Translator Disclaimer
23 March 2010 Search field size and lesion detection performance
Author Affiliations +
In this study, we evaluated the ability of an observer to identify abnormal foci on CT and how that ability is affected by changing the search field size from a whole abdomen to the liver region alone. A 2-Alternate Forced Choice (2 AFC) experimental paradigm was used to quantify observer detection performance. Each AFC experiment yielded the intensity needed to achieve 92% accuracy in lesion detection (I92%). Abdominal images were obtained at an x-ray tube voltage of 120 kV with a CTDIvol of 20 mGy. Circular lesions were generated by projecting spheres onto the image plane, followed by blurring function. Five lesion sizes (5 mm, 7 mm, 10 mm, 12 mm, and 15 mm), and four readers who were extensively trained in AFC methodology, were used in the 2AFC experiments. Each experiment was repeated 4 times to improve the experimental precision, as well as to provide an estimate of experimental uncertainty. For each observer, the experimental order of the 40 experiments was randomized to eliminate learning curve and/or observer fatigue. We measured contrast detail slopes for both Abdomen and Liver search field size, and determined ratio of I92% value for Abdomen search field to the corresponding I92% for the Liver search field (i.e., Rabd:liv). Values of Rabd:liv provides quantitative indicator of the relative difficulty of detection lesions in the whole Abdomen relative to lesion detection restricted to the Liver. The slope of the contrast detail curve for the Abdomen search field was -0.03, whereas the corresponding slope for the Liver search field was -0.18. Rabd:liv ranged between 1.3 and 1.6, with an average of 1.4 ± 0.1. The value of Rabd:liv monotonically increased from 1.35 for 5 mm lesions to nearly 1.6 for the largest 15 mm lesion. The results of our study indicate that limiting the area of search to the liver on a CT of the abdomen improves the detection of mass lesions. This finding is almost certainly related to the fact that the liver provides a relatively homogenous background for identifying abnormalities, while the rest of the abdomen is much more heterogeneous. The clinical relevance of our findings is that CT can have limitations for detection of mass lesions outside of the liver, and sizes of masses necessary for detection are larger outside the liver than within.
© (2010) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Sameer Tipnis, Walter Huda, Andrew Hardie, and Kent Ogden "Search field size and lesion detection performance", Proc. SPIE 7622, Medical Imaging 2010: Physics of Medical Imaging, 76224U (23 March 2010);


Back to Top