A supervised texton based approach for automatic segmentation and measurement of the fetal head and femur in 2D ultrasound images

Abstract

This paper presents a supervised texton based approach for the accurate segmentation and measurement of ultrasound fetal head (BPD, OFD, HC) and femur (FL). The method consists of several steps. First, a non-linear diffusion technique is utilized to reduce the speckle noise. Then, based on the assumption that cross sectional intensity profiles of skull and femur can be approximated by Gaussian-like curves, a multi-scale and multi-orientation filter bank is designed to extract texton features specific to ultrasound fetal anatomic structure. The extracted texton cues, together with multiscale local brightness, are then built into a unified framework for boundary detection of ultrasound fetal head and femur. Finally, for fetal head, a direct least square ellipse fitting method is used to construct a closed head contour, whilst, for fetal femur a closed contour is produced by connecting the detected femur boundaries. The presented method is demonstrated to be promising for clinical applications. Overall the evaluation results of fetal head segmentation and measurement from our method are comparable with the inter-observer difference of experts, with the best average precision of 96.85%, the maximum symmetric contour distance (MSD) of 1.46 mm, average symmetric contour distance (ASD) of 0.53 mm; while for fetal femur, the overall performance of our method is better than the inter-observer difference of experts, with the average precision of 84.37%, MSD of 2.72 mm and ASD of 0.31mm.

Automatic image quality assessment and measurement of fetal head in two-dimensional ultrasound image

“…Multiscale and multiorientational local intensity and texture cues extracted from a texton map specific to the US anatomic structures are then combined for the initial skull boundary detection. Different from our previous work, 16 where a support vector machine (SVM) was used to identify skull segments from the initial boundary candidates, an RF algorithm 28 is used for skull boundary identification. In the following subsections, each step is summarized.…”

“…The generated texton map (Tmap) is used to calculate the texture-oriented gradient that provides significant local information for the initial skull boundary detection. 16 The oriented gradient magnitude at each pixel ðx; yÞ is calculated by employing the χ 2 distance of histograms between two-half discs defined as E Q -T A R G E T ; t e m p : i n t r a l i n k -; e 0 0 1 ; 3 2 6 ; 6 9 7 χ 2 ðg; hÞ ¼ 1 2…”

“…The final global probability of boundary (gPb) is formed as a weighted sum of mPb and sPb. 16,31 This process allows weak boundaries to be determined and excluded from strong boundaries while preserving contour quality. Example results with respect to the mPb, sPb, and gPb are shown in Figs.…”

“…Different from our previous work in Ref. 16 where an SVM is used to identify skull segments from the initial boundary candidates, an RF classifier is used for skull boundary identification. RF is an ensemble technique that uses multiple decision trees.…”

“…Figure 2 shows the main framework of our method. A texton-based US fetal head segmentation is first adopted 16 in which a filter bank is designed to extract texton features specific to US fetal anatomic structures; multiscale local brightness and texture cues are taken into account for an initial skull boundary detection, which is then followed by the identification of the fetal skull using a supervised learning-based method. The image quality assessment step includes skull midline detection and feature calculation from the segmented head region, and classification of US image quality using an RF classifier.…”