(G-273) Impact of Vessel Length on Geometric Characterization of Patient-Specific Total Cavopulmonary Connections

For pediatric patients with single ventricle heart defects, the common palliative surgical technique is the Fontan procedure. This process reroutes the inferior vena cava (IVC) and superior vena cava (SVC) flow directly to the pulmonary artery (PA), creating the total cavopulmonary connection (TCPC). While improving life expectancy and short-term outcomes, patients are at risk of developing long-term complications. Geometric characteristics of Fontan TCPCs, such as angles between vessels and minimum and average vessel diameters, have furthermore been linked with clinically relevant hemodynamic parameters, which in turn have been linked to increased risk of complications for Fontan patients; therefore, accurately analyzing geometric characteristics of the Fontan pathway is of utmost importance.

               As all vessels of the TCPC are cropped in the model preparation phase, this study examines the impact of the length of the right and left pulmonary artery (RPA and LPA) and of the IVC and SVC on the measured diameter and angle data for each vessel, additionally comparing a novel method of calculating diameter, which utilizes cross-sectional area slices, to the previously used and tested method of using inscribed sphere (IS) radius. This study aims to establish an optimal process for geometry analysis of these models to ensure accurate results that can be used to increase efficacy of surgical planning and other applications.

Patient-individualized Fontan TCPCs were modelled using phase contrast magnetic resonance imaging (PC-MRI) from 15 patients. These models were then smoothed and cut with different lengths for the PA using Geomagic. Vascular Modeling ToolKit (VMTK) was, in turn, used to compute vessel centerlines and bifurcation vectors. This data was then used to calculate vessel diameters and angles between vessels. For the diameter calculations, VMTK computes two relevant pieces of information. These include a radius statistic calculated by a function that places a sphere of the greatest radius possible at each point on the centerline and a cross-sectional surface area function that computes the cross-sectional surface area (CSA) for each point on the center line. Both these statistics were used to compute diameters, assuming a circular cross-section of each vessel.

             These processes and VMTK's bifurcation vector functionality were used for TCPC models with vessel lengths of 0-15 mm (2.5mm increments) for each of the 15 patients. The average, minimum, and maximum diameters (Davg, Dmin, Dmax) were all recorded using both methods, and the CSA diameters were divided by their corresponding IS diameter values to find a percent difference between the two. These percentages were then averaged for each length group to illustrate the effect of the vessel length on divergence between the two methods. Bifurcation vectors were also used to find the angles between vessels, and the standard deviations for angle measurements were calculated for each patient.

                The IVC and SVC vessel length was found to have negligible effect on the calculated diameters; however, for the PAs (with both CSA and IS methods), increased length resulted in decreased average diameter. Comparing the methods, average percent difference (by length) ranged from +37-48% and +29-53% for D_avg of LPA and RPA’s, respectively. For both, the percent difference between the two methods’ results generally decreased as the length of the PA increased. The smallest percent differences were seen at 10 and 12.5 mm for LPA and RPA, respectively, with lengths above that slightly increasing the difference (< 1%). The D_min and D_max generally followed these same trends; however, those datasets were characterized by outliers in the 0-2.5 mm range showing much higher and lower percentages. The D_max also had much higher differences than the D_avg and D_min, having percentages upwards of +67% for the LPA and +93% for the RPA.

For the angle calculations, IVC/SVC length again had negligible effect, while lower PA lengths often resulted in angle measurements not being calculated by VMTK, but for 13/15 cases, a length of 7.5 mm or greater resulted in all angle measurements being computed. For the two remaining cases, a length of 10 mm and 15 mm were needed, respectively. In terms of the impact of the length of the angle values, most cases saw standard deviations of 5° or less. Exceptions had deviations of up to 15°; however, they were accompanied by outliers in angle magnitude in the 0-5 mm length range, further establishing that greater PA length increases the accuracy of measuring vessel angles.

In conclusion, a greater length for the PAs decreases the difference between the CSA and IS methods for calculating diameter while also ensuring accurate and complete angle calculation. The CSA method was found to result in minimum and average diameters from additional 20-40% of the tested IS diameters, while for the maximum diameters, this percent was much higher. With this said, increased PA length is linked to more effective geometric characterization, while further study is necessary for determining causes and implications of the IS/CSA method differences.