Intra- and interoperator reliability of manual and semi-automated measurement of fetal nuchal translucency by sonographers with different levels of experience.

OBJECTIVE: To compare the intra- and interoperator reliability of manual and semi-automated nuchal translucency thickness (NT) measurement by sonographers with different levels of experience.

METHODS: From our digital database we selected retrospectively 70 NT images without measurements and of highest quality according to the guidelines of The Fetal Medicine Foundation (FMF). Four operators (two experts who were consultants in fetal medicine, one general consultant in obstetrics and gynecology with average knowledge in obstetric ultrasound and one final-year medical student with no experience in ultrasound), having first attended the FMF's online NT measurement course, measured the fetal NT manually according to FMF guidelines and semi-automatically from the inner to inner borders of the two echogenic lines delineating the NT and from the inner border to the middle of the lower NT line. The semi-automatic measurement algorithm identifies the two NT lines within a manually chosen region of interest and standardizes the caliper placement process. Each operator was blinded to any pre-existing measurements. Each measurement was performed twice by each operator. Intraoperator repeatability was assessed by the within-operator SD and intraclass correlation coefficients (ICC). Interoperator reliability was assessed by comparison with the mean manual measurement of the experts, which was considered as the gold standard.

RESULTS: The mean NT measurement of the four sonographers was between 1.9 and 2.0 mm with both the manual and the semi-automated inner-inner NT measurement and it was between 2.2 and 2.3 mm with the semi-automated inner-middle method. The within-operator SDs of the manual measurement for both experts were similar and significantly lower than those of the non-experts, and all four sonographers performed similarly with the semi-automated measurement. Concerning the interoperator biases relative to the manual measurements of the experts taken as the gold standard, the manual measurements of the student and the general consultant were significantly lower, by 0.18 mm and 0.12 mm. The semi-automated inner-inner measurements of all four sonographers were much closer to the gold standard, although there was still a significant underestimation (mean differences from gold standard between 0.04 and 0.06 mm). With the semi-automated inner-middle measurement there was a significant and more pronounced overestimation of fetal NT compared with the gold standard, relative biases ranging from 0.08 to 0.23 mm above it.

CONCLUSION: Semi-automation of NT measurement leads to improved standardization of the NT assessment process. In particular, non-expert operators improve as the variability in their caliper placement can be reduced to the level of experts. However, the most important contributor to the inter- and intraoperator variability of NT assessment is that of image acquisition, which is not affected by semi-automation of NT measurement.