Purchase this article with an account.
Melissa Trevino, Todd Horowitz, Ismail Turkbey, Peter Choyke, Marcin Czarniecki; Detecting and localizing prostate lesions within half a second. Journal of Vision 2018;18(10):654. doi: 10.1167/18.10.654.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Previous research has shown radiologists can discriminate "abnormal" from "normal" medical images at above-chance levels in a fraction of a second (Carmody, Nodine, & Kundel, 1981; Evans et al., 2013; Evans et al., 2016; Kundel & Nodine, 1975), without necessarily being able to localize the abnormality. This is referred to as gist perception. However, we do not know whether this phenomenon is a general property of expert medical perception or a specific feature of conventional chest radiographs and mammograms. Here we investigated whether radiologists can extract gist from multiparametric magnetic resonance imaging (mpMRI) of the prostate. Stimuli were 100 T2-weighted MRI images of the base, mid and apex regions of the prostate. Lesions (Gleason scores 6-9) were present in 50% of the images. Two groups of participants were tested, radiologists specializing in mpMRI (n = 11) and radiologists not trained in prostate imaging (n = 5). Images were presented for 500 ms, followed by a sector map of the prostate. Participants localized the lesion on the sector map, and then provided a confidence rating. While radiologists specializing in mpMRI performed well above chance at detecting the lesion (d'= 1.1, sd = 0.3), radiologists without prostate training performed at chance level (d'= -0.1, sd = 0.3). Curiously, both groups could localize the lesions above chance. These data extend the phenomenon of gist perception to the mp-MRI domain. Like mammographers, radiologists with prostate training can detect lesions in a glance. Unlike the mammography data, however, localization performance was well above chance, and predicted by detection performance (r = .80), suggesting that there may be enough information available in 500 ms to guide attention to the lesions.
Meeting abstract presented at VSS 2018
This PDF is available to Subscribers Only