Assessing system performance.

Measuring system performance during image-guided procedures is difficult but necessary if the safety, quality, and efficiency of interventional radiology are to be improved. Because overall performance during these procedures is directly related to the operator's skill, this article will focus on how observable manifestations of skill can be elicited and analyzed. The discussion includes issues encountered when the results of "high-stakes" tests such as board certification are used to predict future system performance. The underlying theme is that process improvement uses the scientific method to gain knowledge of a system. The successive cycles of "predict-test-analyze-decide" found in any training program generate an expanding knowledge base that leads to measurable improvements in system performance.