An improved nucleus segmentation for cervical cell images using FCM clustering and BPNN

Abstract

Pap smear test plays an important role for the early diagnosis of cervical cancer in which human cells taken from the cervix of patient are analysed for pre-cancerous changes. The manual analysis of these cells by expert cytologist is labor intensive and time consuming job. In this paper, an improved nucleus segmentation algorithm is proposed using FCM clustering and BPNN. The existing algorithm based on FCM clustering has been improved by finding optimum clusters instead of fixed clusters. Further, shape based features are extracted from each region which act as input to Back Propagation Neural Network (BPNN); to classify regions as nucleus or non-nucleus. Thus false detected regions are removed to produce the accurate segmentation of nucleus regions. The proposed work is evaluated on the public available Herlev dataset. Experimental results show the improvement in performance (precision, recall and Dice Coefficient) of nucleus segmentation by 1%, 7% and 5% respectively compared to existing work.