Accurate segmentation of dental panoramic radiographs with u-NETS

Thorbjorn Louring Koch; Mathis Perslev; Christian Igel; Sami Sebastian Brandt

doi:10.1109/ISBI.2019.8759563

Accurate segmentation of dental panoramic radiographs with u-NETS

Thorbjorn Louring Koch, Mathis Perslev, Christian Igel, Sami Sebastian Brandt

10 Citations (Scopus)

Abstract

Fully convolutional neural networks (FCNs) have proven to be powerful tools for medical image segmentation. We apply an FCN based on the U-Net architecture for the challenging task of semantic segmentation of dental panoramic radiographs and discuss general tricks for improving segmentation performance. Among those are network ensembling, test-time augmentation, data symmetry exploitation and bootstrapping of low quality annotations. The performance of our approach was tested on a highly variable dataset of 1500 dental panoramic radiographs. A single network reached the Dice score of 0.934 where 1201 images were used for training, forming an ensemble increased the score to 0.936.

Original language	English
Title of host publication	ISBI 2019 - 2019 IEEE International Symposium on Biomedical Imaging
Publisher	IEEE
Publication date	2019
Pages	15-19
ISBN (Electronic)	9781538636411
DOIs	https://doi.org/10.1109/ISBI.2019.8759563
Publication status	Published - 2019
Event	16th IEEE International Symposium on Biomedical Imaging, ISBI 2019 - Venice, Italy Duration: 8 Apr 2019 → 11 Apr 2019

Conference

Conference	16th IEEE International Symposium on Biomedical Imaging, ISBI 2019
Country/Territory	Italy
City	Venice
Period	08/04/2019 → 11/04/2019
Sponsor	et al., IEEE Engineering in Medicine and Biology Society (EMB), IEEE Signal Processing Society, The Institute of Electrical and Electronics Engineers (IEEE), UAI, United Imaging Intelligence

Keywords

Deep learning
Dental radiography
Fully convolutional neural network
Pantomogram
Segmentation

Access to Document

10.1109/ISBI.2019.8759563

Cite this

@inproceedings{967fee898e1c48c181306435c1aa0ce0,

title = "Accurate segmentation of dental panoramic radiographs with u-NETS",

abstract = "Fully convolutional neural networks (FCNs) have proven to be powerful tools for medical image segmentation. We apply an FCN based on the U-Net architecture for the challenging task of semantic segmentation of dental panoramic radiographs and discuss general tricks for improving segmentation performance. Among those are network ensembling, test-time augmentation, data symmetry exploitation and bootstrapping of low quality annotations. The performance of our approach was tested on a highly variable dataset of 1500 dental panoramic radiographs. A single network reached the Dice score of 0.934 where 1201 images were used for training, forming an ensemble increased the score to 0.936.",

keywords = "Deep learning, Dental radiography, Fully convolutional neural network, Pantomogram, Segmentation",

author = "Koch, {Thorbjorn Louring} and Mathis Perslev and Christian Igel and Brandt, {Sami Sebastian}",

year = "2019",

doi = "10.1109/ISBI.2019.8759563",

language = "English",

pages = "15--19",

booktitle = "ISBI 2019 - 2019 IEEE International Symposium on Biomedical Imaging",

publisher = "IEEE",

note = "16th IEEE International Symposium on Biomedical Imaging, ISBI 2019 ; Conference date: 08-04-2019 Through 11-04-2019",

}

TY - GEN

T1 - Accurate segmentation of dental panoramic radiographs with u-NETS

AU - Koch, Thorbjorn Louring

AU - Perslev, Mathis

AU - Igel, Christian

AU - Brandt, Sami Sebastian

PY - 2019

Y1 - 2019

N2 - Fully convolutional neural networks (FCNs) have proven to be powerful tools for medical image segmentation. We apply an FCN based on the U-Net architecture for the challenging task of semantic segmentation of dental panoramic radiographs and discuss general tricks for improving segmentation performance. Among those are network ensembling, test-time augmentation, data symmetry exploitation and bootstrapping of low quality annotations. The performance of our approach was tested on a highly variable dataset of 1500 dental panoramic radiographs. A single network reached the Dice score of 0.934 where 1201 images were used for training, forming an ensemble increased the score to 0.936.

AB - Fully convolutional neural networks (FCNs) have proven to be powerful tools for medical image segmentation. We apply an FCN based on the U-Net architecture for the challenging task of semantic segmentation of dental panoramic radiographs and discuss general tricks for improving segmentation performance. Among those are network ensembling, test-time augmentation, data symmetry exploitation and bootstrapping of low quality annotations. The performance of our approach was tested on a highly variable dataset of 1500 dental panoramic radiographs. A single network reached the Dice score of 0.934 where 1201 images were used for training, forming an ensemble increased the score to 0.936.

KW - Deep learning

KW - Dental radiography

KW - Fully convolutional neural network

KW - Pantomogram

KW - Segmentation

UR - http://www.scopus.com/inward/record.url?scp=85073091535&partnerID=8YFLogxK

U2 - 10.1109/ISBI.2019.8759563

DO - 10.1109/ISBI.2019.8759563

M3 - Article in proceedings

AN - SCOPUS:85073091535

SP - 15

EP - 19

BT - ISBI 2019 - 2019 IEEE International Symposium on Biomedical Imaging

PB - IEEE

T2 - 16th IEEE International Symposium on Biomedical Imaging, ISBI 2019

Y2 - 8 April 2019 through 11 April 2019

ER -

Accurate segmentation of dental panoramic radiographs with u-NETS

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this