Ultrasonic Image Segmentation Method based on Improved Fully Convolution Network

DOI

10.2991/icmeit-19.2019.93How to use a DOI?

Keywords

Computer vision; Deep learning; Ultrasound image; Image segmentation.

Abstract

The clinical fetal brain ultrasound image contains a lot of noise, which is impact the classification or recognition results in deep learning tasks. Therefore, proposed an improved fully convolutional network which can automatically eliminate noise and extract effective area in fetal brain ultrasound images. By adding dilated convolution, U-net obtained a larger receptive field. The original U-net neural network was compared with the method by adding dilated convolution one. The experiment of test data shows that the improved segmentation model was effective and robust, and performed better in precise rate, recall rate, f1-score and DICE coefficient. This improved method can be wildly used in other medical image segmentation tasks.

Copyright

© 2019, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Hai Ye
AU  - Kaiping Feng
AU  - Lihong Luo
AU  - Hongning Xie
PY  - 2019/04
DA  - 2019/04
TI  - Ultrasonic Image Segmentation Method based on Improved Fully Convolution Network
BT  - Proceedings of the 3rd International Conference on Mechatronics Engineering and Information Technology (ICMEIT 2019)
PB  - Atlantis Press
SP  - 581
EP  - 584
SN  - 2352-538X
UR  - https://doi.org/10.2991/icmeit-19.2019.93
DO  - 10.2991/icmeit-19.2019.93
ID  - Ye2019/04
ER  -