Phase-based thresholding schemes for segmentation of fetal thigh cross-sectional region in ultrasound images

Abstract

Measurement of thigh circumference has been used as an indication of nutritional adequacy and fetal growth in the womb. Fractional thigh volume is one of the variables for estimating fetal weight that is also calculated using the fetal thigh circumference. This cross-section of the fetal thigh is an intricate part of detecting because it has a less clear boundary and sometimes attached to other tissue in the ultrasound image. This study aims to establish a detection scheme and segmentation of the fetal thigh cross-sectional area automatically on ultrasound images. We propose a segmentation framework that integrates two phase-based thresholding schemes and a gap connecting step. In the first scheme, we combine the phase congruency of the local phase features and the fuzzy entropy method. The second thresholding scheme is a combination of the phase symmetry method and the saliency visual attention model. The proposed framework has the advantage of being able to segment cross-sectional objects more precisely, even though there are other objects attached to them. This method also works well when applied to low-quality ultrasound images taken using a portable ultrasound scanner. This proposed segmentation method exhibits an average performance precision of 0.922 and a Dice similarity of 0.928.