RESUMO
PURPOSE: To evaluate diagnostic performance of PACS-based quantitative gray-scale ultrasound as an objective method in evaluation of pediatric thyroiditis. METHODS: Quantitative measurements of the echo-intensity level of the thyroid were obtained from ultrasound images, retrospectively using a PACS-based tool in 37 children with the tissue-proven diagnosis. Thyroid/muscle ratio was calculated by dividing the mean echo intensity of thyroid by that of adjacent strap muscle. Heterogeneity index (HI) was calculated by dividing thyroid standard deviation (SD) by thyroid mean values. For qualitative evaluation, two radiologists independently reviewed ultrasounds twice for the presence of thyroiditis. A consensus session was performed for patients for whom there was disagreement. Intra- and inter-observer reliability were assessed. Thyroid/muscle ratio and HI were correlated with final pathology. RESULTS: Lymphocytic thyroiditis was found by histopathology in 19/37 (51%). No significant difference between thyroiditis and normal thyroid groups was found for either thyroid/muscle ratio (1.51 and 1.62, respectively, p = .82) or HI (0.23 and.23, respectively, p = .37). A larger proportion of patients for whom the consensus review indicated thyroiditis were confirmed by histopathology than would be expected by chance alone (12/19 (63%), p = .03). There was fair inter-observer agreement (κ with 95% confidence intervals of 0.36 (0.14-0.57), p = .004) and slight intra-observer agreement for each radiologist (κ with 95% confidence intervals of 0.13 (0.17-0.43), p = .39 and 0.17 (0.15-0.49), p = .31). CONCLUSION: Quantitative gray-scale echo intensity analysis of US was not sufficient to diagnose thyroiditis in a pediatric population. Consensus qualitative analysis of ultrasound was more consistent with pathological diagnosis.
