Prediction of KRAS mutation status from H&E foundation model embeddings in non-small cell lung cancer

Marc Robbins, Jessica Loo, Saurabh Vyawahare, Yang Von Wang, Carson Mcneil, Dave Steiner, Sudha Rao, Pok Fai Wong, Ehud Rivlin, Shamira Weaver, Roman Goldenberg
Proceedings of the MICCAI Workshop on Computational Pathology, PMLR 254:170-179, 2024.

Abstract

We predicted KRAS mutation status on non-small cell lung cancer (NSCLC) H&E images from foundation model embeddings. We evaluated a variety of attention-based multiple instance learning (MIL) models and aggregation strategies for a tilewise linear classifier. MIL with self-attention performed the best (AUC=0.822) followed by the minimum over tiles classified with the linear model (AUC=0.810). Self-attention was necessary for MIL to surpass tilewise linear classification when a wide range of aggregation techniques was considered.

Cite this Paper

BibTeX
@InProceedings{pmlr-v254-robbins24a, title = {Prediction of {KRAS} mutation status from H&E foundation model embeddings in non-small cell lung cancer}, author = {Robbins, Marc and Loo, Jessica and Vyawahare, Saurabh and Wang, Yang Von and Mcneil, Carson and Steiner, Dave and Rao, Sudha and Wong, Pok Fai and Rivlin, Ehud and Weaver, Shamira and Goldenberg, Roman}, booktitle = {Proceedings of the MICCAI Workshop on Computational Pathology}, pages = {170--179}, year = {2024}, editor = {Ciompi, Francesco and Khalili, Nadieh and Studer, Linda and Poceviciute, Milda and Khan, Amjad and Veta, Mitko and Jiao, Yiping and Haj-Hosseini, Neda and Chen, Hao and Raza, Shan and Minhas, FayyazZlobec, Inti and Burlutskiy, Nikolay and Vilaplana, Veronica and Brattoli, Biagio and Muller, Henning and Atzori, Manfredo and Raza, Shan and Minhas, Fayyaz}, volume = {254}, series = {Proceedings of Machine Learning Research}, month = {06 Oct}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v254/main/assets/robbins24a/robbins24a.pdf}, url = {https://proceedings.mlr.press/v254/robbins24a.html}, abstract = {We predicted KRAS mutation status on non-small cell lung cancer (NSCLC) H&E images from foundation model embeddings. We evaluated a variety of attention-based multiple instance learning (MIL) models and aggregation strategies for a tilewise linear classifier. MIL with self-attention performed the best (AUC=0.822) followed by the minimum over tiles classified with the linear model (AUC=0.810). Self-attention was necessary for MIL to surpass tilewise linear classification when a wide range of aggregation techniques was considered.} }
Endnote
%0 Conference Paper %T Prediction of KRAS mutation status from H&E foundation model embeddings in non-small cell lung cancer %A Marc Robbins %A Jessica Loo %A Saurabh Vyawahare %A Yang Von Wang %A Carson Mcneil %A Dave Steiner %A Sudha Rao %A Pok Fai Wong %A Ehud Rivlin %A Shamira Weaver %A Roman Goldenberg %B Proceedings of the MICCAI Workshop on Computational Pathology %C Proceedings of Machine Learning Research %D 2024 %E Francesco Ciompi %E Nadieh Khalili %E Linda Studer %E Milda Poceviciute %E Amjad Khan %E Mitko Veta %E Yiping Jiao %E Neda Haj-Hosseini %E Hao Chen %E Shan Raza %E Fayyaz MinhasInti Zlobec %E Nikolay Burlutskiy %E Veronica Vilaplana %E Biagio Brattoli %E Henning Muller %E Manfredo Atzori %E Shan Raza %E Fayyaz Minhas %F pmlr-v254-robbins24a %I PMLR %P 170--179 %U https://proceedings.mlr.press/v254/robbins24a.html %V 254 %X We predicted KRAS mutation status on non-small cell lung cancer (NSCLC) H&E images from foundation model embeddings. We evaluated a variety of attention-based multiple instance learning (MIL) models and aggregation strategies for a tilewise linear classifier. MIL with self-attention performed the best (AUC=0.822) followed by the minimum over tiles classified with the linear model (AUC=0.810). Self-attention was necessary for MIL to surpass tilewise linear classification when a wide range of aggregation techniques was considered.
APA
Robbins, M., Loo, J., Vyawahare, S., Wang, Y.V., Mcneil, C., Steiner, D., Rao, S., Wong, P.F., Rivlin, E., Weaver, S. & Goldenberg, R.. (2024). Prediction of KRAS mutation status from H&E foundation model embeddings in non-small cell lung cancer. Proceedings of the MICCAI Workshop on Computational Pathology, in Proceedings of Machine Learning Research 254:170-179 Available from https://proceedings.mlr.press/v254/robbins24a.html.

Related Material