Towards a Single Solution for Polyp Detection, Localization and Segmentation in Colonoscopy Images

Willem Dijkstra, André Sobiecki, Jorge Bernal, Alexandru C. Telea

2019

Abstract

Colorectal cancer is one of the main causes of cancer death worldwide. Early detection of its precursor lesion, the polyp, is key to ensure patient survival. Despite its gold standard status, colonoscopy presents some drawbacks such as polyp misses. While several computer-based solutions in this direction have been proposed, there is no available solution tackling lesion detection, localization and segmentation at once. We present in this paper a one-shot solution to characterize polyps in colonoscopy images. Our method uses a fully convolutional neural network model for semantic segmentation. Next, we apply transfer learning to provide detection and localization. We tested our method on several public datasets showing promising results, including compliance with technical and clinical requirements needed for an efficient deployment in the exploration room.

Download


Paper Citation


in Harvard Style

Dijkstra W., Sobiecki A., Bernal J. and Telea A. (2019). Towards a Single Solution for Polyp Detection, Localization and Segmentation in Colonoscopy Images.In Proceedings of the 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 4: GIANA, ISBN 978-989-758-354-4, pages 616-625. DOI: 10.5220/0007694906160625


in Bibtex Style

@conference{giana19,
author={Willem Dijkstra and André Sobiecki and Jorge Bernal and Alexandru Telea},
title={Towards a Single Solution for Polyp Detection, Localization and Segmentation in Colonoscopy Images},
booktitle={Proceedings of the 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 4: GIANA,},
year={2019},
pages={616-625},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0007694906160625},
isbn={978-989-758-354-4},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 4: GIANA,
TI - Towards a Single Solution for Polyp Detection, Localization and Segmentation in Colonoscopy Images
SN - 978-989-758-354-4
AU - Dijkstra W.
AU - Sobiecki A.
AU - Bernal J.
AU - Telea A.
PY - 2019
SP - 616
EP - 625
DO - 10.5220/0007694906160625