基於XR導引之肺結節定位

Abstract

肺結節定位是胸腔鏡次肺葉切除手術前的關鍵步驟，傳統上多依賴電腦斷層影像(CT)導引定位。本研究提出一套基於延展實境(XR)的肺結節定位系統。術前在病人身上貼附可在CT中顯影的基準標記並標示位置，據術前CT重建包含基準標記與器官之數位孿生模型。術中依照病人身上標示之特殊圖樣，將平面標記復位後，偵測並估計平面標記在空間中的位置，基準標記與平面標記兩者經配準後達成器官模型與體表的對齊。術者可於系統中規劃染劑定位針的穿刺路徑，並在系統導引下完成肺結節染劑定位。本研究於複合式手術室中進行驗證，共納入15名病患，定位完成後以術中CT評估定位誤差，平均定位誤差為7.9 mm（標準差 5.1 mm）。

Lung nodule localization is a critical step prior to sublobar resection in video-assisted thoracoscopic surgery. Traditionally, localization relies on CT-guided techniques. This study proposes an extended reality (XR)-based localization system. Prior to surgery, radiopaque fiducial markers are attached to the patient's body, and their positions are marked. Digital twin models, including the fiducial markers and relevant organs, are reconstructed from preoperative CT images. Intraoperatively, planar markers are repositioned according to the marked skin patterns. The spatial pose of the planar markers is then detected and estimated. Registration between the fiducial and planar markers aligns the organ models with the body. The system allows the surgeon to plan the puncture path for dye localization, which is then executed under XR guidance. The system was validated in a hybrid operating room in 15 patients. Localization error was evaluated using intraoperative CT. The mean localization error was 7.9 mm (SD: 5.1 mm).