A Dynamic Phantom Model for Research and Quality Control in Cardiac Imaging

Abstract

Background. Accurate myocardial function and chamber quantification is of major interest. Lack of standardization between the different vendors, difficulties in performing validation studies and absence of reference systems for calibration have slowed down the expansion of such technologies in clinical practice. The goal of this study was to build a dynamic cardiac phantom to enable in vitro assessment of echo software algorithms. Methods. Using a polyvinyl alcoholic gel, we built a multimodality phantom model. Three pneumatic cylinders and a computer-driven control system allowed a 3D deformation capability. Sonomicrometer crystals were positioned on the phantom and used as reference for strain. The transducer tip was successively fixed at the apex of the gel for longitudinal strain assessment. Peaks of strain obtained by echo were then compared to the strain recorded by the sonomicrometers. The phantom was also scanned using an ultrasound machine with 3D capabilities and an MRI machine. MRI-volumes were compared to those obtained by 3D-echo. Results. We were able to apply various levels of longitudinal strain (-5 to -22%), and there was a strong and significant correlation between strain measured by tissue Doppler and sonomicrometers (R2 = 0.91, P = 0.0001) as well as between measurements by speckle tracking and sonomicrometers (R2 = 0.97, P < 0.0001). There was also a significant correlation between the volumes assessed by 3D-echo and MRI (R² = 0.94, P < 0.0001). Conclusion. This cardiac phantom model demonstrates realistic and complex deformation and is a promising tool to improve new echo algorithms, test their accuracy and standardize the measurements between different providers.