Conditional Variational Autoencoder for Learned Image Reconstruction

Learned image reconstruction techniques using deep neural networks have recently gained popularity and have delivered promising empirical results. However, most approaches focus on one single recovery for each observation, and thus neglect information uncertainty. In this work, we develop a novel co...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Computation Ročník 9; číslo 11; s. 114
Hlavní autoři: Zhang, Chen, Barbano, Riccardo, Jin, Bangti
Médium: Journal Article
Jazyk:angličtina
Vydáno: Basel MDPI AG 01.11.2021
Témata:
Artificial neural networks
conditional variational autoencoder
deep learning
Epistemology
Image reconstruction
Inverse problems
Machine learning
Mathematical analysis
Medical imaging
Neural networks
Positron emission
Random variables
Statistical methods
Uncertainty
uncertainty quantification
ISSN:2079-3197, 2079-3197
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Learned image reconstruction techniques using deep neural networks have recently gained popularity and have delivered promising empirical results. However, most approaches focus on one single recovery for each observation, and thus neglect information uncertainty. In this work, we develop a novel computational framework that approximates the posterior distribution of the unknown image at each query observation. The proposed framework is very flexible: it handles implicit noise models and priors, it incorporates the data formation process (i.e., the forward operator), and the learned reconstructive properties are transferable between different datasets. Once the network is trained using the conditional variational autoencoder loss, it provides a computationally efficient sampler for the approximate posterior distribution via feed-forward propagation, and the summarizing statistics of the generated samples are used for both point-estimation and uncertainty quantification. We illustrate the proposed framework with extensive numerical experiments on positron emission tomography (with both moderate and low-count levels) showing that the framework generates high-quality samples when compared with state-of-the-art methods.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2079-3197
2079-3197
DOI:10.3390/computation9110114