Human vs. model observers in anatomic backgrounds

Model observers have been compared to human performance detecting low contrast signals in a variety of computer generated backgrounds including white noise, correlated noise, lumpy backgrounds, and two component noise. The purpose of the present paper is to extend this mark by comparing a number of previously proposed model observers (non-prewhitening matched filter, non-prewhitening matched fitler model with an eye filter Hotelling observer and channelized-Gabor Hotelling observer model) to human visual detection performance in real anatomic backgrounds (x-ray coronary angiograms). Human and model observer performance are compared as a function of increasing added white noise. Our results show that three of the four models (the non-prewhitening matched filter, the Hotelling and channelized-Gabor Hotelling) are good predictors of human performance.