Skip to main content
Top

Hint

Swipe to navigate through the chapters of this book

2018 | OriginalPaper | Chapter

Perfect MCMC Sampling in Bayesian MRFs for Uncertainty Estimation in Segmentation

share
SHARE

Abstract

Typical segmentation methods produce a single optimal solution and fail to inform about (i) the confidence/uncertainty in the object boundaries or (ii) alternate close-to-optimal solutions. To estimate uncertainty, some methods intend to sample segmentations from an associated posterior model using Markov chain Monte Carlo (MCMC) sampling or perturbation models. However, they cannot guarantee sampling from the true posterior, deviating significantly in practice. We propose a novel method that guarantees exact MCMC sampling, in finite time, of multi-label segmentations from generic Bayesian Markov random field (MRF) models. For exact sampling, we propose Fill’s strategy and extend it to generic MRF models via a novel bounding chain algorithm. Results on simulated data and clinical brain images from 4 classic problems show that our uncertainty estimates gain accuracy over the state of the art.
Literature
1.
go back to reference Alberts, E., et al.: Uncertainty quantification in brain tumor segmentation using CRFs and random perturbation models. In: IEEE International Symposium on Biomedical Imaging, pp. 428–431 (2016) Alberts, E., et al.: Uncertainty quantification in brain tumor segmentation using CRFs and random perturbation models. In: IEEE International Symposium on Biomedical Imaging, pp. 428–431 (2016)
2.
go back to reference Awate, S., Whitaker, R.: Multiatlas segmentation as nonparametric regression. IEEE Trans. Med. Imaging 33(9), 1803–17 (2014) CrossRef Awate, S., Whitaker, R.: Multiatlas segmentation as nonparametric regression. IEEE Trans. Med. Imaging 33(9), 1803–17 (2014) CrossRef
4.
go back to reference Fill, J.: An interruptible algorithm for perfect sampling via Markov chains. Ann. Appl. Prob. 8(1), 131–62 (1998) MathSciNetCrossRef Fill, J.: An interruptible algorithm for perfect sampling via Markov chains. Ann. Appl. Prob. 8(1), 131–62 (1998) MathSciNetCrossRef
6.
go back to reference Kohli, P., Torr, P.: Measuring uncertainty in graph cut solutions. Comp. Vis. Image Underst. 112, 30–8 (2008) CrossRef Kohli, P., Torr, P.: Measuring uncertainty in graph cut solutions. Comp. Vis. Image Underst. 112, 30–8 (2008) CrossRef
7.
go back to reference Kybic, J.: Bootstrap resampling for image registration uncertainty estimation without ground truth. IEEE Trans. Image Process. 19(1), 64–73 (2010) MathSciNetCrossRef Kybic, J.: Bootstrap resampling for image registration uncertainty estimation without ground truth. IEEE Trans. Image Process. 19(1), 64–73 (2010) MathSciNetCrossRef
8.
go back to reference Le, M., Unkelbach, J., Ayache, N., Delingette, H.: Sampling image segmentations for uncertainty quantification. Med. Image Anal. 34, 42–51 (2016) CrossRef Le, M., Unkelbach, J., Ayache, N., Delingette, H.: Sampling image segmentations for uncertainty quantification. Med. Image Anal. 34, 42–51 (2016) CrossRef
9.
go back to reference Papandreou, G., Yuille, A.: Perturb-and-MAP random fields: using discrete optimization to learn and sample from energy models. In: International Conference on Computer Vision (2011) Papandreou, G., Yuille, A.: Perturb-and-MAP random fields: using discrete optimization to learn and sample from energy models. In: International Conference on Computer Vision (2011)
10.
go back to reference Propp, J., Wilson, D.: Exact sampling with coupled Markov chains and applications to statistical mechanics. Random Struct. Algorithms 9(1), 223–52 (1996) MathSciNetCrossRef Propp, J., Wilson, D.: Exact sampling with coupled Markov chains and applications to statistical mechanics. Random Struct. Algorithms 9(1), 223–52 (1996) MathSciNetCrossRef
Metadata
Title
Perfect MCMC Sampling in Bayesian MRFs for Uncertainty Estimation in Segmentation
Authors
Saurabh Garg
Suyash P. Awate
Copyright Year
2018
DOI
https://doi.org/10.1007/978-3-030-00928-1_76

Premium Partner