Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

PURPOSE: To develop an eye-tracking method applicable to three-dimensional (3D) images, where the abnormality is both moving and changing in size. MATERIALS AND METHODS: Research ethics committee approval was granted to record eye-tracking data from six inexperienced readers who inspected eight short (<30 seconds) endoluminal fly-through videos extracted from computed tomographic (CT) colonography examinations. Cases included true-positive and false-positive polyp detections from a previous study (polyp diameters, 5-25 mm). Eye tracking was performed with a desk-mounted tracker, and readers indicated when they saw a polyp with a mouse click. The polyp location on each video frame was quantified subsequently by using a circular mask. Gaze data related to each video frame were calculated relative to the visible polyp boundary and used to identify eye movements that pursue a polyp target as it changes size and position during fly-through. Gaze data were then related to positive polyp detections by readers. RESULTS: Tracking eye gaze on moving 3D images was technically feasible. Gaze was successfully classified by using pursuit analysis, and pursuit-based gaze metrics were able to help discriminate different reader search behaviors and methods of allocating visual attention during polyp identification. Of a total of 16 perceptual errors, 15 were recognition errors. There was only one visual search error. The largest polyp (25 mm) was seen but not recognized by five of six readers. CONCLUSION: Tracking a reader's gaze during endoluminal interpretation of 3D data sets is technically feasible and can be described with pursuit-based metrics. Perceptual errors can be classified into visual search errors and recognition errors. Recognition errors are more frequent in inexperienced readers.

Original publication




Journal article



Publication Date





924 - 931


Clinical Competence, Colonic Polyps, Colonography, Computed Tomographic, Diagnostic Errors, Eye Movements, Humans, Imaging, Three-Dimensional, Radiographic Image Interpretation, Computer-Assisted