Article Text
Abstract
Introduction The ability to assess packing density during endovascular treatment of aneurysms is limited by multiple factors. Algorithms to quickly predict the effect of coil choice on packing density are not routinely used during coil selection. Additionally, the accuracy of 3D rotational angiography volumes and mathematical models for estimating volume is unknown. The AngioSuite system was developed to expediently perform these calculations. The purpose of this study is to determine the accuracy of 3D rotational angiography, mathematical modeling, and the AngioSuite software system for volume and packing density analysis.
Methods Phantom aneurysm clay models were created to reflect common aneurysm appearances. Digital subtraction angiography was performed on an AXIOM biplane system (Siemens). Aneurysm dimensions were obtained from biplane angiograms after calibration was performed, and transferred to a Leonardo workstation (Siemens) for analysis. Aneurysm dimensions and volumes were obtained from the 3D subtraction angiograms in the same projections, but at multiple window and level values. Measurements obtained from 2D and 3D angiograms were used to estimate volumes using ellipsoid and multi-lobular mathematical models (AngioCalc). Biplane angiographic images were loaded into the AngioSuite system (Cascade Endovascular). Volumes were calculated using calibration performed from the 2D measurements, 3D measurements, and finally, the fiduciary markers. Models were weighed in distilled water at 19.7°C on a Basic BA110 digital scale (Sartorius AG). The weight of the displaced water was measured. Based on the specific gravity of water at the known temperature, the volume of each phantom was then calculated.
Results The mean volumes of all phantoms analyzed with AngioCalc differed by 6.43%±18.05% from measured volumes, using both ellipsoid and bi-lobed models (calibrated 2D angiogram). Using dimensions from a 3D angiogram at the lowest threshold, volume measurements differed by 9.44% ±25.50%. Adjusting window levels for best appearance resulted in measurements differing by 10.95%±32.09%. Mean volumes calculated with the AngioSuite system differed from true volumes by 1.58%±4.99% (calibrated internally from a known marker). Using a 2D system to calibrate, volumes differed by 2.75%±9.44%. Measurements from a 3D system at the lowest threshold led to volumes that differed by 4.70%±11.46%. Windowing to best appearance resulted in volumes that differed by 8.92%±15.19% from the phantoms. Mean volumes calculated on the Leonardo workstation, at the lowest threshold, differed from true volumes by 2.13%±6.02%. Volumes obtained by windowing for best appearance differed by 5.18%±13.47%.
Conclusions The AngioSuite software system provides an easy and accurate means of calculating aneurysm volumes when calibrated from fiduciary markers, with volume accuracy of ±5%, similar to that of 3D rotational angiography. Typical packing densities (40% or less) are thus accurate within ±2%. Mathematical models, as well as 3D rotational angiographic volumes windowed for best appearance, are particularly inaccurate. When using 3D rotational angiography for aneurysm measurement, the user must set the threshold to an appropriately low level to improve accuracy.
Statistics from Altmetric.com
Footnotes
Disclosures B Woodward: Cascade Endovascular. D Forsberg: Cascade Endovascular.