The reliability and validity of three dimensional ultrasound volumetric measurements using an in vitro balloon and in vivo uterine model.

Abstract

To evaluate the reliability and validity of two and three dimensional ultrasound volumetric measurements using balloon and uterine models. Prospetive observational study. Obstetric ultrasound department at a university teaching hospital. Two and three dimensional ultrasound volumetric measurements (with 5, 10 and 15 ultrasonic slices) were performed on 30 different sets of ultrasound images obtained from 15 water filled balloons with volumes ranging from 19 to 697mL. The measurements were performed independently by two observers who were blinded to the true volumes of the balloons. For the uterine model, only three dimensional ultrasonic volume measurements were performed independently on 16 uteri by two observers who were again unaware of the definitive uterine volumes. For the assessment of intra-and inter-rater reliability, the intraclass correlation coefficient was used. The index of concordance between the ultrasonic volumes and those obtained by the reference standard (validity) was assessed with the conventional Pearson's correlation coefficient, limits of agreement method and the intra-class correlation coefficient. High levels of reliability and validity were obtained for both two and three dimensional ultrasound balloon volume measurements. For two dimensional ultrasonic volume measurements, the intra-class correlation coefficient ranged from 0.992 to 0.998 for reliability and validity whereas the Pearson's correlation coefficient for validity was 0.996. With three dimensional ultrasonic volume measurements, the intra-class correlation coefficient ranged from 0.991 to 0.999 for reliability and validity whereas the Pearson's correlation coefficient for validity was 0.999. Both two and three dimensional ultrasonic measurements tended to underestimate the true balloon volume with the largest observed mean difference obtained with three dimensional ultrasound measurements using five ultrasonic slices and the smallest value obtained with three dimensional ultrasound measurements employing 15 ultrasonic slices. The mean difference in volume measurement for two dimensional ultrasound was intermediate between these two values. However, two dimensional ultrasound volume measurement generated the largest range between the limits of agreement whereas the smallest range was obtained with three dimensional ultrasound using 10 ultrasonic slices. The intra-class correlation coefficient for reliability and validity with three dimensional ultrasonic uterine volume estimation ranged from 0.956 to 0.996 whereas the Pearson's correlation coefficient for validity ranged from 0.993 to 0.999). The use of three dimensional ultrasound also consistently under-estimated the actual uterine volumes. The larger the number of ultrasonic slices employed for three dimensional ultrasound, the smaller was the mean difference between the ultrasonic and true uterine volume measurements and the smaller the limits of agreement. The reliability and validity of balloon and uterine volume measurement by three dimensional ultrasound is high. This allows further research on three dimensional ultrasound for measuring pelvic organ volumes in the prediction of pelvic pathology.