Tomographic ultrasound (tUS) for thyroid volume determination

Accurate thyroid volume determination is essential for treatment planning. This feasibility study evaluated the accuracy of a novel tomographic ultrasound (tUS) approach compared to 2D ultrasound, using contrast-enhanced computed tomography (ceCT) as reference standard.

In this prospective, single-center study, 50 patients undergoing neck ceCT were enrolled. Thyroid volume was measured by 2D ultrasound (ellipsoid model), automatedvolume (tUS uncorrected), physician-corrected volume (tUS corrected) and ceCT. Linear regression for proportional bias, log-transformed Bland-Altman analysis, clinical error metrics and correction duration were evaluated.

2D ultrasound showed significant proportional bias (slope=-0.215, p<0.0001) and systematic underestimation (mean difference -6.0mL; geometric mean ratio 0.76). Corrected tUS eliminated proportional bias (p=0.0557) and reduced bias (mean difference +0.8mL; geometric mean ratio 1.04). Corrected tUS achieved lower MAE (4.42mL vs. 6.53mL) and MAPE (19.4% vs. 25.4%) than 2D ultrasound, placing 60.0% vs. 34.0% of cases within ±20% error limits. In enlarged glands (n=19), corrected tUS maintained superior accuracy (MAE 5.82mL, MAPE 12.4%) versus 2D ultrasound (MAE 11.35mL, MAPE 26.5%). Physician correction averaged 2:51minutes, independent of thyroid volume (r=-0.081, p=0.5784).

In this feasibility study, physician-corrected tUS reduced the size-dependent systematic underestimation observed with 2D ultrasound and yielded volumes closer to ceCT, particularly in enlarged glands, within a clinically feasible workflow. Because the best-performing method required manual expert correction and the limits of agreement with ceCT exceeded the predefined clinical threshold, corrected tUS should be regarded as a promising semi-automated approach that requires further validation, including interobserver assessment, before it can be considered interchangeable with ceCT.