Quantitative analysis of antral follicle number and size: a comparison of two-dimensional and automated three-dimensional ultrasound techniques

Transcription

1 Ultrasound Obstet Gynecol 2010; 35: Published online in Wiley InterScience ( DOI: /uog.7505 Quantitative analysis of antral follicle number and size: a comparison of two-dimensional and automated three-dimensional ultrasound techniques S. DEB, B. K. CAMPBELL, J. S. CLEWES and N. J. RAINE-FENNING University of Nottingham, School of Clinical Sciences, Division of Human Development, Nottingham University Research and Treatment Unit in Reproduction (NURTURE), Nottingham, UK KEYWORDS: anti-müllerian hormone; antral follicle count; antral follicle size; automated; ovarian reserve; reliability; SonoAVC TM ; three-dimensional ultrasound ABSTRACT Objectives To compare two-dimensional (2D) ultrasound imaging with automated three-dimensional (3D) ultrasound imaging for the measurement of antral follicle number and size. Methods Twenty-four subjects aged < 40 years underwent transvaginal ultrasound examination (Voluson E8) in the early follicular phase of the menstrual cycle. A 2D ultrasound scan of both ovaries was performed; each antral follicle was identified and then measured by taking the mean of two diameters. A 3D ultrasound dataset of both ovaries was then acquired and analyzed using Sonography-based Automated Volume Count (SonoAVC TM ). The time taken to measure the size of all antral follicles in both ovaries was recorded in seconds for each technique. Antral follicle size was recorded to the nearest millimeter and counts for each 1-mm group were obtained. Antral follicle counts were also grouped according to five predefined size categories: mm, mm, mm, mm and mm. Limits of agreement (LOA) and a paired t-test or Wilcoxon signed ranks test were used to analyze the data depending on their distribution. Results When antral follicle numbers were compared for each 1-mm follicle size group, 2D ultrasound imaging recorded more follicles measuring mm (mean ± SD, 4.11 ± 3.70 vs ± 2.31; P = 0.019) and mm (mean ± SD, 4.63 ± 4.86 vs ± 2.89; P = 0.013) than did SonoAVC. LOA were widest with follicles measuring mm (LOA, 6.38 and 3.43) and mm (LOA, 7.99 and 4.09). The antral follicle count in each of the five predefined size categories was significantly lower with SonoAVC than with 2D ultrasound imaging (P < 0.05). SonoAVC took significantly less time to measure the size and record the number of antral follicles than did 2D ultrasound imaging (mean ± SD, ± s vs ± s; P < 0.001). Conclusions Fewer antral follicles are evident overall when SonoAVC is used to analyze 3D ultrasound data. The clinical significance of this remains to be determined but the automated technique is significantly quicker than is making measurements using 2D ultrasound imaging. Copyright 2010 ISUOG. Published by John Wiley & Sons, Ltd. INTRODUCTION The antral follicle count has been shown to be a significant predictor of ovarian reserve 1 3. The test is performed using either two-dimensional (2D) or three-dimensional (3D) ultrasound imaging, and usually involves counting each and every visible antral follicle measuring 2.0 mm or more, with an upper limit of 9.0 mm 4 6 or 10.0 mm 1,7 9. However, other groups have used different cut-offs based on the hypothesis that the smaller follicles reflect the functional reserve within the ovary, and therefore the probable response to ovarian stimulation more closely The pool of recruitable follicles within each ovary is made of preantral and early antral follicles (0.2 2 mm) that are largely gonadotropin-independent, small antral follicles (2 6 mm) that are selectable owing to their responsiveness to gonadotropins, and larger antral Correspondence to: Dr. S. Deb, NURTURE, B Floor, East Block, Queen s Medical Centre, Nottingham NG7 2UH, UK ( shilpa.deb@nottingham.ac.uk) Accepted: 17 August 2009 Copyright 2010 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPER

2 Quantitative analysis of antral follicle number and size 355 follicles (> 6 mm) that are gonadotropin dependent 13. The follicle destined to become dominant is selected from this pool during the early follicular phase. With the exception of this dominant follicle, all other healthy follicles showing evidence of granulosa cell activity tend not to exceed 6 mm in diameter 14 such that larger follicles are invariably atretic 15. The size of the antral follicle may therefore be more important than the total number and more predictive of the functional ovarian reserve, defined as the expected response to controlled ovarian stimulation. The smaller antral and preantral follicles may also reflect the woman s true fecundity, and relate more closely to oocyte quality and provide a better predictor of the chance of pregnancy. Very few studies have considered the size of antral follicles other than to define the limits of follicular diameter above and below which follicles should not be included in the final count 10,12,16. Those studies that have considered the size of the follicles suggest that those measuring 2 5 mm in diameter correlate most strongly with successful assisted reproduction treatment outcomes 10,16,17. Anti-Müllerian hormone, an important and independent marker of ovarian reserve and a predictor of outcome following assisted reproduction treatment, is mainly produced by the small antral follicles measuring < 6 mm in diameter, rather than by larger follicles which are more likely to be atretic The standard technique used to quantify the size of an antral follicle involves calculation of the mean diameter from two linear measurements of the follicle 11.Thereare no standards to define how this should be performed and no studies to date have reported the intraobserver or interobserver reliability of quantitative antral follicle counts stratified according to the absolute size of the follicle. Measurement reliability is likely to be relatively poor as it will depend on both the identification and subsequent measurement of each follicle. There is a natural variation in the determination of the total number of antral follicles, both within and between observers 22. Furthermore, once a follicle has been identified, there will be further variability in objectively defining its size; such variability is likely to be more marked than that seen in its identification, which is both simple and subjective. Any variability is likely to be exaggerated when there are larger numbers of antral follicles, such as in the polycystic ovary, as more measurements are required overall 2,8. Sonography-based Automated Volume Count (SonoAVC TM, GE Healthcare Ultrasound, Zipf, Austria) represents a new method that could be either used on specific ultrasound machines (Voluson E8 and Voluson-i; GE Healthcare Ultrasound) or on a personal computer with the aid of 4D View software (GE Healthcare Ultrasound). SonoAVC facilitates the automated identification and measurement of antral follicles 23. We have already shown that SonoAVC, when used for follicle tracking during controlled ovarian stimulation, provides automated measurements of follicular diameter and volume that are more reliable and more valid in the assessment of volume than comparable estimations made using 2D ultrasound imaging 23,24. More recently, we have also demonstrated how SonoAVC can be used automatically to determine the total number of antral follicles and that these measurements are more reliable than manual measurements of 2D and 3D ultrasound datasets 25. This study was designed specifically, therefore, to compare measurements of antral follicle number and size made using SonoAVC with those made using 2D ultrasound imaging, in women about to undergo assisted reproduction treatment. METHODS Twenty-four subjects aged < 40 years planning to undergo assisted reproduction treatment for subfertility were recruited prospectively. Subjects with a history of previous ovarian surgery including ovarian cystectomy, ovarian drilling and unilateral oophorectomy were excluded from the study. If the entry criteria were met, subjects underwent transvaginal ultrasound examination during the early follicular phase, cycle days 2 5 in the menstrual cycle, immediately before treatment. Those found to have ovarian cysts or follicles measuring > 10 mm were excluded at this point. The study was performed according to a protocol approved by the Nottingham Research Ethics Committee and the hospital s Research and Development Department. Informed, written consent was obtained before the enrollment of any subject. A single investigator (J.S.C.) performed all of the ultrasound scans using a Voluson E8 Expert (GE Healthcare Ultrasound) and a 5 9-MHz transvaginal volume transducer, which has both 3D and 4D ultrasound scanning modes. This investigator measured the follicles manually using 2D ultrasound imaging, but the results were recorded by another observer (S.D.). The same observer recorded the follicle measurements made following the application of SonoAVC, which was performed by the first investigator (J.S.C.) after acquisition of the 3D ultrasound datasets. The investigator performing the ultrasound scans (J.S.C.) was not therefore aware of the exact measurements made by both methods. The ultrasound assessment involved a 2D ultrasound assessment of the pelvis to exclude any obvious pathology, followed by visualization of ovaries in their transverse and longitudinal planes. Antral follicle number and size were first measured using 2D ultrasound imaging and then, following the acquisition of a 3D ultrasound dataset, with SonoAVC. Our technique of volume ultrasound imaging has been described in detail before 25.Inbrief,itcomprised identification of the area of interest with a box which defined a region of interest and the subsequent acquisition of a 3D ultrasound dataset. This involved the acquisition of a series of 2D ultrasound images during an automated sweep of the ultrasound beam through a predetermined but adjustable angle and speed. For the purpose of this study we used a 90 angle, to ensure that data were acquired from the whole ovary. The high-quality, slow-sweep mode was used, which provides the greatest

3 356 Deb et al. number of 2D ultrasound planes and therefore the highest degree of resolution when the 2D ultrasound planes are reconstructed and displayed as a 3D ultrasound image. The ultrasound settings and technique of data acquisition were standardized and identical for all subjects: gain, 5; speckle reduction imaging, 2; enhance, 2; reject, 25; and harmonics, high. Once the dataset had been obtained, SonoAVC, integrated into the ultrasound machine, was used to provide an automated assessment of the follicles within each ovary. Measurement method 1: two-dimensional ultrasound imaging This method involved measurement of antral follicles during ultrasound assessment of the ovaries and represents the current standard for this determination. Once the ovary was located the observer used the transducer to scroll through the ovary in two planes, longitudinal and transverse, and observe the antral follicles. Each antral follicle was then identified and measured in turn until the whole ovary had been analyzed. Antral follicle size was calculated by taking the mean of two perpendicular diameters, one of which represented the largest dimension, of each follicle (Figure 1). The same process was then repeated for the contralateral ovary. Measurement method 2: SonoAVC The acquired 3D ultrasound datasets were displayed on the ultrasound machine (Voluson E8 Expert) in the multiplanar view. The image display was optimized and the render mode entered to generate a threedimensional volume of interest (VOI). The render box was adjusted to exclude as much extraovarian information as possible and to ensure that the whole ovary was included in the VOI. The threshold settings, which assign transparency associated with fluid to opaque voxels, were maintained for all datasets at a default setting of Figure 1 Transverse view of the ovary on two-dimensional ultrasound imaging. Each antral follicle is measured by taking the mean of two perpendicular linear measurements (calipers). low. Once the dataset had been positioned correctly, SonoAVC was implemented. SonoAVC identifies and quantifies hypoechogenic regions within a 3D ultrasound dataset and provides an automated estimation of their size. Each individual volume is given a specific color, and the automated measurements of its mean diameter (relaxed sphere diameter), its maximum dimensions (x, y, z diameters) and its volume are displayed in descending order from the largest to the smallest 26.The individual follicles were displayed with a specific color and shown together with their dimensions and relative sizes. Postprocessing, involving the manual identification of follicles not included in the automated analysis, was then used to ensure that all antral follicles were counted and measured. Each additional antral follicle identified in this way was given a new color and its dimensions were displayed together with those of follicles that were identified originally. The mean relaxed sphere diameter of each antral follicle in both ovaries displayed as d(v) was recorded and used for data analysis (Figure 2). This measurement was chosen as previous work by our group has shown that it equates more closely to the volume of follicles aspirated at the time of oocyte retrieval than the mean diameter obtained by averaging the three perpendicular (x, y, z) diameters 27. The relaxed sphere diameter is a true 3D ultrasound measurement in that it is derived from the 3D volume using the formula for a sphere and reflects the diameter the follicle would assume should it form a perfect sphere 24. Data collection and analysis The antral follicle population for each subject was recorded to the nearest millimeter, as this reflects the current resolution of the ultrasound system, starting from 2.0 mm up to a maximum of 10.0 mm. For further analysis the antral follicles were then grouped into five categories based on the absolute dimensions of the follicles to reflect the current literature defining which follicles constitute and contribute to the total antral follicle count: mm 10,17, mm 11, mm 16, mm 4,6,28 and mm 1,7 9. The time taken for the whole process was recorded to the nearest second. The clock was started as soon as the baseline scan had been completed and the decision was made to assess the antral follicle population. Before starting the clock the transducer was placed to show a longitudinal section of the uterus. Once the clock had been started, the measurement methods were performed in a random order commencing with either 2D or 3D ultrasound examination. The 2D ultrasound technique included the time taken to locate both ovaries, count all of the antral follicles, and measure and record their individual sizes. The 3D ultrasound technique also commenced with the probe positioned centrally showing a longitudinal view of the uterus and included the time required to locate each ovary, acquire the 3D ultrasound datasets, apply SonoAVC and any necessary postprocessing, and record

4 Quantitative analysis of antral follicle number and size 357 Figure 2 Multiplanar display of an ovarian three-dimensional ultrasound dataset where Sonography-based Automated Volume Count (SonoAVC) has been implemented to define the antral follicles automatically. Each individual follicle is given a specific color and its measurements are displayed on the right of the image in descending order of volume. the results of this analysis in terms of the total antral follicle number and the relative sizes of the follicles. The time taken for both the methods did not include the time taken in each subject to assess the pelvis and exclude pathology. Statistical analysis Statistical analysis was undertaken using SPSS version 15.0 (SPSS, Chicago, IL, USA). The distribution of the data was checked using normal probability plots and the Kolmogorov Smirnov test. The mean value and SD are given for normally distributed data, and the median and range for non-parametric data. The limits of agreement (LOA) as described by Bland and Altman were used to measure the differences between the two methods 29. Depending on the normality of the data, a paired student t- test or one-way ANOVA with Wilcoxon signed ranks test was used to test for differences between the measurement methods in the number of antral follicles overall and within each size category, and in the time taken for measurements. P < 0.05 was considered to be statistically significant. RESULTS Three subjects were excluded from the study, as they were found to have follicles measuring 10 mm; complete data were therefore available for 21 subjects, who had a mean ± SD age of ± 4.29 (range, ) years and body mass index of ± 2.32 (range, ) kg/m 2. The cause of their subfertility related to male factors (8/21; 38.09%), tubal disease (2/21; 9.52%) or was unexplained (11/21; 52.38%). The measurement data and time taken for analysis were both confirmed to be normally distributed. The mean time taken for the automated analysis of the 3D ultrasound dataset with SonoAVC was significantly less than that required for the 2D ultrasound assessment (mean ± SD, ± s vs ± s; mean difference ± SD, ± (95% CI, ); LOA, and 30.06; P< 0.001). Postprocessing was used in all cases and the time recorded for SonoAVC included the additional time needed for this. The mean time taken for SonoAVC without postprocessing was ± s. The mean time required for the postprocessing aspect was ± s. When the numbers of antral

5 358 Deb et al. Table 1 Number of antral follicles stratified by size to the nearest millimeter as measured by two-dimensional (2D) and automated three-dimensional (3D) Sonography-based Automated Volume Count (SonoAVC) ultrasound imaging Antral follicle count Follicle Mean ± SD Mean difference Limits of agreement size (mm) Real-time 2D 3D SonoAVC ± SD (95% CI) Upper Lower P* ± ± ± 2.19 ( 1.64 to 0.48) ± ± ± 2.50 (0.27 to 2.68) ± ± ± 3.08 (0.46 to 3.43) ± ± ± 1.75 ( 0.89 to 0.79) ± ± ± 1.37 ( 0.92 to 0.40) ± ± ± 1.08 ( 0.31 to 0.73) ± ± ± 1.03 ( 0.71 to 0.29) ± ± ± 0.79 ( 0.17 to 0.59) *Paired samples t-test. Table 2 Comparison of two-dimensional (2D) and automated three-dimensional (3D) Sonography-based Automated Volume Count (SonoAVC) ultrasound techniques for the assessment of total antral follicle count according to absolute size of antral follicles Antral follicle count Mean ± SD Mean difference Limits of agreement Follicle size (mm) Real-time 2D 3D SonoAVC ± SD (95% CI) Upper Lower P* ± ± ± 4.03 (0.90 to 4.79) ± ± ± 3.19 (3.72 to 6.80) < ± ± ± 2.56 (1.50 to 3.97) ± ± ± 2.27 (1.43 to 3.62) ± ± ± 2.40 (1.58 to 3.89) *Paired samples t-test. follicles were stratified and analyzed to the nearest millimeter, a significantly higher number of follicles measuring mm (mean ± SD, 4.11 ± 3.70 vs ± 2.31; P = 0.019) and mm (mean ± SD, 4.63 ± 4.86 vs ± 2.89; P = 0.013) were identified with 2D ultrasound than with the automated 3D ultrasound technique (Table 1). The LOA between the two methods were widest with follicles measuring mm (LOA, 6.38 and 3.43) and mm (LOA, 7.99 and 4.09) (Table 1). SonoAVC recorded a significantly lower number of antral follicles than did 2D ultrasound imaging when the total counts were considered and also when the follicles were grouped into each of the five predefined size categories (Table 2). The LOA between the two methods were wide (Table 2). DISCUSSION This is the first study comparing an automated method (SonoAVC) to 2D in the measurement of size of antral follicles. Previous work by our own group has shown that the new automated technique provides a more reliable measure of total follicle number than 2D ultrasound imaging and manual assessment of 3D data both within and between observers 25. The present study extends this work and has shown that SonoAVC provides different results from those of 2D ultrasound imaging when the size of the follicle is considered. Although SonoAVC is an automated technique, postprocessing of the data was required in all cases. The technique is best considered as semiautomated therefore; although the additional processing adds to the assessment time, it is straightforward and the overall time required for analysis was still significantly less than that for measurements made during 2D ultrasound examination when the size of follicle size was also assessed. In both studies we found that SonoAVC identified and measured significantly fewer follicles than 2D ultrasound imaging, resulting in a consistently lower total antral follicle count regardless of the upper and lower size limits used to define the antral follicle population. This may be a true result or reflect fundamental differences in the measurement techniques. An obvious limitation of the present study is the inability to confirm the accuracy of these measurements, but histological work suggests that the automated measurements are more valid than those derived using 2D ultrasound imaging. Weenen et al. examined the ovaries

6 Quantitative analysis of antral follicle number and size 359 of 12 regularly cycling subjects aged between 19 and 44 years who had undergone prophylactic oophorectomy as they were genetically predisposed to an increased risk of ovarian cancer or because they had endometriosis 16. Follicles, at all stages of folliculogenesis, were measured in two perpendicular planes and the mean measurement taken as the diameter of the follicle. Histologically, an average of 18 follicles measuring 2 6 mm was evident in each ovary. This compared favorably with the measurements made with SonoAVC, which identified 20 such follicles, whereas 2D ultrasound imaging identified 26. A smaller number of follicles measuring 1 2 mm were evident histologically, approximately six in each ovary. SonoAVC identified approximately half this number whereas 2D ultrasound imaging failed to identify any. This almost certainly reflects the resolution limits of ultrasonography but may have been further confounded by the subjective nature of 2D ultrasound imaging. The ability of an observer to identify and reliably measure such small structures by 2D ultrasound examination is likely to be time dependent, but a degree of measurement inaccuracy would still be expected if more time was allowed for assessment. Although the subjects in the study by Weenen et al. may be different from those in the present study with regards to the timing of the assessment in the menstrual cycle and in the population studied, the data reported represent the best currently available on the human ovary against which ultrasound imaging can be evaluated. There are obvious limitations to the present evidence base and future validation studies may provide a stronger evidence base against which standards could be set. If we accept that the semiautomated 3D ultrasound measurements are more valid than those made by 2D ultrasound imaging, we need to consider why this new technique is better. It may relate to the image being displayed in a 3D ultrasound multiplanar view, which allows cross-checking of each follicle in three different planes, thus improving spatial orientation, but is more likely to reflect the fact that each follicle is color coded which prevents repeat measurements of the same follicle 24. As regards the objective assessment of follicle size, the 2D ultrasound technique derives the follicular diameter through estimation of the mean of two perpendicular linear measures, whereas SonoAVC uses volumetric information to define a relaxed sphere 24. This might explain the differences seen between the two techniques as 2D ultrasound imaging is more likely to disregard small irregularities in the follicles. Volumetric measurements made with SonoAVC have been shown to provide a more reliable and valid assessment of the estimated diameter of larger follicles than both 2D ultrasound imaging and manual assessment of 3D ultrasound datasets, regardless of the number of linear measurements taken to define the size of the follicle 23,27. Our results show that SonoAVC, despite being a semiautomated method as data postprocessing is invariably required, provides quicker measurements of antral follicle size than does 2D ultrasound imaging. There is least agreement between the two methods with antral follicles measuring mm. If the size of the follicles contributing to the overall antral follicle population is more important than the absolute size of the pool, as current animal and human evidence suggests, these findings have important implications for the research setting and clinical environment. Further research is required to evaluate the biological and clinical importance of quantifying antral follicle size and count. REFERENCES 1. Broekmans FJ, Kwee J, Hendriks DJ, Mol BW, Lambalk CB. A systematic review of tests predicting ovarian reserve and IVF outcome. Hum Reprod Update 2006; 12: Jayaprakasan K, Campbell BK, Clewes JS, Johnson IR, Raine- Fenning NJ. Three-dimensional ultrasound improves the interobserver reliability of antral follicle counts and facilitates increased clinical work flow. Ultrasound Obstet Gynecol 2008; 31: Verhagen TE, Hendriks DJ, Bancsi LF, Mol BW, Broekmans FJ. The accuracy of multivariate models predicting ovarian reserve and pregnancy after in vitro fertilization: a meta-analysis. Hum Reprod Update 2008; 14: Allemand MC, Tummon IS, Phy JL, Foong SC, Dumesic DA, Session DR. Diagnosis of polycystic ovaries by threedimensional transvaginal ultrasound. Fertil Steril 2006; 85: Lam PM, Raine-Fenning N. The role of three-dimensional ultrasonography in polycystic ovary syndrome. Hum Reprod 2006; 21: Nardo LG, Gelbaya TA. Evidence-based approach for the use of ultrasound in the management of polycystic ovary syndrome. Minerva Ginecol 2008; 60: Muttukrishna S, McGarrigle H, Wakim R, Khadum I, Ranieri DM, Serhal P. Antral follicle count, anti-mullerian hormone and inhibin B: predictors of ovarian response in assisted reproductive technology? BJOG 2005; 112: Scheffer GJ, Broekmans FJ, Bancsi LF, Habbema JD, Looman CW, Te Velde ER. Quantitative transvaginal two- and threedimensional sonography of the ovaries: reproducibility of antral follicle counts. Ultrasound Obstet Gynecol 2002; 20: Scheffer GJ, Broekmans FJ, Looman CW, Blankenstein M, Fauser BC, te Jong FH, te Velde ER. The number of antral follicles in normal women with proven fertility is the best reflection of reproductive age. Hum Reprod 2003; 18: Chang MY, Chiang CH, Hsieh TT, Soong YK, Hsu KH. Use of the antral follicle count to predict the outcome of assisted reproductive technologies. Fertil Steril 1998; 69: Haadsma ML, Bukman A, Groen H, Roeloffzen EM, Groenewoud ER, Heineman MJ, Hoek A. The number of small antral follicles (2 6 mm) determines the outcome of endocrine ovarian reserve tests in a subfertile population. Hum Reprod 2007; 22: Pellicer A, Gaitan P, Neuspiller F, Ardiles G, Albert C, Remohi J, Simon C. Ovarian follicular dynamics: from basic science to clinical practice. J Reprod Immunol 1998; 39: Gougeon A. Some aspects of the dynamics of ovarian follicular growth in the human. Acta Eur Fertil 1989; 20: Gougeon A. Ovarian follicular growth in humans: ovarian ageing and population of growing follicles. Maturitas 1998; 30: Gougeon A, Lefevre B. Evolution of the diameters of the largest healthy and atretic follicles during the human menstrual cycle. J Reprod Fertil 1983; 69: Weenen C, Laven JS, Von Bergh AR, Cranfield M, Groome NP, Visser JA, Kramer P, Fauser BC, Themmen AP. Anti-Mullerian hormone expression pattern in the human ovary: potential implications for initial and cyclic follicle recruitment. Mol Hum Reprod 2004; 10:

7 360 Deb et al. 17. Pellicer A, Ardiles G, Neuspiller F, Remohi J, Simon C, Bonilla- Musoles F. Evaluation of the ovarian reserve in young low responders with normal basal levels of follicle-stimulating hormone using three-dimensional ultrasonography. Fertil Steril 1998; 70: Gruijters MJ, Visser JA, Durlinger AL, Themmen AP. Anti- Mullerian hormone and its role in ovarian function. Mol Cell Endocrinol 2003; 211: Kevenaar ME, Meerasahib MF, Kramer P, van de Lang- Born BM, de Jong FH. Groome NP, Themmen AP, Visser JA. Serum anti-mullerian hormone levels reflect the size of the primordial follicle pool in mice. Endocrinology 2006; 147: Fanchin R, Mendez Lozano DH, Frydman N, Gougeon A, di Clemente N, Frydman R, Taieb J. Anti-Mullerian hormone concentrations in the follicular fluid of the preovulatory follicle are predictive of the implantation potential of the ensuing embryo obtained by in vitro fertilization. J Clin Endocrinol Metab 2007; 92: Nardo LG, Christodoulou D, Gould D, Roberts SA, Fitzgerald CT, Laing I. Anti-Mullerian hormone levels and antral follicle count in women enrolled in in vitro fertilization cycles: relationship to lifestyle factors, chronological age and reproductive history. Gynecol Endocrinol 2007; 23: Jayaprakasan K, Walker KF, Clewes JS, Johnson IR, Raine- Fenning NJ. The interobserver reliability of off-line antral follicle counts made from stored three-dimensional ultrasound data: a comparative study of different measurement techniques. Ultrasound Obstet Gynecol 2007; 29: Raine-Fenning N, Jayaprakasan K, Clewes J, Joergner I, Bonaki SD, Chamberlain S, Devlin L, Priddle H, Johnson I. SonoAVC: a novel method of automatic volume calculation. Ultrasound Obstet Gynecol 2008; 31: Raine-Fenning N, Jayaprakasan K, Clewes J. Automated follicle tracking facilitates standardization and may improve work flow. Ultrasound Obstet Gynecol 2007; 30: Deb S, Jayaprakasan K, Campbell BK, Clewes JS, Johnson IR, Raine-Fenning NJ. Intraobserver and interobserver reliability of automated antral follicle counts made using three-dimensional ultrasound and SonoAVC. Ultrasound Obstet Gynecol 2009; 33: Raine-Fenning N, Jayaprakasan K, Chamberlain S, Devlin L, Priddle H, Johnson I. Automated measurements of follicle diameter: a chance to standardize? Fertil Steril 2008; 91: Raine-Fenning N, Deb S, Jayaprakasan K, Clewes J, Hopkisson J, Campbell B. Timing of oocyte maturation and egg collection during controlled ovarian stimulation: a randomized controlled trial evaluating manual and automated measurements of follicle diameter. Fertil Steril 2009 Mar 31 [Epub ahead of print]. 28. Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshop group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod 2004; 19: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: