Human–machine interfaces for medical imaging and clinical interventions

Abstract Until the day when medical diagnostics and clinical interventions are fully automated, all Medical Image Computing and Computer-Assisted Interventions will have, at their core, the system architecture of a “Human–Machine Interface” (HMI). An “interface” in abstract terms is a boundary between two systems (for an HMI, the two systems are the human and the machine; while in an HCI, the two systems are the human and the computer – for this chapter, HCI/HMI will be used interchangeably). More critically, the study of any interface involves characterizing “what” flows between the two. For a Human–Machine Interface, the flows will be characterized in terms central to information, control, and decision theories. In these terms, we can characterize these processes within the context of the information-processing requirements of the task being performed: namely, the observability and controllability of the state of the task being conducted using the interface. Since the applications for Medical and Interventional interfaces span a vast domain, we will begin by highlighting a few representative examples, before moving forward to identify a number of unifying characteristics. We will then identify the key “Design Principles” that emerge when considering such interfaces, before concluding with a unifying framework for the objective evaluation of such interfaces, along with future directions for research in this area.