Abstract
2527
Objectives Accurate quantification of pulmonary vein measurements on CT is needed for a variety of preclinical and clinical applications. Several different processing techniques are available, but their comparability and accuracy is not well known. The objective was to evaluate 2 commercially available processing techniques and compare with pathology results in a large-animal model.
Methods Gated cardiac CT scans were done on 5 canines, both pre- and post-pulmonary vein ablation, using a 64-slice GE Discovery VCT. Multiple pulmonary veins were measured twice at the ostia on these animals using (1) a fully automated vessel analysis (VA) software and (2) a partially automated multiplanar reformat (MPR) software, both on the GE Advanced Windows Workstation. The animals were then euthanized and the software results were compared to pathology measurements. Unpaired T-tests were done to compare differences between these techniques.
Results Both the VA and MPR techniques performed well and were not significantly different from each other (Table). The MPR technique showed slightly lower numbers than VA, both pre- and post-pulmonary ablation. The pathology measurements were smaller than either automated technique. However, neither the VA (p=0.69) nor the MPR (p=0.62) was significantly different from the pathology measurements. Fairly high variances in the data could account for the lack of significance.
Conclusions The results of this study show that if done rigorously, both the fully automated VA software and the partially automated MPR software provide accurate and comparable results for the noninvasive measurement of pulmonary veins in canines. However, the MPR software, more than VA, trended toward the true pathology measurements. Ultimately, other factors (processing time, ease of use, data ouput, cost, etc.) may influence the choice of software
Table
Key: VA: Vessel Analysis, MPR: Multiplanar Reformat; SD: Standard Deviation