Purpose: We determined the diagnostic performance of assessing the nodularity of deep versus superficial surfaces of the liver as a predictor of cirrhosis.
Methods: One sonologist retrospectively reviewed the sonograms of 100 patients at risk for cirrhosis based on clinical data and laboratory tests. A second sonologist reviewed the sonograms of a subset of 25 patients to assess for inter-reader variability, while the first sonologist re-reviewed the sonograms of a different subset of 25 patients for intrareader variability. Sonograms of all patients were obtained with standard sector- or curved-array transducers. Biopsy confirmation of cirrhosis was used as the standard for diagnostic accuracy.
Results: Fifty of the 100 patients had a pathologic diagnosis of cirrhosis. The sensitivity of inspection of the deep surface versus the superficial surface was 86% versus 53% (p = 0.0003), respectively. Sensitivity was not dependent on pathologic type. Intrareader agreement was better for deep than for superficial surface observations. The overall inter-reader agreement was fair and comparable for both deep and superficial surface methods.
Conclusions: The sensitivity for detecting cirrhosis based on surface nodularity is greater for deep than for superficial surface observations. As the prevalence of disease increases, the accuracy of assessment of the deep surface in predicting cirrhosis increases.
Copyright 2002 Wiley Periodicals, Inc.