Automatic Segmentation of Ovarian Follicle Using K-Means Clustering

Abstract

Automatic detection of human ovarian follicles has been of increasing interest in recent years and is a significant area of women's health. Improper development of ovarian follicles has been an important reason for infertility in women. Currently, detection of ovarian follicle is done through diagnostic imaging technique called ultrasonography. Follicles differ in shape and colour. Further, the camouflaging characteristic of ultrasound images and the presence of speckle noise make the follicle detection a challenging task. In this paper, a novel method for automatic recognition of follicles in ultrasound images is proposed. Discrete wavelet transform based k-means clustering is proposed. Discrete wavelet transform is preferred due to its superior spectral temporal resolution that helps in despeckling the ultrasound images. K-means clustering is used to segment the image into different anatomical structures to yield better segmentation. Structural Similarity (SSIM), False Acceptance Rate (FAR) and False Rejection Rate (FRR) are used to demonstrate the efficiency of the proposed method.