Three-dimensional US volume analysis of gastric pseudotumors in a porcine model

Background: Tumor burden is difficult to estimate by endoscopy and conventional EUS. Objective: The purpose of this study was to determine the accuracy and the reliability of a new 3-dimensional (3D) EUS system in a pseudotumor model (Olympus EUS EXERA EU-M60). Design: A pseudotumor model was developed in a porcine stomach. Pseudotumors were created by injecting various volumes of US gel (0.3, 0.5, 0.7, and 1 mL) into porcine stomach specimens, and then the volume was measured in vitro. Two investigators made volume measurements by outlining the cross-sectional area of the pseudotumor at different radial planes. The instrument then automatically calculated the volume based on the outlined cross-sectional areas. The measured volume was compared with the actual volume of the pseudotumor by using a Bland-Altman analysis. Every second, third, fourth, fifth, sixth, and tenth image was measured to calculate the tumor volume and to determine the optimum number of images required for accurate volume determination. Inter- and intraobserver variability, percentage error, Bland-Altman analysis, analysis of variance (ANOVA), and kappa statistic were performed. Setting: This study was performed in an in vitro animal model. Patients: There were no patients involved in this study. Main Outcome Measurements: Accuracy and reliability of pseudotumor volume measurement. Results: When averaging across all measurements, the overall average mean error was 3.25%. The overall inter-rater reliability as measured by intraclass correlation coefficient was 0.78. The overall intra-rater reliability as measured by intraclass correlation coefficient was 0.99. Bland-Altman analysis and ANOVA showed similar low variability for measured volumes based on image frequencies for volume calculations between every other and every sixth image but greater variability for measured volumes based on every tenth image. Larger pseudotumors were measured with a slight decrease in mean percentage error. The kappa statistic for interobserver variability was .61, which demonstrated substantial agreement among observers. Limitations: The major limitation of this technology is the penetration of the US beam to evaluate large tumors, because the US transducer is high frequency (20 MHz) and, therefore, has a limited penetration. Conclusions: In conclusion, the new Olympus EUS EXERA EU-M60 3D US probe allowed for accurate volume measurements of small pseudotumors in porcine stomach model in vitro. There was substantial evaluator agreement, with a low interobserver variability. Larger pseudotumors were measured with a slightly lower percentage error than smaller pseudotumors. Volumes measured with a greater number of radial images were measured slightly more accurately. We plan to test this device in patients with GI tumors in the near future.