An automatic segmentation of cervical intraepithelial neoplasia (CIN3) from parabasal cells

This paper presents a method to automatic segmentation of cervical intraepithelial neoplasia (CIN3) parabasal cervical cells from PAP Smear images. The proposed method based on the structural characteristics of cervical; the region of nucleus by using Active Contour Model (ACM). The energy function is minimized in order to correct the less gradient of image energy area; to correct the cloudy contour. The performance of the proposed method is evaluated by comparing the results from Linear Discriminant Analysis (LDA) to expert diagnosis. The experimental results performing with 100 normal cases and 100 abnormal cases shown classification rate 93.5%.