Combined Hierarchical Watershed Segmentation and SVM Classification for Pap Smear Cell Nucleus Extraction

Abstract

Abstract. In this paper, we propose a two-phase approach to nuclei segmentation/classification in Pap smear test images. The first phase, the segmentation phase, includes a morphological algorithm (watershed) and a hierarchical merging algorithm (waterfall). In the merging step, waterfall uses spectral and shape information as well as the class information. In the second phase, classification, the goal is to obtain nucleus regions and cytoplasm areas by classifying the regions resulting from the first phase based on their spectral and shape features, merging of the adjacent regions belonging to the same class. Between the two phases, three unsupervised segmentation quality criteria were tested in order to determine the best one selecting the best level after merging. The classification of individual regions is obtained using a Support Vector Machine (SVM) classifier. The segmentation and classification results are compared to the segmentation provided by expert pathologists and demonstrate the efficacy of the proposed method.