Interpretable and Interactive Deep Multiple Instance Learning for Dental Caries Classification in Bitewing X-rays

Benjamin Bergner, Csaba Rohrer, Aiham Taleb, Martha Duchrau, Guilherme De Leon, Jonas Rodrigues, Falk Schwendicke, Joachim Krois, Christoph Lippert
Proceedings of The 5th International Conference on Medical Imaging with Deep Learning, PMLR 172:130-149, 2022.

Abstract

We propose a simple and efficient image classification architecture based on deep multiple instance learning, and apply it to the challenging task of caries detection in dental radiographs. Technically, our approach contributes in two ways: First, it outputs a heatmap of local patch classification probabilities despite being trained with weak image-level labels. Second, it is amenable to learning from segmentation labels to guide training. In contrast to existing methods, the human user can faithfully interpret predictions and interact with the model to decide which regions to attend to. Experiments are conducted on a large clinical dataset of 38k bitewings (316k teeth), where we achieve competitive performance compared to various baselines. When guided by an external caries segmentation model, a significant improvement in classification and localization performance is observed.

Cite this Paper

BibTeX
@InProceedings{pmlr-v172-bergner22a, title = {Interpretable and Interactive Deep Multiple Instance Learning for Dental Caries Classification in Bitewing X-rays}, author = {Bergner, Benjamin and Rohrer, Csaba and Taleb, Aiham and Duchrau, Martha and De Leon, Guilherme and Rodrigues, Jonas and Schwendicke, Falk and Krois, Joachim and Lippert, Christoph}, booktitle = {Proceedings of The 5th International Conference on Medical Imaging with Deep Learning}, pages = {130--149}, year = {2022}, editor = {Konukoglu, Ender and Menze, Bjoern and Venkataraman, Archana and Baumgartner, Christian and Dou, Qi and Albarqouni, Shadi}, volume = {172}, series = {Proceedings of Machine Learning Research}, month = {06--08 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v172/bergner22a/bergner22a.pdf}, url = {https://proceedings.mlr.press/v172/bergner22a.html}, abstract = {We propose a simple and efficient image classification architecture based on deep multiple instance learning, and apply it to the challenging task of caries detection in dental radiographs. Technically, our approach contributes in two ways: First, it outputs a heatmap of local patch classification probabilities despite being trained with weak image-level labels. Second, it is amenable to learning from segmentation labels to guide training. In contrast to existing methods, the human user can faithfully interpret predictions and interact with the model to decide which regions to attend to. Experiments are conducted on a large clinical dataset of 38k bitewings (316k teeth), where we achieve competitive performance compared to various baselines. When guided by an external caries segmentation model, a significant improvement in classification and localization performance is observed.} }
Endnote
%0 Conference Paper %T Interpretable and Interactive Deep Multiple Instance Learning for Dental Caries Classification in Bitewing X-rays %A Benjamin Bergner %A Csaba Rohrer %A Aiham Taleb %A Martha Duchrau %A Guilherme De Leon %A Jonas Rodrigues %A Falk Schwendicke %A Joachim Krois %A Christoph Lippert %B Proceedings of The 5th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2022 %E Ender Konukoglu %E Bjoern Menze %E Archana Venkataraman %E Christian Baumgartner %E Qi Dou %E Shadi Albarqouni %F pmlr-v172-bergner22a %I PMLR %P 130--149 %U https://proceedings.mlr.press/v172/bergner22a.html %V 172 %X We propose a simple and efficient image classification architecture based on deep multiple instance learning, and apply it to the challenging task of caries detection in dental radiographs. Technically, our approach contributes in two ways: First, it outputs a heatmap of local patch classification probabilities despite being trained with weak image-level labels. Second, it is amenable to learning from segmentation labels to guide training. In contrast to existing methods, the human user can faithfully interpret predictions and interact with the model to decide which regions to attend to. Experiments are conducted on a large clinical dataset of 38k bitewings (316k teeth), where we achieve competitive performance compared to various baselines. When guided by an external caries segmentation model, a significant improvement in classification and localization performance is observed.
APA
Bergner, B., Rohrer, C., Taleb, A., Duchrau, M., De Leon, G., Rodrigues, J., Schwendicke, F., Krois, J. & Lippert, C.. (2022). Interpretable and Interactive Deep Multiple Instance Learning for Dental Caries Classification in Bitewing X-rays. Proceedings of The 5th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 172:130-149 Available from https://proceedings.mlr.press/v172/bergner22a.html.

Related Material