Objective: The purpose of this study was to evaluate quantitatively observer performance with ultrasonography and sialography in the diagnosis of parotid gland involvement of Sjögren's syndrome.
Study design: Sixty-four parotid gland sialograms and 65 ultrasonograms were prepared for observer performance experiments. They included both modalities in 24 Sjögren's syndrome and 19 nonspecific parotitis cases, 21 normal parotid gland sialograms, and 22 normal ultrasonograms in healthy volunteers. The images were randomly sequenced and presented to five observers who were asked to describe several findings and finally to determine the imaging diagnosis by ranking the abnormal features and the diagnosis on a five-point-rating scale. Observer performance was evaluated on the basis of the reliability of findings interpreted and the diagnostic accuracy of each modality from observers' rating scores.
Results: The diagnostic accuracy of sialography was very high, nearly perfect. The diagnostic accuracy of ultrasonography was lower than that of sialography, resulting from the lower incidence of characteristic findings in the disease groups and lower sensitivity on ultrasonography. In the differentiation of Sjögren's syndrome from the normal, however, the diagnostic accuracy of ultrasonography increased to 80% for all cases, and up to nearly 90% in the advanced sialographic stages.
Conclusion: Ultrasonography is useful for the diagnosis of Sjögren's syndrome in the advanced stages. Taking the noninvasiveness of this technique into account we recommend first applying ultrasonography to the diagnosis of Sjögren's syndrome and performing sialography when no positive findings are detected on ultrasonography.