Medicine has been a prominent driver for visualization research over many years with application areas ranging from diagnostics to intra-interventional guidance. Considerable success has been achieved and many techniques have been adopted in clinical practice. However, the transition from research prototypes to clinical applications also gives rise to new challenges. Clinical application scenarios are high-throughput and there is little time to fiddle with visualization parameters and to tune individual settings. Hence, techniques and interfaces need to be "smart" in the sense that they incorporate domain knowledge to automate such tasks in a goal-oriented manner. In this talk, I will discuss visualization and interaction concepts developed to achieve these goals and present examples for their realization.