Transvaginal three-dimensional ultrasound: reproducibility of ovarian and endometrial volume measurements

Transcription

1 FERTLTY AND STERL'fi' Copyright 1996 American Society for Reproductive Medicine Vol. 66, No.5, November 1996 Printed on acid free paper in U. S. A. Transvaginal three-dimensional ultrasound: reproducibility of ovarian and endometrial volume measurements Amma Kyei-Mensah, M.D.*t Noreen Maconochie, Ph.D. Jamal Zaidi, M.D.* Rudi Pittrof, M.D.* Stuart Campbell, M.D.* Seang Lin Tan, M.D.*~ The London Women's Clinic, University College London Hospitals, King's College School of Medicine and Dentistry, London, United Kingdom; Radcliffe nfirmary, Oxford University, Oxford, United Kingdom; and Royal Victoria Hospital, McGill University, Montreal, Quebec, Canada Objective: To assess the reproducibility of ovarian and endometrial volume measurements obtained using transvaginal three-dimensional (3D) ultrasound (US). Design: Prospective clinical study. Setting: A tertiary referral center for assisted reproduction. Patients: Forty women undergoing ovarian stimulation for VF-ET using the long protocol of GnRH agonist.. ntervention: Three observers independently measured 20 stored ovarian scanned volumes and 20 endometrial volumes. Also, ovarian volume was calculated from three diameters obtained by two-dimensional (2D) US. Main Outcome Measure: Analysis of variance, the paired Student's t-test, and calculation of intraclass and interclass correlation coefficients were used for statistical analysis. Results: Three-dimensional ovarian volume measurements were not significantly different from ovarian volume calculated from three diameters (7.98 versus 7.58 ml). The mean endometrial volume measurement was 3.56 ml. The intraobserver coefficient of variation for both ovarian and endometrial volume was 8%. The interobserver coefficient of variation was 9% for ovarian volume and 11% for endometrial volume. ntraclass and interclass correlation coefficients were 0.95 and 0.95 for ovarian volume and 0.90 and 0.82, respectively, for endometrial volume. Conclusion: Transvaginal 3D US produces highly reproducible ovarian and endometrial volume measurements. Fertil Steril 1996;66: Key Words: Reproducibility, three-dimensional US Real time two-dimensional (2D) pelvic ultrasonography is a relatively accurate and reliable method of determining ovarian size and morphology (1, 2) Received February 26, 1996; revised and accepted June 5, * The London Women's Clinic. t Department of Reproductive Endocrinology, University College London Hospitals. Reprint requests and present address: Amma Kyei-Mensah, M.D., Maternity Unit, Department of Obstetrics and Gynaecology, Royal Berkshire Hospital, London Road, Reading RG1 5AN, Berkshire, United Kingdom (FAX: ). Department of Public Health and Primary Care, Radcliffe nfirmary. Academic Department of Obstetrics and Gynaecology, King's College School of Medicine and Dentistry. ~ Department of Obstetrics and Gynaecology, Royal Victoria Hospital, McGill University. and for monitoring follicular and endometrial growth during normal menstrual cycles as well as in assisted conception cycles (3, 4). The advent of computerized three-dimensional (3D) ultrasound (US) systems allowing the acquisition and storage of volume data permits visualization of the transverse plane of the pelvis for the first time. This ultrasonic innovation has diverse applications, e.g., assessment of embryonic brain development (5), evaluation of fetal malformations (6), and detection of congenital uterine abnormalities (7). This study was designed to determine the reproducibility of transvaginal 3D US volume measurements of the ovary and endometrium, which must be established clearly before a significant role can be assigned to this technique in clinical practice. 718 Kyei-Mensah et al. Reproducibility of 3D US volume measurements Fertility and Sterility

2 MATERALS AND METHODS n this study 40 patients were recruited from a group of women with regular menstrual cycles who were undergoing ovarian stimulation for VF -ET using the long protocol of GnRH agonist therapy (8). The study was approved by the nstitutional Review Board of the clinic, and all patients gave informed consent. All 3D scans were obtained by A.K.-M. using the Combison 530 system (Kretztechnik AG, Zi pf, Austria) with a Voluson transvaginal 7.5-MHz volume transducer (Kretztechnik AG). This system provides views of the coronal or C-plane, which is parallel to the transducer face in addition to conventional 2D US. Twenty patients were scanned on day 2 of the menstrual cycle to provide ovarian scanned volumes, and an additional 20 were scanned on the day when hcg was administered to trigger final oocyte maturation before oocyte collection to provide endometrial scanned volumes. All scanned volumes were stored digitally. The ultrasonographic investigation of each patient involved systematic examination of the morphology of the uterus and ovaries in 2D mode, as previously described (1). Then the system was switched into "volume mode," leading to the appearance of a moveable sector on the screen, which defined the "region of interest." The ovary (or endometrium) was centralized carefully within this sector and the patient was instructed to remain very still. The fast volume acquisition setting using low line density was activated. Volume data was captured by holding the transducer stationary while its crystal mechanically rotated through 360 for about 8 seconds, storing the resulting sections in the computer memory. The scanned ovarian or endometrial volume was displayed on the screen in three orthogonal planes. The perpendicular orientation of these planes is maintained throughout any translation or rotation so that serial translation actually produces an US tomogram from which volumetric data can be captured and stored digitally on an 88 MB removable cartridge hard disc (SyQuest; Technology nc., Fremont, CA) for subsequent analysis. Reliability of the 3D volume data was assessed by three observers (A.K.-M., J.Z., and R.P.) who were all experienced in pelvic ultrasonography. They independently analyzed all the stored scans and were unaware of each other's results. Three-dimensional volume measurement involved highlighting the plane that permitted the clearest view of the ovarian endometrial outline across its entire width. A green border appeared, and any contour changes reting from serial sections were subsequently red to this plane. The location of the section was 0 indicated in the other planes by the coordinated movement of small dot cursors and arrows. Starting from the tissue margin, the changing contour of the ovary in the measurement plane was out~ lined using a rollerball calibrator (Fig. 1). Similarly, the changing contour of the endometrial margin was outlined at the myometrial-endometrial interface (Fig. 2, A and B). For ovarian volume measurement, 10 serial slices were taken across the ovary from one pole to the other. Endometrial measurement involved taking serial sections from the fundal part of the endometrial cavity down to the level of the internal os. The area and volume of each section appeared simultaneously at the bottom of the screen, but these measurements were concealed until completion of the last section to prevent operator bias. Each ovary and endometrium was measured three times by each observer and the volumes were recorded. When analyzing the data, the three replicate measurements on each subject made by each observer were summarized as mean values. Ovarian volume was calculated also using the conventional formula (D! X D2 X D3 }7r/6, where D was the maximum diameter in the transverse, anteroposterior, and longitudinal axes of the scanned volume. This is the method employed with 2D US and therefore provided a means of comparing the two methods of volume measurement without the additional bias caused by using a different ultrasound machine. The reproducibility of measurements of ovarian and endometrial volume was determined by estimating components of variance using analysis of variance for repeated measurements (9). Subjects and observers were assumed to be chosen randomly from their respective populations, and the data were Figure 1 Section of the ovary outlined for volume measurement using 3D ultrasonography. Kyei-Mensah et al. Reproducibility of 3D US volume measurements 719

3 used to compare the two methods of obtaining ovarian volume measurements (2D or 3D US). For each subject, the estimate of ovarian volume using a particular method was taken to be the mean of the three observer measurements. A P-value of <0.05 was considered statistically significant. RESULTS Figure 2 (A) Endometrial volume measurement using 3D ultrasonography. (B ) Endometrial volume measurement: identification of the end-point, i.e., internal os may be difficult and cause observer differences. checked for statistical normality before any analyses were performed. Repeatability was expressed as the intraclass correlation coefficient (10), Rintra, and the interclass correlation coefficient (11), Rinter. These indicators of how reliable a method is in terms of measuring a subject's "true" value lie between 0 and 1, where 1 indicates total reliability. They can be interpreted as the correlation between any two (or more) measurements of the same subject, when those measurements are made by the same (intraclass coefficient) or different (interclass coefficient) observers, and are calculated as the proportion of the total variance in the volume measurement that can be attributed to real (biological) subject differences. ntraobserver and interobserver ' coefficients of variation (12) were calculated also as (SDloverall mean) X 100% using the estimates of within-observer and between-observer variance to obtain the SD as appropriate. The paired Student's t-test was 720 A paired t- test provided no evidence of a difference between the 3D and 2D US measurements of ovarian volume (P = 0.11). The intraobserver and interobserver coefficients of variation for 3D US measurement were low, at 8% and 9%, and at 8% and 11 %, for ovarian and endometrial volume, respectively (Table 1). The intraclass correlation coefficient (Rintra) for ovarian volume obtained using 3D US was 0.95: this indicates that a single measurement of the ovarian volume of anyone subject by anyone observer is 95% reliable as a measure of the subject's "true" ovarian volume. Rintra for endometrial volume obtained using 3D US was 0.90, indicating that a single measurement of endometrial volume in anyone subject by anyone observer is 90% reliable as a measure of the "true" endometrial volume. nterobserver agreement for 3D US measurements, expressed as Rinten was 0.95 for ovarian volume and 0.82 for endometrial volume, indicating a high level of agreement between any two measurements by different (randomly chosen) observers of any (randomly chosen) subject. DSCUSSON The most important finding in this study is the high degree of reproducibility of ovarian and endometrial volume measurements obtained using 3D US both within and between observers, demonstrated by high intraclass and interclass correlation coefficients. The results suggest a slightly higher degree of reproducibility of ovarian volume measurements, particularly between observers. This may be because the ovarian outline is usually well defined and outlined easily if scanning conditions are favorable. n contrast, even when sufficient endometrial volume data have been captured, considerable interobserver differences may occur in determining the position of the internal os during volume assessment, and also the endometrial-myometrial interface may be difficult t visualize, particularly in those patients with a type B endometrium (13), which is typically isoechogenic with respect to the myometrium. These variations will lead inevitably to a greater range of volume measurements and Kyei-Mensah et al. Reproducibility of 3D US volume measurements Fertility and Sterility

4 Table 1 Coefficients of Variation for 3D volume measurements* ntraobserver nterobserver Overall mean Within-observer Between-observer coefficient of coefficient of (ml) variance variance variation (%) variation (%) Ovarian volume (0.68) 0.46 (0.68) 8 9 Endometrial volume (0.28) 0.14 (0.37) 8 11 * Values are expressed as the mean of three measurements made by each observer, with SD in parentheses. SD = vvariance. higher interobserver variability. An investigation is assessed ideally in the population in which it will be most useful; in this study, the timing of scans was appropriate for the evaluation of physiologic or pathologic processes in the ovary and endometrium and the study population was well defined. Therefore 3D US can be recommended confidently as a reliable technique in these circumstances. The stored volumes produced by the Combison 530 US system (Kretztechnik AG) are in effect recorded real-time scans that can be analyzed days, weeks, or years later by experienced investigators. Effective use of this system does not preclude, however, scanning experience, and the ultrasonographer must have sufficient expertise to capture satisfactory volume data for storage and subsequent analysis. Three-dimensional US will be affected adversely by the same conditions that impair 2D US, e.g., obesity and poor tissue plane definition, so that several attempts may be needed before useful information can be acquired. mages that may have been visualized incompletely using real-time 2D US can be projected in any orientation using the 3D system, providing an opportunity for more detailed morphological examination that may clarify any diagnostic ambiguities. Any new measurement technique requires careful validation before being incorporated into clinical practice or research. n a separate study, we demonstrated that 3D US produces more precise follicular volume measurements over the clinically relevant volume range during VF-ET cycles (14). This information taken with the current study findings justifies investigation of possible clinical applications of 3D US in gynecology. The most promising clinical application of 3D US is likely to be in the detection of uterine anomalies, because the coronal or C-plane view parallel to the transducer face permits visualization of the endometrial outline between the uterine angles. Jurkovic et al. (7) assessed the potential value of 3D US for the diagnosis of congenital uterine abnormalities in 61 patients. There were no false-positive or false-negative diagnoses using this technique, and specificity and positive predictive value were higher compared with 2D US. Measurement of a fundal cleft or uterine septum, previously inaccessible with 2D US, can be achieved; in experienced hands 3D US may prove to be an efficient first-line investigation and possibly a safe, noninvasive alternative to hysterosalpingography. The advantage of 3D rather 2D US for ovarian follicular volume measurement has not been demonstrated clearly. There were no systematic differences in ovarian volumes calculated using either US measurement technique. Oocyte recovery and cleavage rates in VF-ET cycles have been shown to be higher in follicles with a volume of 3 to 7 ml (15), and further studies are now warranted to determine whether more precise follicular volume measurement actually translates into higher pregnancy rates during assisted conception treatment cycles. n cases of multi follicular development, follicle size and numbers can be assessed more easily using 3D US because systematic translation of the scanned volume reduces the likelihood of measuring the same follicle twice. The scan consultation time is not prolonged, and the storage facility means that patients do not need to be present for analysis. Three-dimensional US can be used to measure the volume of pelvic masses such as ovarian neoplasms or uterine fibroids. There may be a role for this technology in the field of tumor monitoring because serial examinations can be stored and may be used to provide an ultrasonographic record of any regression or growth of the tumor in response to therapy. Further studies are warranted to investigate this potential clinical application. There is no general consensus on the importance of endometrial thickness for prediction of implantation and pregnancy rates in assisted conception therapy. Some studies have suggested that measurement of endometrial thickness is a useful parameter (13), whereas other studies have disputed its value (16). The ability to quantify the volume of the endometrium using 3D US may help to resolve this issue because cycle outcome can be correlated with a quantitative parameter rather than endometrial thickness, which is prone to greater subjective variation in measurement. Similarly, the importance of endometrial volume in association with abnormal proliferative conditions, e.g., hyperplasia, carcinoma could be investigated using 3D US. f endometrial volume has more predictive value for carcinoma Vol. 66, No.5, November 1996 Kyei-Mensah et al. Reproducibility of 3D US volume measurements 721

5 l- than endometrial thickness, then 3D US may be useful as a screening test in postmenopausal patients. n summary, ovarian and endometrial volume measurements obtained using the Combison 530 3D system (Kretztechnik AG) are highly reproducible. A further advantage of this system is conferred by the on-line storage facility and its ability to display structures in the transverse plane of the pelvis. The validation of this technique should encourage further investigation of its use in clinical gynecology for diagnostic, monitoring, and therapeutic purposes. REFERENCES 1. Campbell S, Goessens L, Goswamy R, Whitehead M. Realtime ultrasonography for determination of ovarian morphology and volume. A possible early screening test for ovarian cancer? Lancet 1982;1: Goswamy RK, Campbell S, Royston JP, Bhan V, Battersby RH, Hall VJ, et al. Ovarian size in postmenopausal women. Br J Obstet Gynaecol 1988;95: Adams JM, Tan SL, Wheeler MJ, Morris DV, Jacobs HS, Franks S. Uterine growth in the follicular phase of spontaneous ovulatory cycles and during luteinizing hormone-releasing hormone induced cycles in women with normal or polycystic ovaries. Fertil Steril 1988;49: Shoham Z, Di Carlo C, Patel A, Conway GS, Jacobs HS. s it possible to run a successful ovulation induction program based solely on ultrasound monitoring? The importance of endometrial measurements. Fertil Steril 1991; 56: Blaas HG, Eik-Nes SH, Kiserud T, Berg S, Angelson B, Olstad B. Three-dimensional imaging ofthe brain cavities in human embryos. Ultrasound Obstet Gynecol 1995;5: Merz E, Bahlmann F, Weber G. Volume scanning in the evaluation of fetal malformations: a new dimension in prenatal diagnosis. Ultrasound Obstet Gynecol 1995;5: Jurkovic D, Geipel A, Gruboek K, Jauniaux E, Natucci M, Campbell S. Three-dimensional ultrasound for the assessment of uterine anatomy and detection of congenital anomalies: a comparison with hysterosalpingography and two-dimensional sonography. Ultrasound Obstet Gynecol 1995; 5: Tan S-L, Kingsland C, Campbell S, Mills C, Bradfield J, Alexander N, et al. The long protocol of administration of gonadotropin-releasing hormone agonist is superior to the short protocol for ovarian stimulation for in vitro fertilization. Fertil Steril 1992;57: Armitage P, Berry G. Components of variance. n: Armitage P, Berry G, editors. Statistical methods in medical research. Vol. 3. Oxford: Blackwell Scientific Publications, 1994: Armitage P, Berry G. ntraclass correlation. n: Armitage P, Berry G, editors. Statistical methods in medical research. Vol. 3. Oxford: Blackwell Scientific Publications, 1994: Scherjon SA, Kok JH, Oosting H, Zondervan HA. ntra-observer and inter-observer reliability of the pulsatility index calculated from pulsed Doppler flow velocity waveforms in three fetal vessels. Br J Obstet Gynaecol 1993; 100: Armitage P, Berry G. Measures of variation. n: Armitage P, Berry G, editors. Statistical methods in medical research. Vol. 3. Oxford: Blackwell Scientific Publications, 1994: Gonen Y, Caspar RF. Prediction of implantation by the sonographic appearance of the endometrium during controlled ovarian stimulation for in vitro fertilization. J n Vitro Fert Embryo Transfer 1990;7: Kyei-Mensah A, Zaidi J, PittrofR, Shaker A, Campbell S, Tan S-L. Transvaginal three-dimensional ultrasound: accuracy of follicular volume measurements. Fertil Steril 1996;65: 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 Steril1994; 62:1205-l. 16. Fleischer AC, Herbert CM, Sacks GA, Wentz AC, Entman SS, James AE Jr. Sonography of the endometrium during conception and nonconception cycles of in vitro fertilization and embryo transfer. Fertil Steril 1986;46:442-7., :1 722 Kyei-Mensah et al. Reproducibility of 3D US volume measurements Fertility and Sterility