Pergamon Ultrasound in Med. & Biol., Vol. 20, No. 2, pp. 157-165, 1994 Copyright © 1994 Elsevier Science Ltd Printed in the USA. All rights reserved 0301-5629/94 $6.00 + .00 OOriginal Contribution IN VITRO EVALUATION OF THREE-DIMENSIONAL ULTRASONOGRAPHY IN VOLUME ESTIMATION OF ABDOMINAL ORGANS ODD HELGE GILJA, t NILS THUNE,* KNUT MATRE, t TRYGVE HAUSKEN, t SVEIN ODEGAARD t and ARNOLD BERSTADt *Medical Department A, Haukeland Hospital, University of Bergen and *Christian Michelsen Research, Bergen, Norway (Received 11 May 1993; in final form 23 August 1993) Abstract--The purpose of this study was to evaluate the accuracy and precision of a three-dimensional (3D) ultrasound system in volume estimation of abdominal organs in vitro. A mechanical sector scanner coupled to a stepping motor recorded images of porcine stomach and kidneys. The transducer was tilted through 88" yielding 81 frames, and volume estimation was performed digitally after interactive manual contour indication and organ reconstruction in 3 dimensions. This 3D system showed good correlation (r = 0.998) between estimated and true volumes. Volume estimation of stomach and kidneys using 3.25 MHz probe demonstrated limits of agreement of 0.877 to 1.146 and 1.007 to 1.125, respectively, depicting estimated volumes as a proportion of true volumes in 95% of the examinations. Intra- and interobserver variation of the tracing procedure revealed low values. We conclude that this 31) ultrasound system performs high accuracy and precision in volume estimation. Key Words: Abdominal organs, Kidney, Stomach, Three-dimensional reconstruction, Volume measurements, Ultrasonic imaging, Ultrasonography. INTRODUCTION Volume estimation of organs and pathologic tissue, like neoplasms, may be of significance in diagnostic procedures and follow up of diseases. Two-dimen- sional (2D) ultrasound has proven to be a valuable, noninvasive method in volume calculation of both pa- renchymal and hollow organs. However, volume calcu- lation performed with 2D ultrasound imaging is based on geometric assumptions (Szebeni and Beleznay 1992), which may introduce significant errors. Previ- ous studies on bladder volume measurements have demonstrated errors from 15% to 95% using 2D ultra- sound (Griffiths et al. 1986). Scanning systems for three-dimensional (3D) ul- trasound imaging have been developed and applied for volume estimation. An in vitro trial on six kidneys demonstrated worst error of -8.3% and no systematic bias in volume estimation (Brinkley et al. 1982). Other authors computed volumes in vitro on phantoms with a mean error of 1.6% using a phased-array transducer, Address correspondence to: Odd Helge Gilja, Medical Depart- ment A, Haukeland Hospital, University of Bergen, N-5021 Bergen, Norway. 157 and indicated that 3D ultrasound did not introduce sig- nificant new errors compared to 2D ultrasound (King et al. 1991, 1992). They argue that evaluation of one specific 3D ultrasound system should not be regarded as true for other 3D systems. In a previous study on phantoms with the present mechanical 3D ultrasound scanner, we found a mean error of 0.78 mL ___2.65 mL in the volume range 1- 45 mL, and an interobserver error of 0.6% ___ 5.0%. Ongoing clinical research on the stomach, the gallblad- der and liver tumors emphasize the need for method- validation on organs. The purpose of this study was to evaluate the accuracy and precision of a mechanical 3D ultrasound scanner for volume estimation on paren- chymal and hollow organs in vitro. MATERIALS AND METHODS Fresh animal organs were acquired from a local slaughterhouse, and both sheep and pig kidneys were harvested in order to obtain a wide range of sizes. Porcine stomachs were required for measurements of a fluid filled hollow organ. The layers of the porcine stomach wall have proven to resemble human layers