Whole-slide-imaging Cancer Metastases Detection and Localization with Limited Tumorous Data

Yinsheng He, Xingyu Li
Medical Imaging with Deep Learning, PMLR 227:877-887, 2024.

Abstract

Recently, various deep learning methods have shown significant successes in medical image analysis, especially in the detection of cancer metastases in hematoxylin and eosin (H&E) stained whole-slide images (WSIs). However, in order to obtain good performance, these research achievements rely on hundreds of well-annotated WSIs. In this study, we tackle the tumor localization and detection problem under the setting of few labeled whole slide images and introduce a patch-based analysis pipeline based on the latest reverse knowledge distillation architecture. To address the extremely unbalanced normal and tumorous samples in training sample collection, we applied the focal loss formula to the representation similarity metric for model optimization. Compared with prior arts, our method achieves similar performance by less than ten percent of training samples on the public Camelyon16 dataset. In addition, this is the first work that show the great potential of the knowledge distillation models in computational histopathology. Our python implementation will be publically accessible upon paper acceptance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v227-he24a, title = {Whole-slide-imaging Cancer Metastases Detection and Localization with Limited Tumorous Data}, author = {He, Yinsheng and Li, Xingyu}, booktitle = {Medical Imaging with Deep Learning}, pages = {877--887}, year = {2024}, editor = {Oguz, Ipek and Noble, Jack and Li, Xiaoxiao and Styner, Martin and Baumgartner, Christian and Rusu, Mirabela and Heinmann, Tobias and Kontos, Despina and Landman, Bennett and Dawant, Benoit}, volume = {227}, series = {Proceedings of Machine Learning Research}, month = {10--12 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v227/he24a/he24a.pdf}, url = {https://proceedings.mlr.press/v227/he24a.html}, abstract = {Recently, various deep learning methods have shown significant successes in medical image analysis, especially in the detection of cancer metastases in hematoxylin and eosin (H&E) stained whole-slide images (WSIs). However, in order to obtain good performance, these research achievements rely on hundreds of well-annotated WSIs. In this study, we tackle the tumor localization and detection problem under the setting of few labeled whole slide images and introduce a patch-based analysis pipeline based on the latest reverse knowledge distillation architecture. To address the extremely unbalanced normal and tumorous samples in training sample collection, we applied the focal loss formula to the representation similarity metric for model optimization. Compared with prior arts, our method achieves similar performance by less than ten percent of training samples on the public Camelyon16 dataset. In addition, this is the first work that show the great potential of the knowledge distillation models in computational histopathology. Our python implementation will be publically accessible upon paper acceptance.} }
Endnote
%0 Conference Paper %T Whole-slide-imaging Cancer Metastases Detection and Localization with Limited Tumorous Data %A Yinsheng He %A Xingyu Li %B Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2024 %E Ipek Oguz %E Jack Noble %E Xiaoxiao Li %E Martin Styner %E Christian Baumgartner %E Mirabela Rusu %E Tobias Heinmann %E Despina Kontos %E Bennett Landman %E Benoit Dawant %F pmlr-v227-he24a %I PMLR %P 877--887 %U https://proceedings.mlr.press/v227/he24a.html %V 227 %X Recently, various deep learning methods have shown significant successes in medical image analysis, especially in the detection of cancer metastases in hematoxylin and eosin (H&E) stained whole-slide images (WSIs). However, in order to obtain good performance, these research achievements rely on hundreds of well-annotated WSIs. In this study, we tackle the tumor localization and detection problem under the setting of few labeled whole slide images and introduce a patch-based analysis pipeline based on the latest reverse knowledge distillation architecture. To address the extremely unbalanced normal and tumorous samples in training sample collection, we applied the focal loss formula to the representation similarity metric for model optimization. Compared with prior arts, our method achieves similar performance by less than ten percent of training samples on the public Camelyon16 dataset. In addition, this is the first work that show the great potential of the knowledge distillation models in computational histopathology. Our python implementation will be publically accessible upon paper acceptance.
APA
He, Y. & Li, X.. (2024). Whole-slide-imaging Cancer Metastases Detection and Localization with Limited Tumorous Data. Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 227:877-887 Available from https://proceedings.mlr.press/v227/he24a.html.

Related Material