Research opportunities in creating medical images

Computer vision is traditionally defined as the analysis of the content of images. In the context of medical imaging, this naturally leads to a focus on segmentation and registration. I argue that there is an underappreciated opportunity for computer vision researchers in the domain of image creation. I illustrate the technical challenges involved, with an emphasis on magnetic resonance (MR) imaging. These problems have unusual features that make them intellectually interesting, beyond their obvious practical importance.