Visual perception studies and observer models in medical imaging.

摘要: Most academic radiologists will be familiar with receiver operating characteristic (ROC) studies. Fundamental studies of human observer performance are now usually performed by forced-choice methods. Both methods based on signal detection theory. The ROC method gives an curve true-positive versus false-positive probabilities. area under the curve, A Z , can used a summary measure. In method, observers given 2 or more images one containing signal. observer's task is to select option most likely contain percentage correct responses, PC, Precise comparison limited very controlled experiments in which signals (simulated lesions for example) carefully designed and discrimination true random noise. Under these conditions, theory predicts simple relationship between measures results consistent There description experimental data analysis. has also been considerable work development theoretic models. Human experiment have evaluate Models that correlate well turn preliminary design new imaging systems selection image quality metrics comparing equipment performance, this article provide during last 30 years evaluating capabilities, models metrics.