A comparison of two- and three-dimensional ultrasonographic methods for evaluation of ovarian follicle counts and classification of polycystic ovarian morphology.

Objectives: To determine the level of agreement across assessments of follicle number per ovary (FNPO) and classifying of polycystic ovarian morphology (PCOM; FNPO ≥25) with the use of various real-time (RT) and off-line two-dimensional (2D) and three-dimensional (3D) ultrasonographic methods.

Design: Method comparison study.

Setting: University-based clinical research unit.

Patient(s): Sixteen women with and without PCOM.

Intervention: Thirty-two ovaries were analyzed with the use of eight ultrasonographic methods: 2D-Grid (reference method), 2D-RT, 2D-RT with Grid, multiplanar view (MPV), MPV-RT, tomographic ultrasound imaging (TUI), TUI-RT, and semiautomated volume calculation (SonoAVC).

Main Outcome Measure(s): FNPO, PCOM status, and time to obtain FNPO. Clinical feasibility, defined as the time taken to obtain FNPO, also was evaluated.

Result(s): 2D-RT overestimated FNPO versus 2D-Grid (3 ± 9 follicles) owing to overcounting in non-PCOM ovaries (6 ± 6 follicles). However, systematic bias was not detected when a grid overlay was incorporated (2D-RT with Grid). SonoAVC underestimated FNPO (-3 ± 5 follicles), particularly in PCOM ovaries (-4.1 ± 5.0 follicles). No bias in FNPO was detected between MPV, TUI, or TUI-RT versus 2D-Grid. 2D-RT significantly misclassified ovaries as PCOM. All methods except MPV took less time to complete FNPO assessments compared with 2D-Grid.

Conclusion(s): Variability in FNPO across ultrasonographic methods limits their interchangeable use, particularly when a precise metric is needed. 2D-RT may be problematic owing to its propensity to misclassify PCOM. 2D-RT with Grid and MPV-RT could represent clinically feasible alternatives to obtain FNPO and classify PCOM. Efforts to reduce variation in FNPO will clarify the relevance of PCOM in women's health.