Inverse problems are typically concerned with the interpretation of indirectly measured data that are related to the quantities of interest (images) through models that describe the data acquisition. In practice, one has to deal not only with noisy or incomplete data, but also with simplified or imperfect models that cannot capture the mechanisms of data acquisition in full complexity. Correctly accounting for these imperfections is crucial for the development of stable numerical algorithms. The aim of this workshop is bringing together researchers working on different approaches to such problems arising in imaging in order to highlight similarities as well as differences and foster further collaboration.

A lattice analogue of the residual method for inverse problems with imperfect forward models

Yury Korolev (University of Cambridge)

All-at-once formulation and regularization of inverse problems

Barbara Kaltenbacher (Alpen-Adria-Universität Klagenfurt)

Deep learning for trivial inverse problems

Peter Maass (University of Bremen)

Spatio-temporal imaging by joint motion and image reconstruction and its application to spat-temporal MRI

Angelica I. Aviles-Rivero (University of Cambridge)

Bayesian imaging inverse problems with partially unknown models

Marcelo Pereyra (Harriott-Watt University)

Improved source estimation in EEG with Bayesian modelling of the unknown skull conductivity

Alexandra Koulouri (Aristotle University of Thessaloniki)

A time-regularized blind deconvolution method via non-convex optimisation

Audrey Repetti (Heriot-Watt University, Edinburgh)

Accounting for model-errors in PDE-constrained optimization

Tristan van Leeuwen (Utrecht University)

Organizers:

Martin Burger (University of Muenster)

Yury Korolev (University of Cambridge)

Keywords:

image reconstruction, inverse problems, partial differential equation models