Supplementary Material for: Clinical Implementation of In-House Developed MR-Based Patient-Specific 3D Models of Liver Anatomy

Knowledge of patient-specific liver anatomy is key to patient safety during major hepatobiliary surgery. Three-dimensional (3D) models of patient-specific liver anatomy based on diagnostic MRI images can provide essential vascular and biliary anatomical insight during surgery. However, a method for generating these is not yet publicly available. This paper describes how these 3D models of the liver can be generated using open source software, and then subsequently integrated into a sterile surgical environment. The most common image quality aspects that degrade the quality of the 3D models as well possible ways of eliminating these are also discussed. Per patient, a single diagnostic multiphase MRI scan with hepatospecific contrast agent was used for automated segmentation of liver contour, arterial, portal, and venous anatomy, and the biliary tree. Subsequently, lesions were delineated manually. The resulting interactive 3D model could be accessed during surgery on a sterile covered tablet. Up to now, such models have been used in 335 surgical procedures. Their use simplified the surgical treatment of patients with a high number of liver metastases and contributed to the localization of vanished lesions in cases of a radiological complete response to neoadjuvant treatment. They facilitated perioperative verification of the relationship of tumors and the surrounding vascular and biliary anatomy, and eased decision-making before and during surgery.

在大型肝胆外科手术中，掌握患者个体化的肝脏解剖知识是保障患者安全的核心要素。基于诊断性磁共振成像（MRI）图像构建的患者个体化肝脏解剖三维（3D）模型，可在手术中提供关键的血管与胆道解剖细节，但目前尚无公开可用的此类模型生成方法。本文详述了如何利用开源软件生成此类肝脏3D模型，并将其集成至无菌手术环境中的具体流程，同时探讨了导致3D模型质量下降的常见图像质量问题及对应的解决思路。针对每位患者，我们采用单次注射肝特异性造影剂的诊断性多相MRI扫描，实现肝脏轮廓、动脉、门静脉、静脉系统以及胆道树的自动化分割，随后由人工手动勾画病灶区域。最终生成的交互式3D模型可在手术中通过无菌防护平板设备进行访问操作。截至目前，此类模型已应用于335例外科手术中，其应用简化了多发肝转移瘤患者的外科治疗流程，可在新辅助治疗达到影像学完全缓解的病例中辅助定位已消失的病灶，有助于围手术期验证肿瘤与周围血管、胆道解剖结构的毗邻关系，并为术前及术中的决策制定提供了便利。