Diagnosis Support System for Cervical Cancer Using Image Processing

Abstract

In this paper, we propose a method for a screening system aimed at reducing the workload of cytologists in cervical cancer diagnosis. Since image processing is required according to the morphological characteristics of the cells, we first classified the cells according to their types. For nuclear extraction, we proposed two methods. Then, we detected malignant cells by the size of the nuclear area and calculated the accuracy of the detection.