MR volumetric measurements of the myomatous uterus: improved reliability of stereology over linear measurements.

Rationale and Objectives: Stereology is a simple, fast method for object segmentation that involves counting the number of intersections of a randomly positioned grid over an object. The objectives of this study were to determine observer reliability in making stereologic- and ellipsoid-based measurements of uterine and leiomyoma volumes and to test the agreement between these two methods of measurement.
Materials and Methods: Two observers made uterine and dominant leiomyoma volume measurements on MR images in 30 patients using stereology and the popular ellipsoid-based technique. Stereologic volume measurements were made from high-resolution T2 images in two perpendicular planes (axial and sagittal). Ellipsoid volume was calculated by multiplying the maximal sagittal, anteroposterior, and transverse dimensions by pi/6. For these measurements, interobserver reliability was tested with paired t-tests and percent differences were determined. A mean stereologic volume and a mean ellipsoid volume were determined and tested for agreement with a paired t-test. Percent differences were also calculated.
Results: Stereologic measurements demonstrated excellent interobserver reliability with 0.3% difference in mean uterine volumes (P = .69) and 0.3% difference (P = .81) in mean leiomyoma volumes. The ellipsoid method resulted in poorer interobserver reliability with 7% difference (P = .01) in mean uterine volumes and 4% difference (p = .24) in mean leiomyoma volumes. The ellipsoid method also significantly overestimated uterine volumes by 14% (P < .01) compared with stereology.
Conclusion: Stereology provided high interobserver reliability for leiomyoma and overall uterine volume measurements and was more reliable than the ellipsoid method, which uses linear measurements. Stereology appears well suited when precise volume measurements are desired for assessing response to uterine arterial embolization treatments.