Geoffrey D. Rubin MD, Presenter: Medical Advisory Board, Fovia, Inc Research grant, General Electric Company Consultant, Medtronic, Inc Consultant, TriVascular, Inc Consultant, TeraRecon, Inc

Martin Tall, Abstract Co-Author: Nothing to Disclose

Kingshuk Roychoudhury, Abstract Co-Author: Nothing to Disclose

Justus E. Roos MD, Abstract Co-Author: Nothing to Disclose

David Seungwon Paik PhD, Abstract Co-Author: Nothing to Disclose

Sandy Napel PhD, Abstract Co-Author: Medical Advisory Board, Fovia, Inc Medical Advisory Board, Vital Images, Inc Consultant, Carestream Health, Inc

Donald Le Ly BSC, Abstract Co-Author: Nothing to Disclose

To determine the factors that affect observer attention when searching for lung nodules in volumetric CT data.

The gaze cone is a perceptual construct centered on foveal gaze within which an object may attract and cause refocusing of attention. A 3x3x5 cm lung mass was simulated and imbedded into 5-cm thick subvolumes (SV) extracted from 3 unenhanced lung MDCT scans (64-row, 1.25 mm thickness, 0.7 mm increment). The mass was visible on all transverse sections in the SV. Two collections of 30 solid secondary (2°) nodules were simulated with 4-6 mm and 7-8 mm diameters, respectively. A total of 207 SVs were created containing one 2° nodule imbedded at a random depth at a distance from the edge of the mass of 2.5, 5, 7.5, or 10 mm, and along rays cast every 45° from the center of the mass. For each SV, a movie was created of the sections from superior to inferior at 3 sections/s. Six radiologists observed each movie once. Gaze tracking hardware assured that vision was centered on the 3 cm mass throughout playback. Upon completion of each viewing the radiologist assigned a confidence rating (0-5) to the detection of a 2° nodule and indicated its location on the screen. Detection sensitivity was analyzed relative to 2° nodule size, radial position, distance from the mass, and local lung complexity (LLC), defined on each transverse section within a 3 cm cylindrical ROI centered on the 2° nodule based upon the number and combined area of discrete lung opacities (e.g., vessel cross-sections) present above a -500 HU threshold.

Using a proportional odds logistic regression model and eliminating redundant predictors, models fit individually to each reader resulted in the following decreasing order of association based on greatest reduction in Akaike Information Criterion (rAIC): 2° nodule size (6/6 readers, max p<0.001, avg. rAIC 38.56), distance from central mass (6/6 readers, max p<0.001, avg. rAIC 16.03), LLC:number (5/6 readers, max p=0.05, avg. rAIC 7.93), and LLC:area (3/6 readers, max p=0.03, avg. rAIC 2.54).

When searching for lung nodules in CT scans, the size of the gaze cone varies with nodule size and lung background complexity

Mapping the gaze cone is an initial step in understanding the variations with which readers search volumetric imaging data and the impact of search on lesion detection.

Rubin, G, Tall, M, Roychoudhury, K, Roos, J, Paik, D, Napel, S, Ly, D, Perception of Lesions in the Gaze Cone Periphery: Impact of Lesion Size, Distance, and Local Lung Complexity on Detection.  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL. http://archive.rsna.org/2010/9013537.html