Deep Learning Approach to Semantic Segmentation in 3D Point Cloud Intra-oral Scans of Teeth

Farhad Ghazvinian Zanjani, David Anssari Moin, Bas Verheij, Frank Claessen, Teo Cherici, Tao Tan, Peter H. N. de With
; Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, PMLR 102:557-571, 2019.

Abstract

Accurate segmentation of data, derived from intra-oral scans (IOS), is a crucial step in a computer-aided design (CAD) system for many clinical tasks, such as implantology and orthodontics in modern dentistry. In order to reach the highest possible quality, a segmentation model may process a point cloud derived from an IOS in its highest available spatial resolution, especially for performing a valid analysis in finely detailed regions such as the curvatures in border lines between two teeth. In this paper, we propose an end-to-end deep learning framework for semantic segmentation of individual teeth as well as the gingiva from point clouds representing IOS. By introducing a non-uniform resampling technique, our proposed model is trained and deployed on the highest available spatial resolution where it learns the local fine details along with the global coarse structure of IOS. Furthermore, the point-wise cross-entropy loss for semantic segmentation of a point cloud is an ill-posed problem, since the relative geometrical structures between the instances (e.g. the teeth) are not formulated. By training a secondary simple network as a discriminator in an adversarial setting and penalizing unrealistic arrangements of assigned labels to the teeth on the dental arch, we improve the segmentation results considerably. Hence, a heavy post-processing stage for relational and dependency modeling (e.g. iterative energy minimization of a constructed graph) is not required anymore. Our experiments show that the proposed approach improves the performance of our baseline network and outperforms the state-of-the-art networks by achieving $0.94$ IOU score.

Cite this Paper


BibTeX
@InProceedings{pmlr-v102-ghazvinian-zanjani19a, title = {Deep Learning Approach to Semantic Segmentation in 3D Point Cloud Intra-oral Scans of Teeth}, author = {{Ghazvinian Zanjani}, Farhad and {Anssari Moin}, David and {Verheij}, Bas and {Claessen}, Frank and {Cherici}, Teo and {Tan}, Tao and {de With}, {Peter H. N.}}, booktitle = {Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning}, pages = {557--571}, year = {2019}, editor = {M. Jorge Cardoso and Aasa Feragen and Ben Glocker and Ender Konukoglu and Ipek Oguz and Gozde Unal and Tom Vercauteren}, volume = {102}, series = {Proceedings of Machine Learning Research}, address = {London, United Kingdom}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v102/ghazvinian-zanjani19a/ghazvinian-zanjani19a.pdf}, url = {http://proceedings.mlr.press/v102/ghazvinian-zanjani19a.html}, abstract = {Accurate segmentation of data, derived from intra-oral scans (IOS), is a crucial step in a computer-aided design (CAD) system for many clinical tasks, such as implantology and orthodontics in modern dentistry. In order to reach the highest possible quality, a segmentation model may process a point cloud derived from an IOS in its highest available spatial resolution, especially for performing a valid analysis in finely detailed regions such as the curvatures in border lines between two teeth. In this paper, we propose an end-to-end deep learning framework for semantic segmentation of individual teeth as well as the gingiva from point clouds representing IOS. By introducing a non-uniform resampling technique, our proposed model is trained and deployed on the highest available spatial resolution where it learns the local fine details along with the global coarse structure of IOS. Furthermore, the point-wise cross-entropy loss for semantic segmentation of a point cloud is an ill-posed problem, since the relative geometrical structures between the instances (e.g. the teeth) are not formulated. By training a secondary simple network as a discriminator in an adversarial setting and penalizing unrealistic arrangements of assigned labels to the teeth on the dental arch, we improve the segmentation results considerably. Hence, a heavy post-processing stage for relational and dependency modeling (e.g. iterative energy minimization of a constructed graph) is not required anymore. Our experiments show that the proposed approach improves the performance of our baseline network and outperforms the state-of-the-art networks by achieving $0.94$ IOU score.} }
Endnote
%0 Conference Paper %T Deep Learning Approach to Semantic Segmentation in 3D Point Cloud Intra-oral Scans of Teeth %A Farhad Ghazvinian Zanjani %A David Anssari Moin %A Bas Verheij %A Frank Claessen %A Teo Cherici %A Tao Tan %A Peter H. N. de With %B Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2019 %E M. Jorge Cardoso %E Aasa Feragen %E Ben Glocker %E Ender Konukoglu %E Ipek Oguz %E Gozde Unal %E Tom Vercauteren %F pmlr-v102-ghazvinian-zanjani19a %I PMLR %J Proceedings of Machine Learning Research %P 557--571 %U http://proceedings.mlr.press %V 102 %W PMLR %X Accurate segmentation of data, derived from intra-oral scans (IOS), is a crucial step in a computer-aided design (CAD) system for many clinical tasks, such as implantology and orthodontics in modern dentistry. In order to reach the highest possible quality, a segmentation model may process a point cloud derived from an IOS in its highest available spatial resolution, especially for performing a valid analysis in finely detailed regions such as the curvatures in border lines between two teeth. In this paper, we propose an end-to-end deep learning framework for semantic segmentation of individual teeth as well as the gingiva from point clouds representing IOS. By introducing a non-uniform resampling technique, our proposed model is trained and deployed on the highest available spatial resolution where it learns the local fine details along with the global coarse structure of IOS. Furthermore, the point-wise cross-entropy loss for semantic segmentation of a point cloud is an ill-posed problem, since the relative geometrical structures between the instances (e.g. the teeth) are not formulated. By training a secondary simple network as a discriminator in an adversarial setting and penalizing unrealistic arrangements of assigned labels to the teeth on the dental arch, we improve the segmentation results considerably. Hence, a heavy post-processing stage for relational and dependency modeling (e.g. iterative energy minimization of a constructed graph) is not required anymore. Our experiments show that the proposed approach improves the performance of our baseline network and outperforms the state-of-the-art networks by achieving $0.94$ IOU score.
APA
Ghazvinian Zanjani, F., Anssari Moin, D., Verheij, B., Claessen, F., Cherici, T., Tan, T. & de With, P.H.N.. (2019). Deep Learning Approach to Semantic Segmentation in 3D Point Cloud Intra-oral Scans of Teeth. Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, in PMLR 102:557-571

Related Material