Information Theoretic Regularisation in Multimodal Medical Imaging

Simon Arridge
University College London
Computer Science

Multimodal imaging is of increasing interest. Several data sets with
complementary information are measured either simultaneously or sequentially
and multiple image reconstruction problems need to be solved. In the
sequential case image registration may also be required. In some cases
one image reconstruction problem may be well posed (e.g. well-sampled CT or
MRI) whereas the other is illposed (e.g Optical Tomography). In this talk I will discuss some methods for using a robust reconstruction as a prior for regularisation of the complementary ill-posed problem. I will discuss the use of
structured priors and a new method using information theoretic metrics as a regularisation functional.

Back to Mathematical Problems, Models and Methods in Biomedical Imaging