Uncertainty-Aware Training of Neural Networks for Selective Medical Image Segmentation

Yukun Ding, Jinglan Liu, Xiaowei Xu, Meiping Huang, Jian Zhuang, Jinjun Xiong, Yiyu Shi
; Proceedings of the Third Conference on Medical Imaging with Deep Learning, PMLR 121:156-173, 2020.

Abstract

State-of-the-art deep learning based methods have achieved remarkable performance on medical image segmentation. Their applications in the clinical setting are, however, limited due to the lack of trustworthiness and reliability. Selective image segmentation has been proposed to address this issue by letting a DNN model process instances with high confidence while referring difficult ones with high uncertainty to experienced radiologists. As such, the model performance is only affected by the predictions on the high confidence subset rather than the whole dataset. Existing selective segmentation methods, however, ignore this unique property of selective segmentation and train their DNN models by optimizing accuracy on the entire dataset. Motivated by such a discrepancy, we present a novel method in this paper that considers such uncertainty in the training process to maximize the accuracy on the confident subset rather than the accuracy on the whole dataset. Experimental results using the whole heart and great vessel segmentation and gland segmentation show that such a training scheme can significantly improve the performance of selective segmentation.

Cite this Paper


BibTeX
@InProceedings{pmlr-v121-ding20a, title = {Uncertainty-Aware Training of Neural Networks for Selective Medical Image Segmentation}, author = {Ding, Yukun and Liu, Jinglan and Xu, Xiaowei and Huang, Meiping and Zhuang, Jian and Xiong, Jinjun and Shi, Yiyu}, pages = {156--173}, year = {2020}, editor = {Tal Arbel and Ismail Ben Ayed and Marleen de Bruijne and Maxime Descoteaux and Herve Lombaert and Christopher Pal}, volume = {121}, series = {Proceedings of Machine Learning Research}, address = {Montreal, QC, Canada}, month = {06--08 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v121/ding20a/ding20a.pdf}, url = {http://proceedings.mlr.press/v121/ding20a.html}, abstract = {State-of-the-art deep learning based methods have achieved remarkable performance on medical image segmentation. Their applications in the clinical setting are, however, limited due to the lack of trustworthiness and reliability. Selective image segmentation has been proposed to address this issue by letting a DNN model process instances with high confidence while referring difficult ones with high uncertainty to experienced radiologists. As such, the model performance is only affected by the predictions on the high confidence subset rather than the whole dataset. Existing selective segmentation methods, however, ignore this unique property of selective segmentation and train their DNN models by optimizing accuracy on the entire dataset. Motivated by such a discrepancy, we present a novel method in this paper that considers such uncertainty in the training process to maximize the accuracy on the confident subset rather than the accuracy on the whole dataset. Experimental results using the whole heart and great vessel segmentation and gland segmentation show that such a training scheme can significantly improve the performance of selective segmentation.} }
Endnote
%0 Conference Paper %T Uncertainty-Aware Training of Neural Networks for Selective Medical Image Segmentation %A Yukun Ding %A Jinglan Liu %A Xiaowei Xu %A Meiping Huang %A Jian Zhuang %A Jinjun Xiong %A Yiyu Shi %B Proceedings of the Third Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2020 %E Tal Arbel %E Ismail Ben Ayed %E Marleen de Bruijne %E Maxime Descoteaux %E Herve Lombaert %E Christopher Pal %F pmlr-v121-ding20a %I PMLR %J Proceedings of Machine Learning Research %P 156--173 %U http://proceedings.mlr.press %V 121 %W PMLR %X State-of-the-art deep learning based methods have achieved remarkable performance on medical image segmentation. Their applications in the clinical setting are, however, limited due to the lack of trustworthiness and reliability. Selective image segmentation has been proposed to address this issue by letting a DNN model process instances with high confidence while referring difficult ones with high uncertainty to experienced radiologists. As such, the model performance is only affected by the predictions on the high confidence subset rather than the whole dataset. Existing selective segmentation methods, however, ignore this unique property of selective segmentation and train their DNN models by optimizing accuracy on the entire dataset. Motivated by such a discrepancy, we present a novel method in this paper that considers such uncertainty in the training process to maximize the accuracy on the confident subset rather than the accuracy on the whole dataset. Experimental results using the whole heart and great vessel segmentation and gland segmentation show that such a training scheme can significantly improve the performance of selective segmentation.
APA
Ding, Y., Liu, J., Xu, X., Huang, M., Zhuang, J., Xiong, J. & Shi, Y.. (2020). Uncertainty-Aware Training of Neural Networks for Selective Medical Image Segmentation. Proceedings of the Third Conference on Medical Imaging with Deep Learning, in PMLR 121:156-173

Related Material