs well for stain varied images containing focused nuclei.