Cross-Modality Priors in Tomography
Multimodality techniques are increasingly used in tomography. They include methods where a subject is imaged with two or more modalities either sequentially or simultaneously, as well as coupled techniques where user-controlled probes generate stimulated signals read by a complementary measurement system. Examples of the former include Positron Emission Tomography/Magnetic Resonance Imaging (PET-MRI), or (Near Infra Red/Magnetic Encephalography (NIR-MEG); examples of the latter include PhotoAcoustic Tomography (PAT) and Magnetic Resonance Elastography (MRE). Whenever inverse problems are ill-posed, regularisation is required to convey stability to the reconstruction. In the multimodality techniques mentioned above, regularisation may be imposed by proposing that one of the modalities is worse-posed and should be guided by the other which is well, or at least better, posed. Such methods include local structural priors, non-local priors, and joint information theoretic priors. A more recent suggestion is to jointly solve the two inverse problems using a prior that reflects both within modality and cross-modality information.
In this minisymposium we will present recent progress in these techniques.