This repository contains code and supplementary material for the article "Quantitative comparison of deep learning-based image reconstruction methods for low-dose and sparse-angle CT applications".

Information on the environment setup can be found in the folder supp_material/environment_details.

Trained network parameters for the experiments on the Apple CT datasets are available from the supplementing zenodo record. Parameters for the LoDoPaB-CT Dataset of those reconstructors implemented in DIVαℓ can be found in the supplementary repository supp.dival.

Saved reconstructions are provided in two separate records:

• LoDoPaB-CT Challenge Reconstructions
• Apple CT Test Reconstructions.