The search termination criterion mediating under exploration of 3D volumetric images

3D volumetric images are prevalent in industries ranging from radiology and oncology to airport luggage screenings. With 3D medical images, however, radiologists are visually overburdened by the vast amount of image data requiring inspection. Radiologists and trained (non-radiologist) observers often under-explore 3D images—producing misses of small targets that are undetectable in the visual periphery. We investigate why observers under-explore 3D images and what quitting criterion they use when reporting, “target absent.” Six trained observers participated in two eye-tracking experiments to analyze three possible explanations for the under-exploration in 3D search: (1) a time or (2) a count of eye movements, or (3) an area explored by the Useful Field of View (UFOV) search-termination criterion. Experiment 1 derived measured and estimated UFOVs per observer. We tested the detectability of a small and large target embedded in 1/f noise as a function of retinal eccentricity. Observers also estimated how well they could see the two targets in their visual periphery. Experiment 2 had participants engage in a 2D and a 3D search for the two targets. Neither time nor a fixed number of eye movements criterion could explain the search termination in reported target-absent trials in 2D or 3D. However, the area explored with estimated but not measured UFOVs was consistent across targets (2D and 3D). Moreover, while performing the 3D search, people covered the 2D image plane with their estimated and measured UFOVs to the same extent as in the 2D search task, leaving much of the 3D volume unexplored.