Counting follicles to define polycystic ovary morphology: should we change the way we do it?

Counting follicles to define polycystic ovary morphology: should we change the way we do it?

VOLUME 115, ISSUE 3, P605-606, MARCH 01, 2021


Didier Dewailly, M.D.


Since the Rotterdam consensus in 2003, the ultrasound aspect of polycystic ovary morphology (PCOM) has been part of the diagnostic classification of polycystic ovary syndrome (PCOS), in association with at least 1 of the 2 items “oligo-anovulation” and “hyperandrogenism.” At the time, the official definition of PCOM was several follicles of 2–9 mm per ovary (FNPO) ≥12 or an ovarian volume (OV) ≥10 mL. Since then, this definition has evolved, mainly due to the improved performance of ultrasound equipment, which now offers images with much better resolution because of high-frequency endovaginal probes (≥8 MHz). This evolution exclusively concerns the FNPO, the measurement of OV not impacted by technical progress, but this parameter is much less sensitive than the FNPO, which remains the main measurement. With the improvement of follicle counting, the FNPO threshold ≥12 to define PCOM has become obsolete, and a recent international guideline recommends a threshold ≥20 with high-frequency endovaginal probes (≥8 MHz) (1).

Read the full text here. 


Go to the profile of Heidi Vanden Brink
over 1 year ago

Response to Editorial by Didier Dewailly

By: Heidi Vanden Brink and Marla Lujan

An engaging and thoughtful response by Dr. Dewailly is well received and brings to light excellent points regarding the clinical utility versus physiological relevance of obtaining follicle counts on ultrasonography. We wish to draw attention to the relevance of our findings as they pertain to the diagnostic accuracy of polycystic ovarian morphology (PCOM) and to stress the need for precision in follicle counts to ensure the efficacy of translational research in this area. Last, we provide our thoughts on how to move forward.

Although we documented a risk of false positives in the detection of PCOM by overcounting follicles, the focus on overcounting needs to be interpreted with caution. The rater in our study may not be representative of the average clinician, as the rater had extensive training in ovarian image analysis as part of our research group. In our experience, without training or significant experience, raters are prone to skipping over follicles, particularly in cases of follicle clustering, poor image quality, and ovarian enlargement, which ultimately results in an underestimation of follicle number. Until other investigators report on their levels of agreement using real-time versus off-line methods, we cannot be certain of any propensity for over- or under-diagnosis of PCOM at this time. As such, we feel the important take-home message of our study relates to the risk of misdiagnosis in general when using real-time approaches.

Continuing to use two-dimensional real-time (2D-RT) approaches to obtain follicle counts and accepting its relative imprecision will not address the current controversy related to the clinical utility or biological relevance of PCOM. Rather, we propose that it may be worth counting follicles with more reproducible methods and attempt to establish best practices for the clinical evaluation of ovarian morphology. PCOM enables the full phenotypic classification of PCOS, which we know exists on a spectrum of severity. We and others have shown that ultrasonographic features of PCOM reflect the degree of reproductive disturbance and emerging research supports that follicle populations may also reflect the degree of metabolic involvement in reproductive dysfunction1. With efforts now starting to parse out reproductive versus metabolic etiologies towards PCOS phenotypes, reproducible documentation of ovarian features and the presence or absence of PCOM are, in our opinion, more important than ever. If we command precision at the level of research, but not in clinical practice, this leads to a lack of external validity of research methods and greatly impacts their translational potential.

So, how do we command precision if we cannot spend 20 minutes per ovary conducting rigorous, off-line assessments? We agree that 2D-Grid is laborious and not suitable for clinical use. However, this is where we see the promise of 2D-RT with Grid or an analogous off-line version as we conducted in the present study. The premise of the grid overlay is to simply partition the ovary into more manageable compartments – particularly, in cases of ovarian enlargement and pronounced follicle clustering, wherein keeping a mental tally across a large area with new follicles constantly coming into view is extremely challenging – if not impossible. In its simplest iteration, the sonographer could simply perform real-time counts focusing on one half of the ovary at a time and summing the two mental tallies. Transitioning to an off-line option could be more preferable in certain clinical settings as acquiring 2D cineloops (videoclips of a sweep through the entire ovary) or 3D volume datasets affords the opportunity to complete a 2 – 3 minute assessment of follicle number per ovary (FNPO) using a variety of approaches after the patient (or participant) has left the examination. Importantly, off-line approaches allow for the ability to re-evaluate images and seek corroboration by others when cases are challenging.

We see a few ways to move the field forward. We have started to view ultrasound imaging analogous to an endocrine assay in that ultrasonographic assessments are subject to definable levels of variability, notably inter-method and inter-individual variability. We have contemplated whether FNPO is truly a single number, or perhaps an estimate with a confidence interval. Indeed, without histological confirmation, FNPO is at best an approximation. The implications of this notion are provocative. First, perhaps we no longer consider a diagnostic threshold, but rather a diagnostic range in order to account for the variability and to ensure clinical applicability. Alternatively or in concert, thresholds for PCOM could be internally defined based on setting-specific preferences for data collection and image analysis methods, which is consistent with the current International Guideline’s2 position for defining diagnostic thresholds for other cardinal features such as biochemical hyperandrogenism.

Our goal is to establish rigor and reproducibility of ultrasonographic imaging as it relates to Women’s Health. The ovary is an integration site of reproductive and metabolic inputs. Emerging research seeks to clarify these integrative pathways and has demonstrated that features of ovarian morphology no longer need to reflect simply binary diagnostic indicators of abnormalities. Rather, ovarian features may indicate intervention responsiveness 3–5, the severity of PCOS phenotype6–8 or degree of metabolic involvement in reproductive dysfunction1. There is no better time than now to shift our sights to establishing rigor in the ultrasonographic assessments of ovarian features. Ultimately, we believe that precision of care for women with PCOS would be best served by demanding precision in our ultrasonographic assessments in both clinical and research settings.



  1. Vanden Brink H, Pea J, Lujan ME. Ultrasonographic features of ovarian morphology capture nutritional and metabolic influences on the reproductive axis: implications for biomarker development in ovulatory disorders. Curr Opin Biotechnol. 2021;70:42-47. doi:10.1016/j.copbio.2020.10.008
  2. Teede HJ, Misso ML, Costello MF, Dokras A, Laven J, Moran L, Piltonen T, Norman RJ, Andersen M, Azziz R, Balen A, Baye E, Boyle J, Brennan L, Broekmans F, Dabadghao P, Devoto L, Dewailly D, Downes L, Fauser B, Franks S, Garad RM, Gibson-Helm M, Harrison C, Hart R, Hawkes R, Hirschberg A, Hoeger K, Hohmann F, Hutchison S, Joham A, Johnson L, Jordan C, Kulkarni J, Legro RS, Li R, Lujan M, Malhotra J, Mansfield D, Marsh K, McAllister V, Mocanu E, Mol BW, Ng E, Oberfield S, Ottey S, Peña A, Qiao J, Redman L, Rodgers R, Rombauts L, Romualdi D, Shah D, Speight J, Spritzer PM, Stener-Victorin E, Stepto N, Tapanainen JS, Tassone EC, Thangaratinam S, Thondan M, Tzeng CR, Van Der Spuy Z, Vanky E, Vogiatzi M, Wan A, Wijeyaratne C, Witchel S, Woolcock J, Yildiz BO. Recommendations from the international evidence-based guideline for the assessment and management of polycystic ovary syndrome. Hum Reprod. 2018;33(9):1-17. doi:10.1093/humrep/dey256
  3. Christ JP, Falcone T. Bariatric Surgery Improves Hyperandrogenism, Menstrual Irregularities, and Metabolic Dysfunction Among Women with Polycystic Ovary Syndrome (PCOS). Obes Surg. 2018;28(8):2171-2177. doi:10.1007/s11695-018-3155-6
  4. Kazemi M, Pierson RA, McBreairty LE, Chilibeck PD, Zello GA, Chizen DR. A randomized controlled trial of a lifestyle intervention with longitudinal follow-up on ovarian dysmorphology in women with polycystic ovary syndrome. Clin Endocrinol (Oxf). 2020;92(6):525-535. doi:10.1111/cen.14179
  5. Nylander M, Frøssing S, Clausen H V., Kistorp C, Faber J, Skouby SO. Effects of liraglutide on ovarian dysfunction in polycystic ovary syndrome: a randomized clinical trial. Reprod Biomed Online. 2017;35(1):121-127. doi:10.1016/j.rbmo.2017.03.023
  6. Alviggi C, Conforti A, Rosa P De, Strina I, Palomba S, Vallone R, Gizzo S, Borrelli R, Andersen CY, Placido G De, Guerriero S. The distribution of stroma and antral follicles differs between insulin-resistance and hyperandrogenism related polycystic ovarian syndrome. Front Endocrinol (Lausanne). 2017;8(MAY):1-7. doi:10.3389/fendo.2017.00117
  7. Reid SP, Kao C, Pasch L, Shinkai K, Cedars MI, Huddleston HG. Ovarian morphology is associated with insulin resistance in women with polycystic ovary syndrome : a cross sectional study. Published online 2017:1-7. doi:10.1186/s40738-017-0035-z
  8. Christ JP, Vanden Brink H, Brooks ED, Pierson R a., Chizen DR, Lujan ME. Ultrasound features of polycystic ovaries relate to degree of reproductive and metabolic disturbance in polycystic ovary syndrome. Fertil Steril. 2015;103(3):787-794. doi:10.1016/j.fertnstert.2014.12.094