Intraobserver and interobserver reliability of automated antral follicle counts made using three-dimensional...

To assess the reliability of automated measurements of the total antral follicle count (AFC) made using Sono-Automatic Volume Count (SonoAVC), and to compare these to two-dimensional (2D) and manual three-dimensional (3D) techniques.

Fifty-five subjects aged under 40 years who had 3D transvaginal ultrasound examination in the early follicular phase of their menstrual cycle were prospectively recruited. 3D datasets were acquired and subsequently analyzed. The total AFC (2-10 mm antral follicles) was calculated by two observers using three independent methods: 2D real-time equivalent (2D-RTE), 3D manual multiplanar view (3D-MPV), and SonoAVC. For measurements made using SonoAVC, the initial automated count (sAVC-AA) was recorded and postprocessing (sAVC-PP) then applied to identify follicles that had been missed or incorrectly included. Intraclass correlation and limits of agreement were used to evaluate the methods.

The intra- and interobserver reliability of measurements of total AFC was best with SonoAVC with postprocessing followed by 3D-MPV and 2D-RTE. The initial count calculated by sAVC-AA missed follicles and this was reflected in the significantly lower mean total AFC (6.51 +/- 4.79) than that made after postprocessing techniques (sAVC-PP, 18.42 +/- 10.53, P < 0.001; 3D-MPV, 19.38 +/- 10.85, P < 0.001; and 2D-RTE, 19.26 +/- 10.55, P < 0.001). The mean total AFC became more comparable with postprocessing (sAVC-PP) but still remained significantly lower than counts made with 2D-RTE and 3D-MPV (P < 0.05).

SonoAVC with postprocessing is a reliable method for measuring total AFC. It takes longer to perform, because of the need for postprocessing, and obtains values that are lower than those obtained by the 2D and 3D-MPV techniques. However, the AFC obtained by sAVC-PP is likely to be lower because this method measures and color codes each follicle preventing recounting.