Matthias Ehrhardt: Combined Image Reconstruction for Combined Medical Imaging

Abstract:

Recent years have seen a trend that different medical imaging modalities get combined into the same system and can probe the patient either one at a time or even simultaneously. An example for simultaneous multi-modality systems are combined positron emission tomography (PET) and magnetic resonance imaging (MRI) scanners which allow simultaneous imaging of structure and function of the human body. The PET data suffers from low resolution and high noise while under-sampling in MRI does not guarantee a unique solution. As function follows structure, the solutions of these inverse problems are a priori expected to show similar shapes. We will discuss models that can exploit the expected similarity in structure and either use one modality to guide the other or to jointly reconstruct both modalities. Our numerical results indicate that combining the imaging modalities can lead to sharper, better defined and less biased PET images and MRI images of good image quality from a lot less data.