Model of Watershed Segmentation in Deep Learning Method to Improve Identification of Cervical Cancer at Overlay Cells

Abstract

Cervical cancer is a disease that is very scary for women because it is the cause of death among women. To be aware of this disease is to do an early examination through the Pap Smear (PS) test. In terms identifying overlapping cancer cells, it still has low accuracy. Therefore, this research was carried out with the aim of getting the level of cell separation with high accuracy. This study uses a model to develop the Watershed segmentation technique in the Deep Learning Method. The data tested in this study comes from the RepomedUNM dataset. The amount of data tested is 420 overlapping images with the formulation of 1,260 test images. The results of this study can very well separate each overlapping cell with an average Intersection over Union (IoU) score of 0.9061. Each result can be divided fully by the whole of its area, so the final results of overlapping cells were successfully separated with an average score of 0.945. Therefore, this research can be used as a reference in identifying cervical cancer cells.