Hierarchical median narrow band for level set segmentation of cervical cell nuclei

This paper presents a novel hierarchical nuclei segmentation algorithm for isolated and overlapping cervical cells based on narrow band level set implementation. Our method applies new multiscale analysis to estimate the number of clusters in each image region containing cells, which turns into input algorithm. We assess results three public cell databases. Overall, our outperformed six state-of-the-art methods concerning correctly segmented Dice coefficient reached values equal or higher than 0.90. also carried out classification experiments using features extracted from proposed pipeline achieved highest average accuracy 0.89 0.77 two-class three-class problems, respectively. These demonstrated suitability integrate decision support systems screening. • A segmentation. cluster detection strategy within initialization Nuclei clumps cells. preserve intrinsic shape texture nuclei. Classification tasks improved with results.