Publication Details

Segmentation of Head and Neck Organs at Risk Using CNN with Batch Dice Loss

KODYM Oldřich, ŠPANĚL Michal and HEROUT Adam. Segmentation of Head and Neck Organs at Risk Using CNN with Batch Dice Loss. In: Pattern Recognition, 40th German Conference, GCPR 2018 Proceedings. LNCS, volume 11269, vol. 2018. Stuttgart: Springer International Publishing, 2018, pp. 105-114. ISBN 978-3-030-12938-5. ISSN 0302-9743.
Czech title
Segmentace orgánů hlavy a krku s využitím CNN s batch Dice loss
Type
conference paper
Language
english
Authors
Keywords

Convolutional Neural Networks, Computed Tomography, Multi-label Segmentation, Head and Neck Radiotherapy

Abstract

This paper deals with segmentation of organs at risk (OAR) in head and neck area in CT images which is a crucial step for reliable intensity modulated radiotherapy treatment. We introduce a convolution neural network with encoder-decoder architecture and a new loss function, the batch soft Dice loss function, used to train the network. The resulting model produces segmentations of every OAR in the public MICCAI 2015 Head And Neck Auto-Segmentation Challenge dataset. Despite the heavy class imbalance in the data, we improve accuracy of current state-of-the-art methods by 0.33 mm in terms of average surface distance and by 0.11 in terms of Dice overlap coefficient on average. 

Published
2018
Pages
105-114
Journal
Lecture Notes in Computer Science, vol. 2018, no. 11269, ISSN 0302-9743
Proceedings
Pattern Recognition, 40th German Conference, GCPR 2018 Proceedings
Series
LNCS, volume 11269
Conference
German Conference on Pattern Recognition, Stuttgart, DE
ISBN
978-3-030-12938-5
Publisher
Springer International Publishing
Place
Stuttgart, DE
DOI
EID Scopus
BibTeX
@INPROCEEDINGS{FITPUB11677,
   author = "Old\v{r}ich Kodym and Michal \v{S}pan\v{e}l and Adam Herout",
   title = "Segmentation of Head and Neck Organs at Risk Using CNN with Batch Dice Loss",
   pages = "105--114",
   booktitle = "Pattern Recognition, 40th German Conference, GCPR 2018 Proceedings",
   series = "LNCS, volume 11269",
   journal = "Lecture Notes in Computer Science",
   volume = 2018,
   number = 11269,
   year = 2018,
   location = "Stuttgart, DE",
   publisher = "Springer International Publishing",
   ISBN = "978-3-030-12938-5",
   ISSN = "0302-9743",
   doi = "10.1007/978-3-030-12939-2\_8",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11677"
}
Back to top