Paper
12 March 2009 Validation of an improved abnormality insertion method for medical image perception investigations
Mark T. Madsen, Gregory R. Durst, Robert T. Caldwell, Kevin M. Schartz, Brad H. Thompson, Kevin S. Berbaum
Author Affiliations +
Abstract
The ability to insert abnormalities in clinical tomographic images makes image perception studies with medical images practical. We describe a new insertion technique and its experimental validation that uses complementary image masks to select an abnormality from a library and place it at a desired location. The method was validated using a 4-alternative forced-choice experiment. For each case, four quadrants were simultaneously displayed consisting of 5 consecutive frames of a chest CT with a pulmonary nodule. One quadrant was unaltered, while the other 3 had the nodule from the unaltered quadrant artificially inserted. 26 different sets were generated and repeated with order scrambling for a total of 52 cases. The cases were viewed by radiology staff and residents who ranked each quadrant by realistic appearance. On average, the observers were able to correctly identify the unaltered quadrant in 42% of cases, and identify the unaltered quadrant both times it appeared in 25% of cases. Consensus, defined by a majority of readers, correctly identified the unaltered quadrant in only 29% of 52 cases. For repeats, the consensus observer successfully identified the unaltered quadrant only once. We conclude that the insertion method can be used to reliably place abnormalities in perception experiments.
© (2009) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mark T. Madsen, Gregory R. Durst, Robert T. Caldwell, Kevin M. Schartz, Brad H. Thompson, and Kevin S. Berbaum "Validation of an improved abnormality insertion method for medical image perception investigations", Proc. SPIE 7263, Medical Imaging 2009: Image Perception, Observer Performance, and Technology Assessment, 726305 (12 March 2009); https://doi.org/10.1117/12.813772
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Medical imaging

Tomography

Lung

Radiology

Visualization

Chest

Computed tomography

Back to Top