Premium
Evaluation of observer performance in detecting blood vessels on simulated angiographic images
Author(s) -
Sandor Tamas,
Swensson Richard G.
Publication year - 1978
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.594510
Subject(s) - observer (physics) , receiver operating characteristic , parametric statistics , nonparametric statistics , artificial intelligence , computer science , computer vision , set (abstract data type) , medical imaging , blood vessel , pattern recognition (psychology) , mathematics , statistics , medicine , physics , quantum mechanics , psychiatry , programming language
This paper presents methods for evaluating the effects of technical factors on observers’ detection of small blood vessels in angiographic images. A fixed set of physical parameters was used to generate computer‐simulated angiographs that contained blood vessels of various diameters. Observers’ judgments about each set of images permitted estimation of the receiver operating characateristic (ROC) curves for vessel detection and of the probability of correctly locating the vessels. Both parametric and nonparametric methods were used to characerize the ROC curves. Three different, but theoretically related, measures of vessel detectability varied systematically with changes in the blood vessel's diameter, reflecting changes in observer performance.