Q-Metrics for Early Detection of Cervical Cancer

A. Das; A. Kar; D. Bhattacharyya

doi:10.37591/joipprp.v1i1.22

Abstract

The most prevalent form of cancer in women worldwide is uterine cervical cancer. Through screening programs aimed at detecting precancerous lesions most cases of cervical cancer can be prevented. In this article, Q-metrics has been proposed for carrying out automated image segmentation of uterine cervical cancer. The validation of detection of cervical lesions is an important issue in medical image processing because it has a specific impact on surgical planning. We evaluated the segmentation accuracy on the basis of a four-sample validation metric against the estimated gold standard, which was derived from several domain experts’ manual segmentations by a novel algorithm. The distribution functions of the lesion and control pixel data were parametrically assumed to be a mixture of two beta distributions with different shape parameters. We estimated the corresponding receiver operating characteristic (R.O.C.) curve and Dice similarity coefficient (D.S.C.) in all possible decision thresholds. The automated segmentation yielded satisfactory accuracy with varied optimal thresholds.

Abstract

Keywords