3D ultrasonography is as accurate as low-dose CT in thyroid volumetry

The purpose of this study was to compare thyroid volumetry by three-dimensional mechanically swept ultrasonography (3DmsUS) and low-dose computed tomography (ldCT).

30 subjects referred for radioiodine therapy of benign thyroid diseases were subjected to 3DmsUS and ldCT. A prerequisite of 3DmsUS analyses was that the scans had to capture the entire thyroid, excluding therefore cases with a very large volume or retrosternal portions. The 3DmsUS data were transformed into a DICOM format, and volumetry calculations were performed via a multimodal workstation equipped with standard software for cross-sectional imaging. Volume was calculated applying both the ellipsoid model and a manually tracing method. Statistical analyses included 95% confidence intervals (CI) of the means and limits of agreement according to Bland and Altman, the latter including 95% of all expected values.

Volumetric measurements by 3DmsUS and ldCT resulted in very high, significant correlation coefficients, r = 0.997 using the ellipsoid model and r = 0.993 with the manually tracing method. The mean relative differences of the two imaging modalities proved very small (-1.2±4.0% [95% CI -2.62; 0.28] using the ellipsoid model; -1.1±5.2% [95% CI -2.93; 0.80] using the manually tracing method) and the limits of agreement sufficiently narrow (-9.1% to 6.8%; -11.3% to 9.2%, respectively).

For moderately enlarged thyroids, volumetry with 3DmsUS proved comparable to that of ldCT, irrespective of whether the ellipsoid model or the manually tracing method was applied. Thus, 3DmsUS qualifies as a potential alternative to ldCT, provided that the organ is completely accessible. The use of a standard workstation for cross-sectional imaging with routine software did not prove problematic.