Medical image segmentation is inherently an ambiguous task due to factors
such as partial volumes and variations in anatomical definitions. While in most
cases the segmentation uncertainty is around the border of structures of
interest, there can also be considerable inter-rater differences. The class of
conditional variational autoencoders (cVAE) offers a principled approach to
inferring distributions over plausible segmentations that are conditioned on
input images. Segmentation uncertainty estimated from samples of such
distributions can be more informative than using pixel level probability
scores. In this work, we propose a novel conditional generative model that is
based on conditional Normalizing Flow (cFlow). The basic idea is to increase
the expressivity of the cVAE by introducing a cFlow transformation step after
the encoder. This yields improved approximations of the latent posterior
distribution, allowing the model to capture richer segmentation variations.
With this we show that the quality and diversity of samples obtained from our
conditional generative model is enhanced. Performance of our model, which we
call cFlow Net, is evaluated on two medical imaging datasets demonstrating
substantial improvements in both qualitative and quantitative measures when
compared to a recent cVAE based model.
Description
[2006.02683] Uncertainty quantification in medical image segmentation with Normalizing Flows
%0 Journal Article
%1 selvan2020uncertainty
%A Selvan, Raghavendra
%A Faye, Frederik
%A Middleton, Jon
%A Pai, Akshay
%D 2020
%K flows uncertainty
%T Uncertainty quantification in medical image segmentation with
Normalizing Flows
%U http://arxiv.org/abs/2006.02683
%X Medical image segmentation is inherently an ambiguous task due to factors
such as partial volumes and variations in anatomical definitions. While in most
cases the segmentation uncertainty is around the border of structures of
interest, there can also be considerable inter-rater differences. The class of
conditional variational autoencoders (cVAE) offers a principled approach to
inferring distributions over plausible segmentations that are conditioned on
input images. Segmentation uncertainty estimated from samples of such
distributions can be more informative than using pixel level probability
scores. In this work, we propose a novel conditional generative model that is
based on conditional Normalizing Flow (cFlow). The basic idea is to increase
the expressivity of the cVAE by introducing a cFlow transformation step after
the encoder. This yields improved approximations of the latent posterior
distribution, allowing the model to capture richer segmentation variations.
With this we show that the quality and diversity of samples obtained from our
conditional generative model is enhanced. Performance of our model, which we
call cFlow Net, is evaluated on two medical imaging datasets demonstrating
substantial improvements in both qualitative and quantitative measures when
compared to a recent cVAE based model.
@article{selvan2020uncertainty,
abstract = {Medical image segmentation is inherently an ambiguous task due to factors
such as partial volumes and variations in anatomical definitions. While in most
cases the segmentation uncertainty is around the border of structures of
interest, there can also be considerable inter-rater differences. The class of
conditional variational autoencoders (cVAE) offers a principled approach to
inferring distributions over plausible segmentations that are conditioned on
input images. Segmentation uncertainty estimated from samples of such
distributions can be more informative than using pixel level probability
scores. In this work, we propose a novel conditional generative model that is
based on conditional Normalizing Flow (cFlow). The basic idea is to increase
the expressivity of the cVAE by introducing a cFlow transformation step after
the encoder. This yields improved approximations of the latent posterior
distribution, allowing the model to capture richer segmentation variations.
With this we show that the quality and diversity of samples obtained from our
conditional generative model is enhanced. Performance of our model, which we
call cFlow Net, is evaluated on two medical imaging datasets demonstrating
substantial improvements in both qualitative and quantitative measures when
compared to a recent cVAE based model.},
added-at = {2020-06-05T11:33:40.000+0200},
author = {Selvan, Raghavendra and Faye, Frederik and Middleton, Jon and Pai, Akshay},
biburl = {https://www.bibsonomy.org/bibtex/24480979a3be15ec0ceaf405efe69c9cf/kirk86},
description = {[2006.02683] Uncertainty quantification in medical image segmentation with Normalizing Flows},
interhash = {3c679fa873e0b49cae02eb6bb0963523},
intrahash = {4480979a3be15ec0ceaf405efe69c9cf},
keywords = {flows uncertainty},
note = {cite arxiv:2006.02683Comment: 12 pages},
timestamp = {2020-06-05T11:33:40.000+0200},
title = {Uncertainty quantification in medical image segmentation with
Normalizing Flows},
url = {http://arxiv.org/abs/2006.02683},
year = 2020
}