PaperSwipe

Conditional Normalizing Flows for Low-Dose Computed Tomography Image Reconstruction

Published 5 years agoVersion 1arXiv:2006.06270

Authors

Alexander Denker, Maximilian Schmidt, Johannes Leuschner, Peter Maass, Jens Behrmann

Categories

eess.IV

Abstract

Image reconstruction from computed tomography (CT) measurement is a challenging statistical inverse problem since a high-dimensional conditional distribution needs to be estimated. Based on training data obtained from high-quality reconstructions, we aim to learn a conditional density of images from noisy low-dose CT measurements. To tackle this problem, we propose a hybrid conditional normalizing flow, which integrates the physical model by using the filtered back-projection as conditioner. We evaluate our approach on a low-dose CT benchmark and demonstrate superior performance in terms of structural similarity of our flow-based method compared to other deep learning based approaches.

Conditional Normalizing Flows for Low-Dose Computed Tomography Image Reconstruction

5 years ago
v1
5 authors

Categories

eess.IV

Abstract

Image reconstruction from computed tomography (CT) measurement is a challenging statistical inverse problem since a high-dimensional conditional distribution needs to be estimated. Based on training data obtained from high-quality reconstructions, we aim to learn a conditional density of images from noisy low-dose CT measurements. To tackle this problem, we propose a hybrid conditional normalizing flow, which integrates the physical model by using the filtered back-projection as conditioner. We evaluate our approach on a low-dose CT benchmark and demonstrate superior performance in terms of structural similarity of our flow-based method compared to other deep learning based approaches.

Authors

Alexander Denker, Maximilian Schmidt, Johannes Leuschner et al. (+2 more)

arXiv ID: 2006.06270
Published Jun 11, 2020

Click to preview the PDF directly in your browser