Uncertainty quantification in non-rigid image registration via
stochastic gradient Markov chain Monte Carlo
stochastic gradient Markov chain Monte Carlo
File(s)
Working Paper
Author(s)
Type
Working Paper
Abstract
We develop a new Bayesian model for non-rigid registration of
three-dimensional medical images, with a focus on uncertainty quantification.
Probabilistic registration of large images with calibrated uncertainty
estimates is difficult for both computational and modelling reasons. To address
the computational issues, we explore connections between the Markov chain Monte
Carlo by backpropagation and the variational inference by backpropagation
frameworks, in order to efficiently draw samples from the posterior
distribution of transformation parameters. To address the modelling issues, we
formulate a Bayesian model for image registration that overcomes the existing
barriers when using a dense, high-dimensional, and diffeomorphic transformation
parametrisation. This results in improved calibration of uncertainty estimates.
We compare the model in terms of both image registration accuracy and
uncertainty quantification to VoxelMorph, a state-of-the-art image registration
model based on deep learning.
three-dimensional medical images, with a focus on uncertainty quantification.
Probabilistic registration of large images with calibrated uncertainty
estimates is difficult for both computational and modelling reasons. To address
the computational issues, we explore connections between the Markov chain Monte
Carlo by backpropagation and the variational inference by backpropagation
frameworks, in order to efficiently draw samples from the posterior
distribution of transformation parameters. To address the modelling issues, we
formulate a Bayesian model for image registration that overcomes the existing
barriers when using a dense, high-dimensional, and diffeomorphic transformation
parametrisation. This results in improved calibration of uncertainty estimates.
We compare the model in terms of both image registration accuracy and
uncertainty quantification to VoxelMorph, a state-of-the-art image registration
model based on deep learning.
Date Issued
2022-05-11
Citation
2022
Publisher
ArXiv
Copyright Statement
© 2022 The Author(s)
Sponsor
Engineering and Physical Sciences Research Council (EPSRC)
Engineering & Physical Science Research Council (EPSRC)
Identifier
http://arxiv.org/abs/2110.13289v1
Grant Number
EP/S013687/1
EP/S013687/1
Subjects
cs.CV
cs.CV
Notes
MELBA Special Issue: Uncertainty for Safe Utilization of Machine Learning in Medical Imaging (UNSURE) 2020
Publication Status
Published