基於有限元素法模型模擬肝臟形變並應用於血管定位

Abstract

在微創手術中，知道在肝臟切割手術中肝臟內部血管及腫瘤位置是困難的，醫師只能憑藉術前的電腦斷層掃瞄影像當作導引大約知道肝臟中血管及腫瘤分佈位置，但肝臟會因為不同的手術操作而產生形變，使得術前的資料產生變動。藉由血管位置計算的幫助，醫師可以更直覺的操作手術且可以降低可能切割到大的肝臟血管(>3mm)或腫瘤的風險。 在本論文中，使用的是從電腦斷層掃瞄影像所建立的有限元素法模型去模擬形變，藉由分析已形變(手術中)及未形變(手術前)肝臟表面差異得到位移邊界條件並用來模擬肝臟形變，在有限元素模擬之後，可以得到形變後(手術中)的血管及腫瘤分布位置。我們利用模擬資料以及體外豬肝實驗去驗證演算法，在模擬實驗中，平均肝臟表面的誤差在0.35mm，平均內部的誤差在0.28mm，在體外豬肝實驗中，平均肝臟表面誤差在0.86mm，平均內部誤差在2.26mm。

In minimally invasive surgery, knowing the position of internal structures of tissue such as vessels and tumors is a challenge. Surgeons only depend on preoperative CT images to know positions of vessels and tumors, but liver is deformed by different surgical procedures. The preoperative information is no longer precise. With the help of vessel-tracking system, surgeons can operate the surgical tasks more intuitively and decrease the risk from cutting large vessels and tumors. In this thesis, finite element model built from preoperative CT is used to calculate deformations. The difference between intraoperative and preoperative surface data is used as displacement boundary conditions applied to simulate the deformations. After finite element computation, the intraoperative positions of vessels and tumors are obtained. The proposed algorithms are first validated through simulation data based on surgical scenario in minimally invasive surgery, then validated through ex vivo porcine liver experiment data. The mean error of liver surface is 0.35mm and mean error of internal structure is 0.28mm in simulation. The mean error of liver surface is 0.86mm and mean error of internal structures is 2.26mm in ex vivo porcine liver experiment.