Comparison of two‐dimensional imaging to three‐dimensional modeling of intrahepatic portosystemic shunts using computed tomography angiography

Abstract

AbstractComputed tomography angiography (CTA) is used for the diagnosis of intrahepatic portosystemic shunts (IHPSS). When planning for transcatheter intervention, caudal vena cava (CVC) measurements are typically obtained from two‐dimensional (2D) imaging to aid in stent selection. We hypothesized that clinically applicable three‐dimensional (3D) IHPSS models can be generated, and CVC measurements will not differ between 2D images and 3D models. Computed tomography angiography datasets from client‐owned dogs with IHPSS at the University of Georgia Veterinary Teaching Hospital from 2016 to 2022 were analyzed. Materialise Mimics 25.0 and 3‐matic 17.0 were used for 3D modeling. Caudal vena cava diameters were measured in 2D dorsal and transverse planes 20 mm cranial and caudal from the shunt ostium and were compared with CVC diameters from 3D models. Length was measured in the 2D dorsal plane between midpoints of each diameter and compared to the 3D model length. Data are presented as mean (SD), and intraclass correlation coefficients were performed. Three‐dimensional models were generated for 32 IHPSS (15 right‐, 12 left‐, and five central‐divisional). Two‐dimensional dorsal and transverse area‐associated diameter measurements were 16.7 mm (5.6) and 15.5 mm (4.2) cranial; 14.9 mm (4.2) and 14.3 mm (3.7) caudal. Three‐dimensional area‐associated diameter measurements were 15.3 mm (4.4) cranial and 14.0 mm (3.6) caudal. The 2D length was 61.5 mm (7.1) compared with 3D 59.9 mm (7.2). Intraclass correlation coefficients comparing 2D and 3D diameters were all >0.80, indicating very good agreement, with good agreement (>0.60) for length. Clinically applicable 3D IHPSS models can be generated using engineering software. Measurements from 3D models are consistent with 2D planar imaging. Both 2D CTA and 3D virtual models can be utilized for preprocedural planning, depending on clinician preference.