A Dual Adversarial Calibration Framework for Automatic Fetal Brain Biometry
Abstract
This paper presents a novel approach to automatic fetal brain biometry motivated by needs in low- and medium-income countries. Specifically, we leverage high-end (HE) ultrasound images to build a biometry solution for low-cost (LC) point-of-care ultrasound images. We propose a novel unsupervised domain adaptation approach to train deep models to be invariant to significant image distribution shift between the image types. Our proposed method, which employs a Dual Adversarial Calibration (DAC) framework, consists of adversarial pathways which enforce model invariance to; i) adversarial perturbations in the feature space derived from LC images, and ii) appearance domain discrepancy. Our Dual Adversarial Calibration method estimates transcerebellar diameter and head circumference on images from low-cost ultrasound devices with a mean absolute error (MAE) of 2.43mm and 1.65mm, compared with 7.28 mm and 5.65 mm respectively for SOTA.
Cite
Text
Gao et al. "A Dual Adversarial Calibration Framework for Automatic Fetal Brain Biometry." IEEE/CVF International Conference on Computer Vision Workshops, 2021. doi:10.1109/ICCVW54120.2021.00363Markdown
[Gao et al. "A Dual Adversarial Calibration Framework for Automatic Fetal Brain Biometry." IEEE/CVF International Conference on Computer Vision Workshops, 2021.](https://mlanthology.org/iccvw/2021/gao2021iccvw-dual/) doi:10.1109/ICCVW54120.2021.00363BibTeX
@inproceedings{gao2021iccvw-dual,
title = {{A Dual Adversarial Calibration Framework for Automatic Fetal Brain Biometry}},
author = {Gao, Yuan and Lee, Lok Hin and Droste, Richard and Craik, Rachel and Beriwal, Sridevi and Papageorghiou, Aris T. and Noble, J. Alison},
booktitle = {IEEE/CVF International Conference on Computer Vision Workshops},
year = {2021},
pages = {3239-3247},
doi = {10.1109/ICCVW54120.2021.00363},
url = {https://mlanthology.org/iccvw/2021/gao2021iccvw-dual/}
}