@inproceedings{Bai:2017:10.1007/978-3-319-66185-8_29,
author = {Bai, W and Oktay, O and Sinclair, M and Suzuki, H and Rajchl, M and Tarroni, G and Glocker, B and King, A and Matthews, P and Rueckert, D},
doi = {10.1007/978-3-319-66185-8_29},
pages = {253--260},
publisher = {Springer Verlag},
title = {Semi-supervised learning for network-based cardiac MR image segmentation},
url = {http://dx.doi.org/10.1007/978-3-319-66185-8_29},
year = {2017}
}

Download

TY  - CPAPER
AB  - Training a fully convolutional network for pixel-wise (or voxel-wise) image segmentation normally requires a large number of training images with corresponding ground truth label maps. However, it is a challenge to obtain such a large training set in the medical imaging domain, where expert annotations are time-consuming and difficult to obtain. In this paper, we propose a semi-supervised learning approach, in which a segmentation network is trained from both labelled and unlabelled data. The network parameters and the segmentations for the unlabelled data are alternately updated. We evaluate the method for short-axis cardiac MR image segmentation and it has demonstrated a high performance, outperforming a baseline supervised method. The mean Dice overlap metric is 0.92 for the left ventricular cavity, 0.85 for the myocardium and 0.89 for the right ventricular cavity. It also outperforms a state-of-the-art multi-atlas segmentation method by a large margin and the speed is substantially faster.
AU  - Bai,W
AU  - Oktay,O
AU  - Sinclair,M
AU  - Suzuki,H
AU  - Rajchl,M
AU  - Tarroni,G
AU  - Glocker,B
AU  - King,A
AU  - Matthews,P
AU  - Rueckert,D
DO  - 10.1007/978-3-319-66185-8_29
EP  - 260
PB  - Springer Verlag
PY  - 2017///
SN  - 0302-9743
SP  - 253
TI  - Semi-supervised learning for network-based cardiac MR image segmentation
UR  - http://dx.doi.org/10.1007/978-3-319-66185-8_29
UR  - http://hdl.handle.net/10044/1/49165
ER  - 

Download