Current status of methodology for assessment of image quality by human observer studies

Receiver operating characteristic (ROC) analysis is now widely accepted as the best way of assessing image quality in terms of the accuracy with which human observers can perform image-based classification tasks. After a brief look backward, this talk will address the questions of "Where are we?" and "Where are we going?" in the development and use of ROC methodology. Particular emphasis will be placed upon identifying the still-evolving conceptual threads of ROC analysis, clarifying current methodological limitations and controversies, and highlighting areas in which future effort might fruitfully be focused.