Computer-Aided Surgery System Using Three-Dimensional Tomographic Images for Endoscopic Retrograde Cholangiopancreatography (ERCP): Preliminary Study

Abstract

Endoscopic retrograde cholangiopancreatography (ERCP) is a widely used technique for diagnosing and treating biliopancreatic diseases such as biliary cancers and choledocholithiasis. It enables minimally invasive procedures such as stent placement and stone removal, and is commonly performed worldwide. However, selective cannulation of the bile and pancreatic ducts requires a high level of operator experience, and the technical difficulty and risks of complications, such as post-ERCP pancreatitis, remain major issues. This study proposes a computer-aided surgery (CAS) system that integrates preoperative CT with intraoperative fluoroscopy to assist bile and pancreatic duct cannulation. The proposed method generates digitally reconstructed radiographic (DRR) images from 3D CT data and evaluates similarity with ERCP images using normalized mutual information (NMI) and normalized cross-correlation (NCC). Registration is formulated as a six-degree-of-freedom (three rotational and three translational) optimization problem. In 20 clinical ERCP cases, the proposed method achieved stable convergence and improved similarity metrics. Visual assessments confirmed good alignment of vertebrae and ribs in both overlay and tiled comparison views. Minor misalignment remained due to nonrigid deformation caused by respiration or posture shifts. This approach may enhance procedural stability and safety to complement operator skill. We report initial validation results and discuss future developments.