Attention-Guided Prostate Lesion Localization and Grade Group Classification with Multiple Instance Learning

Ekaterina Redekop, Karthik V. Sarma, Adam Kinnaird, Anthony Sisk, Steven S. Raman, Leonard S. Marks, William Speier, Corey W. Arnold
Proceedings of The 5th International Conference on Medical Imaging with Deep Learning, PMLR 172:975-987, 2022.

Abstract

Lesion localization is a component of prostate magnetic resonance imaging (MRI) evaluation and is essential for targeted biopsy by enabling registration with real-time ultrasound. Most previous work on prostate cancer localization has focused on classification or segmentation assuming the availability of radiology annotations. In this work, we propose to use an unsupervised attention-based multiple instance learning (MIL) method in an application for the classification and localization of clinically significant prostate cancer. We train our model end-to-end with only image-level labels instead of relying on voxel-level annotations. We extend MIL method by operating both on patches and the whole size images to learn local and global features, which improves classification and localization performance. To better leverage the relationships between multi-modal data, we use an architecture with multiple encoding paths, where each path processes one image modality. The model was developed on a dataset containing 986 multiparametric prostate MRIs and achieved $0.75 \pm 0.03$ AUROC using 3-fold cross-validation in prostate cancer Grade Group classification. Lesion localization analysis showed 70-80% sensitivity for GG $\ge$ 3 at less than one false positive (FP) per patient and 65% of GG2 at one FP per patient.

Cite this Paper


BibTeX
@InProceedings{pmlr-v172-redekop22a, title = {Attention-Guided Prostate Lesion Localization and Grade Group Classification with Multiple Instance Learning}, author = {Redekop, Ekaterina and Sarma, Karthik V. and Kinnaird, Adam and Sisk, Anthony and Raman, Steven S. and Marks, Leonard S. and Speier, William and Arnold, Corey W.}, booktitle = {Proceedings of The 5th International Conference on Medical Imaging with Deep Learning}, pages = {975--987}, 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/redekop22a/redekop22a.pdf}, url = {https://proceedings.mlr.press/v172/redekop22a.html}, abstract = {Lesion localization is a component of prostate magnetic resonance imaging (MRI) evaluation and is essential for targeted biopsy by enabling registration with real-time ultrasound. Most previous work on prostate cancer localization has focused on classification or segmentation assuming the availability of radiology annotations. In this work, we propose to use an unsupervised attention-based multiple instance learning (MIL) method in an application for the classification and localization of clinically significant prostate cancer. We train our model end-to-end with only image-level labels instead of relying on voxel-level annotations. We extend MIL method by operating both on patches and the whole size images to learn local and global features, which improves classification and localization performance. To better leverage the relationships between multi-modal data, we use an architecture with multiple encoding paths, where each path processes one image modality. The model was developed on a dataset containing 986 multiparametric prostate MRIs and achieved $0.75 \pm 0.03$ AUROC using 3-fold cross-validation in prostate cancer Grade Group classification. Lesion localization analysis showed 70-80% sensitivity for GG $\ge$ 3 at less than one false positive (FP) per patient and 65% of GG2 at one FP per patient.} }
Endnote
%0 Conference Paper %T Attention-Guided Prostate Lesion Localization and Grade Group Classification with Multiple Instance Learning %A Ekaterina Redekop %A Karthik V. Sarma %A Adam Kinnaird %A Anthony Sisk %A Steven S. Raman %A Leonard S. Marks %A William Speier %A Corey W. Arnold %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-redekop22a %I PMLR %P 975--987 %U https://proceedings.mlr.press/v172/redekop22a.html %V 172 %X Lesion localization is a component of prostate magnetic resonance imaging (MRI) evaluation and is essential for targeted biopsy by enabling registration with real-time ultrasound. Most previous work on prostate cancer localization has focused on classification or segmentation assuming the availability of radiology annotations. In this work, we propose to use an unsupervised attention-based multiple instance learning (MIL) method in an application for the classification and localization of clinically significant prostate cancer. We train our model end-to-end with only image-level labels instead of relying on voxel-level annotations. We extend MIL method by operating both on patches and the whole size images to learn local and global features, which improves classification and localization performance. To better leverage the relationships between multi-modal data, we use an architecture with multiple encoding paths, where each path processes one image modality. The model was developed on a dataset containing 986 multiparametric prostate MRIs and achieved $0.75 \pm 0.03$ AUROC using 3-fold cross-validation in prostate cancer Grade Group classification. Lesion localization analysis showed 70-80% sensitivity for GG $\ge$ 3 at less than one false positive (FP) per patient and 65% of GG2 at one FP per patient.
APA
Redekop, E., Sarma, K.V., Kinnaird, A., Sisk, A., Raman, S.S., Marks, L.S., Speier, W. & Arnold, C.W.. (2022). Attention-Guided Prostate Lesion Localization and Grade Group Classification with Multiple Instance Learning. Proceedings of The 5th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 172:975-987 Available from https://proceedings.mlr.press/v172/redekop22a.html.

Related Material