The search model is a statistical parameterization of a perceptual model of radiological interpretation (1-3) a key aspect of which is the radiologist does not assign equal attention units to all locations in the image. Rather, image viewing begins with a brief holistic interpretation requiring a few hundred milliseconds during which information is collected predominantly by peripheral vision and regions that appear suspicious for cancer or flagged for further examination. The radiologist then examines these regions individually using higher resolution foveal vision and makes decisions whether or not to report them. In other words the viewing involves two stages – holistic followed by region-by-region cognitive analysis and decision making.
Evidence for the Kundel-Nodine model is to be found in eye-position recordings made on radiologists. By monitoring corneal reflections from an infrared light-source one can measure the line-of-gaze of an observer (4) as a function of time. Shown below are eye-position recordings for a mammogram for two observers: (a) an inexperienced observer and (b) an expert radiologist. Brief fixations, defined as locations where the observer’s gaze duration exceeded 100 ms, are indicated by the small circles. Clustered fixations with a total dwell-time exceeding one second are indicated by the large high-contrast circles. These are the regions flagged by the holistic mechanism. It is believed (5) that the observer makes conscious decisions to report or not to report only at the flagged locations. The large low-contrast circle indicates a cancer. Notice that the inexperienced observer flagged 4 normal regions as possible lesion candidates and missed the lesion whereas the radiologist found the lesion without flagging any normal regions. These two aspects of expertise, finding lesions while not flagging normal regions, constitute two parameters of the search model. The third parameter is the lesion signal-to-noise ratio.
Images are courtesy of Dr. Claudia-Mello Thoms and Prof. Harold Kundel.
1. Kundel HL, Nodine CF. A visual concept shapes image perception. Radiology 1983; 146:363-368.
2. Kundel HL, Nodine CF. Modeling visual search during mammogram viewing. Proc. SPIE 2004; 5372:110-115.
3. Nodine CF, Kundel HL. Using eye movements to study visual search and to improve tumor detection. RadioGraphics 1987; 7:1241-1250.
4. Duchowski AT. Eye Tracking Methodology: Theory and Practice. Clemson, SC: Clemson University, 2002.
5. Hillstrom A. Repetition effects in visual search. Percept Psychophys. 2000; 2:800-817.