Prospective evaluation of automated follicle monitoring in 58 in vitro fertilization cycles: follicular volume as a new indicator of oocyte maturity Adela Rodrıguez-Fuentes, M.D., a Jairo Hernandez, Ph.D., a Rocıo Garcıa-Guzman, M.D., a Elena Chinea, M.Sc., a Laura Iaconianni, M.D., b and Angela Palumbo, M.D., Ph.D. a a Centro de Asistencia a la Reproduccion Humana de Canarias, La Laguna, Tenerife, Spain; and b Centro Ecografico EcoBI, Rome, Italy Objectives: To assess the practical use of SonoAVC in an IVF program, and to establish new criteria for hcg administration based on follicular volume. Design: Prospective clinical study. Setting: Private IVF Center. Patient(s): Fifty-eight women with infertility undergoing IVF. Intervention(s): Two dimensional (2D) and three dimensional (3D) scanning on the day of hcg administration. Main Outcome Measure(s): Image quality, mean follicular diameter obtained by 2D and 3D sonography, follicular volume, number of oocytes retrieved, number of mature oocytes, time needed for each examination. Result(s): Approximately 60% of the patients included in this study had good image quality and could be monitored by 3D scans with subsequent application of the SonoAVC software. When image quality is good, measurements obtained by the automated mode are comparable to those obtained manually in 62% of cases. Automated monitoring is significantly quicker than conventional manual monitoring. Follicles with a measured volume R0.6 cc on the day of hcg administration are associated with the finding of mature oocytes at the time of egg retrieval. Conclusion(s): SonoAVC allows reliable evaluation of stimulated ovaries, and may help us establish new criteria for timing hcg administration based on follicular volume estimation rather than follicular size. Software improvements are needed to improve universal patient use. (Fertil Steril Ò 2010;93:616 20. Ó2010 by American Society for Reproductive Medicine.) Key Words: Follicular monitoring, three-dimensional ultrasound, SonoAVC, follicular volume, in vitro fertilization Ovulation induction relies on monitoring of treatment using both ultrasound and plasma hormonal levels. Accurate follicular monitoring by transvaginal ultrasound is of paramount importance for the success of in vitro fertilization (IVF) (1). In conventional two-dimensional (2D) ultrasonography, both ovaries are visualized and all visible follicles are counted and measured. Usually, the two greatest diameters are measured for each follicle and the mean follicular diameter is calculated. Accurate monitoring is time consuming for the provider and for the patient. When many follicles are present in each ovary, the prolonged process may result in crowded waiting rooms and frustrated patients. Measuring Received December 23, 2008; revised February 10, 2009; accepted February 18, 2009; published online April 18, 2009. A.R-F. has nothing to disclose. J.H. has nothing to disclose. R.G-G. has nothing to disclose. E.C. has nothing to disclose. L.I. has nothing to disclose. A.P. has nothing to disclose. All instruments and software were purchased by our department, and no support was received by industry or instruments and software manufacturers. Reprint requests: Angela Palumbo, M.D., Ph.D., Centro de Asistencia a la Reproduccion Humana de Canarias, C/ Doctor Zamenhof 14, E-38240, La Laguna, Tenerife, Spain (FAX: 922 63 28 79; E-mail: apalumbo@ull. es). multiple follicles may also be inaccurate. In fact, inter- and intraobserver variation has been estimated to be approximately 20% when multiple follicles are present (2). General Electric recently released a new software, called SonoAVC (Automatic Volume Calculation), which allows automated measurement of follicular diameters and volumes from a three-dimensional (3D) volume of each ovary (3, 4). If this technology is proven reliable, it may become a great asset for fertility clinics. This system is highly likely to improve efficiency in busy IVF practices. More importantly, it may also prove to be more accurate than conventional follicular measurements using 2D technologies. One major innovation of this new technology is its ability to provide a rapid and reliable measurement of follicular volume. This should more accurately reflect follicular size than measurement of mean follicular diameter, particularly in hyperstimulated ovaries, which often contain very irregularly shaped follicles. Follicular volume can be also measured using Virtual Organ Computer-aided AnaLysis (VOCAL, GE Medical Systems, Zipf, Austria), a 3D tool that has been proven very reliable for volume assessment (5). However, application of VOCAL to individual follicles in stimulated ovaries is both time consuming and subject to error. 616 Fertility and Sterility â Vol. 93, No. 2, January 15, 2010 0015-0282/10/$36.00 Copyright ª2010 American Society for Reproductive Medicine, Published by Elsevier Inc. doi:10.1016/j.fertnstert.2009.02.058
The purpose of the present study was to test the clinical applicability of SonoAVC in a cohort of unselected patients undergoing follicular monitoring for IVF. The specific objectives were: [1] to assess the quality of the SonoAVC image in individual patients, [2] to investigate the correlation of SonoAVC measurements to 2D measurements, [3] to compare the time necessary for follicular monitoring using SonoAVC to traditional monitoring by 2D ultrasound, and [4] to evaluate the relationship between follicular volumes on the day of hcg administration and the number of mature oocytes retrieved and establish cutoff values of follicular volume that aid in the timing of hcg administration in IVF cycles. FIGURE 1 (a) Example of a patient with good quality of image. All follicles were picked up automatically and no postprocessing work was necessary. (b) Example of a patient with poor quality of image: note that one follicle needed manual measurement. MATERIALS AND METHODS Fifty-eight women undergoing ovulation induction for IVF were recruited prospectively and monitored using a General Electric Voluson E8 Expert instrument (GE Medical Systems, Zipf, Austria) under institutional review board approval. All patients underwent monitoring using both conventional 2D ultrasound and 3D automated follicle count on the day of hcg administration. Follicular monitoring of each ovary was first performed manually using conventional 2D measurements with documentation of the two greatest diameters of each follicle. This was followed by a 3D volume acquisition of each ovary. All scans were performed using standardized techniques by one of two trained ultrasonographers. The volumes obtained were then analyzed by the same investigator, first using VOCAL to delimit the ovarian contour, then using the SonoAVC software (SonoAVC, Automatic Volume Calculation: GE Medical Systems). The following variables were analyzed: quality of the 3D image, number of follicles, mean follicular diameter, follicular volume, number of oocytes retrieved, number and percent of mature oocytes, and time necessary to perform the study. Image Quality Quality of automated three-dimensional imaging was considered good when >90% of follicles were measured automatically and no or minimal post processing work was needed using the SonoAVC software. Imaging was considered medium poor when >10% of the follicles present required manual measurement and a significant amount of time was spent in postprocessing work (Fig. 1a and b). Correlation between 2D and 3D Datasets Measurements obtained by conventional 2D ultrasound were compared with those obtained by 3D ultrasound using the SonoAVC software. Time The following time intervals were recorded for each individual patient and compared: [1] time required to perform the 2D measurements, [2] time necessary for volume acquisition, [3] Rodrıguez-Fuentes. Measurement of follicle volume using SonoAVC. Fertil Steril 2010. time employed to apply the SonoAVC to each 3D volume, and [4] the sum of intervals 2 and 3. Follicular Volume To establish the cutoff value at which follicular volume (as measured on the day of hcg administration) correlates with the retrieval of mature oocytes, we calculated the ratio between selected distinct cutoff values and the number of mature oocytes retrieved, as well as their inverse values (the ratio between the number of mature oocytes and the number of follicles at or above each cutoff value). Results were tabulated and the follicular volume best approximating the number of mature oocytes was estimated by that cutoff value that resulted in both ratios approximating unity. Statistical analysis was performed using the SPSS software (SPSS Inc., Chicago, IL). For the purpose of this analysis data Fertility and Sterility â 617
obtained in each ovary were analyzed separately, because some patients only had one ovary and results obtained in the two ovaries of the same patient sometimes differed in terms of image quality. To establish the correlation between manual and automatic measurements we used the Pearson s correlation coefficient, with a significance level of P<.05. RESULTS Ninety-two ovaries from 58 patients were included in the analysis. Quality of image was good (3.1% of follicles needed manual measurement) in 53 (57.6%) ovaries and medium poor (28.13% of follicles needed manual measurement) in 39 (42.4%) ovaries (Table 1). Correlation between 2D and 3D Datasets When we analyzed the whole sample of 92 ovaries, the correlation between measurements obtained manually by conventional 2D ultrasound and those obtained using 3D ultrasound and the SonoAVC software was good for 51% of ovaries (P>.05; n ¼ 47), whereas significant differences between the two methods were detected in 49% of ovaries (P<.05; n ¼ 45) (Table 2a). However, when good and medium poor quality ovarian images were studied separately, good quality images allowed for good correlation (P>.05) in 62.3% of cases (n ¼ 33), whereas medium poor quality images allowed for good correlation (P>.05) in only 35.9% of cases (n ¼ 14) (Table 2b and 2c). Time The average time necessary to perform follicular monitoring in patients with >10 follicles was 9.6 minutes, compared with 5.6 minutes for automated monitoring. The latter time allotment includes both volume acquisition and data analysis. Volume acquisition itself (the time during which the patient occupies the ultrasound room) required an average of 2 minutes for automated monitoring. Thus, each patient would save an average of 7.6 minutes and the sonographer an average of 4 minutes per patient. Follicular Volume Manually calculated volumes were always higher than those obtained using automated methods. Follicles with an automated volume >0.6 cc on the day of hcg administration are likely to yield mature oocytes. Table 3 illustrates the relationship between the number of mature oocytes and different follicular volumes as measured on the day of hcg administration: when a cutoff of 0.6 cc is chosen, the maximum number of follicles at or above this cutoff is closest to the number of mature oocytes retrieved. In other words, 98.6% of follicles R0.6 cc will give mature oocytes, whereas all other volumes studied either underestimate or overestimate the number of mature oocytes. DISCUSSION This study demonstrates that automated follicular monitoring is reliable and quick, provided the image quality is good. In this investigation all scans were performed by the same two ultrasonographers using a standardized technique, thereby minimizing the intra- and interobserver variation. In busy IVF programs that may require the services of several different sonographers, automated scanning may improve the accuracy of monitoring to an even greater degree. At first glance the correlation between the manual and automated measurements may appear poor, and statistical analysis of the datasets shows significant differences in a high percentage of patients. These differences reflect the fact that, by definition, 2D methods exclude the third follicular diameter. When the measured follicle is other than spherical, the mean follicular diameter will typically be an overestimate. On the other hand, 3D measurements represent the real follicular size. Because criteria for hcg used in ovulation induction cycles are based on 2D methodologies, these criteria must be redefined. The most important innovation of automated follicle monitoring is the introduction of a new parameter, follicular volume, in clinical practice. Since the introduction of transvaginal ultrasound (6, 7), this probably represents the biggest breakthrough in reproductive imaging. Currently, most IVF programs base their decisions for hcg administration on diameter measurements of the two or three biggest TABLE 1 Quality of the ultrasound image in 92 ovaries. Number of ovaries Mean number of follicles measured manually (%) Good quality a 53 (57.6%) 3.1% Medium poor quality b 39 (42.4%) 28% a Less than 10% of follicles needed manual measurement. b More than 10% of follicles needed manual measurement. 618 Rodrıguez-Fuentes et al. Measurement of follicle volume using SonoAVC Vol. 93, No. 2, January 15, 2010
TABLE 2 Correlation between measurements obtained manually by conventional two dimensional ultrasound and those obtained by three dimensional ultrasound using the SonoAVC software. A) Correlation of manual versus automated in all patients No difference (P>.05) 47 (51%) Significant difference (P<.05) 45 (49%) B) Correlation of manual versus automated in good quality images No difference (P>.05) 33 (62.3%) Significant difference (P<.05) 20 (37.7%) C) Correlation of manual versus automated in medium poor quality images No difference (P>.05) 14 (64.1%) Significant difference (P<.05) 25 (35.9%) Note: Pearson s correlation was used for statistical analysis. follicles using 2D sonography. SonoAVC permits a quick estimation of the number of mature oocytes and the ratio of mature versus immature oocytes that is based on follicular volume, a completely novel concept. The reliability and reproducibility of follicular volume measurement by SonoAVC has been reported by Salama et al. (8), who found that the mean follicular diameter measured automatically at the time of egg retrieval is comparable to the actual follicular volume extrapolated from the amount of follicular fluid aspirated. Their study is conceptually different from ours in that we evaluated follicular volume on the day of hcg administration. Our goal was to test the hypothesis that it may be possible to predict the number of mature oocytes retrieved based on follicular volume as measured on the day of hcg administration. Our results indeed indicate that this is the case: the number of follicles with a volume at or above 0.6 cc on the day of hcg corresponds or is very close to the number of mature oocytes that will be retrieved. The lowest follicular volume associated with mature oocytes is 0.6 cc. In summary, SonoAVC provides relevant information on follicular volume that may help clinicians redefine and perhaps improve the ultrasound criteria for hcg administration. Because optimization of ovulation induction and timing of hcg administration is critical for in vitro fertilization success, the introduction of follicular volume as a new parameter in clinical practice may contribute to improved TABLE 3 Relationship between the number of mature oocytes and different follicular volumes. Follicle volume cutoff Number of mature oocytes/number of follicles in range Number of follicles in range/number of mature oocytes R0.8 cc 121.26% 82.47% R0.7 cc 109.07% 91.69% R0.6 cc 100.88% 98.60% R0.5 cc 94.11% 106.26% R0.4 cc 88.74% 112.69% Note: When a cutoff of 0.6 cc is chosen, the maximum number of follicles at or above this cutoff is closest to the number of mature oocytes retrieved. Follicular volumes are those measured on the day of hcg. Fertility and Sterility â 619
pregnancy rates, especially in the most difficult cases, such as high or poor responders. In our hands, after optimization of presets, the time necessary for 3D scanning of both ovaries is 5.6 minutes (including data analysis). The time necessary for 2D scanning is much more variable and depends on optimization of the image as well as the number of follicles to be measured. As expected, the amount of time saved using the automated technique increases with the number of follicles to be measured. Further, the time required for analysis does not require patients to be in the office, which improves efficiency and increases patient convenience and acceptability. At present, the biggest limitations of SonoAVC are the result of poor image quality in a proportion of patients. After application of SonoAVC to a large number of patients, we now estimate that approximately 5% of cases cannot be monitored by the automatic mode (analysis of the 3D image is difficult and live scanning is required) and another 15% require such extensive postprocessing work that the automatic mode is not useful. In the remaining 80% of patients, however, the advantages of the automatic mode are clearcut: [1] increased accuracy of measurements, [2] improved efficiency, and [3] decision making based on follicular volumes. In conclusion, we believe that SonoAVC introduces a new and important perspective on follicular monitoring. The concept of follicular volume should improve our reliability in choosing follicular measurements that predict mature oocytes. Specific software improvements are still needed to improve the consistency of image quality. Acknowledgments: We thank Danny J. Schust, M.D., for his critical review of the manuscript. REFERENCES 1. Wittmaack FM, Kreger DO, Blasco L, Tureck RW, Mastroianni L Jr, Lessey BA. Effect of follicular size on oocyte retrieval, fertilization, cleavage, and embryo quality in in vitro fertilization cycles: a 6-year data collection. Fertil Steril 1994;62:1205 10. 2. Deutch TD, Abuhamad AZ, Matson DO, Bocca S, Stadtmauer LA, Oehningr SC. Automated calculation of ovarian follicular diameters using three dimensional sonography in women undergoing in vitro fertilization (IVF): a prospective evaluation of a novel software. Fertil Steril 2007;88: S80. 3. Deutch TD, Abuhamad AZ, Matson DO, Bocca S, Stadtmauer LA, Oehningr SC. Clinical evaluation of a novel automated follicular assessment software program in women undergoing controlled ovarian hyperstimulation for in vitro fertilization using three dimensional sonography. Fertil Steril 2007;88:S80. 4. Raine-Fenning N, Jayaprakasan K, Clewes J. Automated follicle tracking facilitates standardization and may improve work flow. Ultrasound Obstet Gynecol 2007;30:1015 8. 5. Raine-Fenning NJ, Clewes JS, Kendall NR, Bunkheila AK, Campbell BK, Johnson IR. The interobserver reliability and validity of volume calculation from three-dimensional ultrasound datasets in the in vitro setting. Ultrasound Obstet Gynecol 2003;21:283 91. 6. Timor-Tritsch IE, Bar-Yam Y, Elgali S, Rottem S. The technique of transvaginal sonography with the use of a 6.5 MHz probe. Am J Obstet Gynecol 1988;158:1019 24. 7. Shapiro BS, DeCherney AH. Ultrasound and infertility. J Reprod Med 1989;34:151 5. 8. Salama S, Arbo E, Lamazou F, Levaillant J-M, Frydman R, Fanchin R. Reproducibility and reliability of automated volumetric measurement of preovulatory follicles: is it time to switch towards follicle volumes? Fertil Steril 2008;90:S20. 620 Rodrıguez-Fuentes et al. Measurement of follicle volume using SonoAVC Vol. 93, No. 2, January 15, 2010