An eye-tracking inspired method for standardised plane extraction from fetal abdominal ultrasound volumes
Lists
Only you see this
Abstract
We present an eye-tracking inspired method for the automatic extraction of standardised image planes from 3D fetal abdominal ultrasound volumes. Firstly, we recorded the eye movements of 10 clinicians navigating 150 fetal abdominal ultrasound volumes to find the standardised plane. We derived a pictorial structures model to mimic their behavior; candidate anatomical landmarks were identified by an integral channel features based sliding window detector trained on 1000 2D fetal ultrasound images, and their optimal configuration determined using a probabilistic model trained on the relative positions of anatomical landmarks in the same 1000 images. When tested on a further 80 3D fetal ultrasound volumes, our algorithm performed better than state-of-the-art methods which did not use knowledge of eye movements, achieving a standardised plane extraction accuracy of 92.5% and demonstrating, for the first time, that eye-tracking is a powerful tool for informing the design of computer vision algorithms for 3D ultrasound image analysis.
BibTeX
@inproceedings{ahmed2016eye,
author = {Maryam Ahmed and J. Alison Noble},
title = {{An eye-tracking inspired method for standardised plane extraction from fetal abdominal ultrasound volumes}},
booktitle = {ISBI},
year = {2016},
doi = {10.1109/isbi.2016.7493454},
}Loading related content...