MedSelect: Selective Labeling for Medical Image Classification Using Meta-Learning

Damir Vrabac, Akshay Smit, Yujie He, Andrew Y. Ng, Andrew L. Beam, Pranav Rajpurkar
Proceedings of The 5th International Conference on Medical Imaging with Deep Learning, PMLR 172:1301-1310, 2022.

Abstract

We propose a selective labeling method using meta-learning for medical image interpretation in the setting of limited labeling resources. Our method, MedSelect, consists of a trainable deep learning model that uses image embeddings to select images to label, and a non-parametric classifier that uses cosine similarity to classify unseen images. We demonstrate that MedSelect learns an effective selection strategy outperforming baseline selection strategies across seen and unseen medical conditions for chest X-ray interpretation. We also perform an analysis of the selections performed by MedSelect comparing the distribution of latent embeddings and clinical features, and find significant differences compared to the strongest performing baseline. Our method is broadly applicable across medical imaging tasks where labels are expensive to acquire.

Cite this Paper

BibTeX
@InProceedings{pmlr-v172-vrabac22a, title = {MedSelect: Selective Labeling for Medical Image Classification Using Meta-Learning}, author = {Vrabac, Damir and Smit, Akshay and He, Yujie and Ng, Andrew Y. and Beam, Andrew L. and Rajpurkar, Pranav}, booktitle = {Proceedings of The 5th International Conference on Medical Imaging with Deep Learning}, pages = {1301--1310}, 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/vrabac22a/vrabac22a.pdf}, url = {https://proceedings.mlr.press/v172/vrabac22a.html}, abstract = {We propose a selective labeling method using meta-learning for medical image interpretation in the setting of limited labeling resources. Our method, MedSelect, consists of a trainable deep learning model that uses image embeddings to select images to label, and a non-parametric classifier that uses cosine similarity to classify unseen images. We demonstrate that MedSelect learns an effective selection strategy outperforming baseline selection strategies across seen and unseen medical conditions for chest X-ray interpretation. We also perform an analysis of the selections performed by MedSelect comparing the distribution of latent embeddings and clinical features, and find significant differences compared to the strongest performing baseline. Our method is broadly applicable across medical imaging tasks where labels are expensive to acquire.} }
Endnote
%0 Conference Paper %T MedSelect: Selective Labeling for Medical Image Classification Using Meta-Learning %A Damir Vrabac %A Akshay Smit %A Yujie He %A Andrew Y. Ng %A Andrew L. Beam %A Pranav Rajpurkar %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-vrabac22a %I PMLR %P 1301--1310 %U https://proceedings.mlr.press/v172/vrabac22a.html %V 172 %X We propose a selective labeling method using meta-learning for medical image interpretation in the setting of limited labeling resources. Our method, MedSelect, consists of a trainable deep learning model that uses image embeddings to select images to label, and a non-parametric classifier that uses cosine similarity to classify unseen images. We demonstrate that MedSelect learns an effective selection strategy outperforming baseline selection strategies across seen and unseen medical conditions for chest X-ray interpretation. We also perform an analysis of the selections performed by MedSelect comparing the distribution of latent embeddings and clinical features, and find significant differences compared to the strongest performing baseline. Our method is broadly applicable across medical imaging tasks where labels are expensive to acquire.
APA
Vrabac, D., Smit, A., He, Y., Ng, A.Y., Beam, A.L. & Rajpurkar, P.. (2022). MedSelect: Selective Labeling for Medical Image Classification Using Meta-Learning. Proceedings of The 5th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 172:1301-1310 Available from https://proceedings.mlr.press/v172/vrabac22a.html.

Related Material