Analysis of Fetus Image Using 2D Ultrasound Images

I. Introduction

Measurement of fetus parameters during each trimester or regular interval has to be done to avoid last minute complications involved during delivery and monitoring fetus growth stage by stage. Ultrasound imaging modality is used most commonly to measure the fetus parameters such as bi-parietal diameter (BPD), head circumference (HC), femur length (FL), abdomen circumference (AC), humerus length (HL), nuchal translucency (NT) and crown rump length (CRL). Since ultrasound imaging is non-invasive, low cost and it has no harmful radiations unlike other imaging modalities, makes it a convenient modality for imaging fetus. However, segmentation of ultrasound image is challenging. Since it has low contrast boundaries, high speckle noise and low signal to noise ratio and difficult to obtain accurate measurements. So more effort is needed for manipulating the parameters. It is time consuming process and also it will vary across sonographers. Hence reliable, accurate and semi-automatic or automatic method is needed.

This paper proposes the semi-automatic segmentation is carried out using active contour level set method. This will be helpful robust diagnosis of fetus and reducing human variability. Method for segmenting fetal ultrasound images a Conditional Random Field (CRF) based framework to handle challenges in segmenting fetal ultrasound images. The CRF framework uses wavelet based texture features for representing the ultrasound image and Support Vector Machines (SVM) for initial label prediction (Gupta & Sisodia, 2011).

A shape-guided variational segmentation method for extracting the fetus envelope on 3D obstetric ultrasound images is used. Segmentation framework that combines three different types of information: pixel intensity distribution, shape prior on the fetal envelope and a back model varying with fetus age to compensate the contrast (Dahdouh et al., 2013). A Bayesian formulation of the partition problem between the amniotic fluid and the fetal tissues integrates statistical models of the intensity distributions in each tissue class and regularity constraints on the contours (Anquez, 2013).

Measurement of the Nuchal Translucency thickness is made to identify the Down syndrome in screening first trimester fetus. The mean shift analysis and canny operators are utilized for segmenting the nuchal translucency region and the exact thickness has been estimated using Blob analysis. It is observed from the results that the fetus in the 14th week of Gestation is expected to have a nuchal translucency thickness of 1.85 ± 0.48 mm (Nirmala, 2009). The shape sensitive derivative class separable segmentation scheme for the Ultrasound fetal images is used. The energy cost function is optimized with topological asymptotic expansion for feature extraction. The speckle present in the image has been removed by improved iterative median filter (Priestly Shan & Madhesawaran, 2009).

The automatic detection and measurement of fetal anatomical structures that directly exploits a large database of expert annotated fetal anatomical structures in ultrasound images and it learns automatically to distinguish between the appearance of the object of interest and background by training a constrained probabilistic boosting tree classifier (Carnerio, 2009). Segmentation of ultrasound images using a new speed term based on local phase and local orientation derived from the monogenic signal, which makes the algorithm robust to attenuation artifact (Belaid et al., 2011). With a morphologic filtering, it establishes the edge map and extracts a preliminary contour by the gradient vector flow (GVF) snake to estimate the NT parameters of the fetus (Yn-Hui et al., 2008). The block diagram of the methodology proposed in this paper is given in Figure 1.

Figure 1.

Block diagram of proposed system

Ii. Methodology