Advanced Rotational X-Ray System with 3D Reconstruction (3DRx): Novel Clinical Applications in Endoscopic Retrograde Cholangiography (ERC)

Abstract

Introduction: Using conventional X-ray systems to perform ERC can be challenging and time-consuming due to image (IMG) Foreshortening (underestimating actual length) and Overlapping structures (F&O). This may lead to erroneous diagnoses, inaccurate measurements, or unsuccessful interventions. The newest C-arms systems can rotate rapidly around the pt to acquire multiple IMGs. These IMGs are subsequently sent to a digital workstation where they are reconstructed into a 3D volume dataset (VDS), which can be rotated freely to inspect the ducts of interest from different angles without administering extra radiation or contrast. This capability allows selection of the best working view with the least amount of IMG F&O. Methods: ERC was performed with a rotational C-arm system (Philips MultiDiagnost Eleva Flat Detector). Standard (Std) views were acquired first, followed by rotational X-ray IMGs at a rate of 15 fps as follows: pt is uncoupled from the respirator to eliminate respiratory motion artifacts; contrast is injected followed by a 6-sec 180o acquisition. 3D volumetric reconstruction was accomplished instantly using proprietary software designed for GI applications. To assess the discrepancy (% IMG foreshortening) between the Std view (L1) and the computer-predicted optimal view (L2), we used a validated 3D modeling technique to generate a 3D duct model of the Hilum (HLM) taken from the rotational IMGs of 7 pts. A quantitative software utility was used to select the optimal HLM view and to measure the %F&O. The target ducts extend 15 mm upstream (Rt/Lt main hepatic ducts; RHD/LHD) and 15 mm downstream (CHD) from the hilum. Additionally an IMG content study (# and longitudinal extent of lesions) was performed in 5 pts with HLM strictures. IMG contents acquired by Std and 3DRx views were reviewed in a double-blind fashion. Results: The optimal views of individual CHD-RHD (S1), CHD-LHD (S2), CHD-HLM (H1) segments were evaluated. The computer-predicted optimal gantry orientation (LAO-RAO, CRAN-CAUD) were: (37.57° ± 28.40°, 0.14° ± 13.83°) for H1, (29.71° ± 29.69°, -18.14° ± 12.06°) for S1, and (42.71° ± 19.44°, -12.29° ± 13.61°) for S2. The overall (average of H1, S1, S2) %F&O of L1 and L2 were (12.52% ± 9.93%, 9.76% ± 30.89%) and (1.48% ± 1.76%, 0.33% ± 0.91%), respectively. In IMG content comparisons, diagnoses based on Std views attributed duct strictures to wrong liver segments in 3/5 pts, and missed additional lesions seen on rotational acquisitions and/or 3D VDS in 4/5 pts. Conclusions: By rotating 3D structures to any projection, optimal patient-specific views of any duct of interest can be determined. For hilar pathology, 3DRx appears to improve diagnostic accuracy.