肝臟腫瘤是華人常見的疾病，早期診斷與治療是避免肝腫瘤惡化很重要的措施。近年來快速發展的高強度聚焦超音波(HIFU)以高溫方式破壞腫瘤細胞，此一對病患傷害最小的無創熱治療方式極具發展潛力。因為病人的呼吸會造成肝臟腫瘤位移，若從人體呼吸的生理反應，例如胸口起伏、呼吸氣體流量，推算並預測肝腫瘤位置，將可以達到與呼吸同步治療，提昇治療準確度。本研究目標是基於實驗室現有的超音波影像輔助機器手臂定位HIFU技術，發展一具備呼吸補償功能之超音波影像輔助機械手臂HIFU腫瘤燒灼追蹤系統。 研究依據三位實驗者肝臟血管位移與胸口起伏量的實驗值，設計了一組機構以模擬人體在呼吸狀態下肝腫瘤位移與胸口起伏關係，並在不同的胸口起伏狀態下，以超音波掃描腫瘤假體、重建三維腫瘤模型並計算其空間位置，之後將各腫瘤空間位置用多項式函數擬合以模擬腫瘤位移軌跡，並建立模擬呼吸起伏的雷射量測値與腫瘤位移軌跡的對應關係。在完成機械手臂、光學式定位器及三維腫瘤模型對位後，機械手臂追蹤系統即依據雷射量測值與自動控制HIFU聚焦點依循腫瘤位移軌跡移動，過程中並就運動延遲時間進行補償，以達到與呼吸同步的精確腫瘤燒灼治療。 腫瘤假體燒灼實驗分成靜態燒灼與動態燒灼:靜態單點燒灼誤差約為1mm，三維模型切面多點燒灼結果顯示可均勻地完成整個假體切面；動態單點追蹤燒灼實驗結果則顯示延遲時間補償後，追蹤誤差與腫瘤位移軌跡曲線擬合誤差分別為0.65±0.31mm和1.72±1.26mm。切面燒灼實驗追蹤誤差與腫瘤位移軌跡曲線擬合誤差分別為0.85±0.39mm和3.04±1.24mm。其中軌跡曲線擬合誤差大，主因是腫瘤取樣位置數目偏低所造成的。 ;Liver tumor is a common disease in Chinese society. Early diagnosis and treatment of liver tumor are highly recommended to avoid worsening. High Intensity Focused Ultrasound (HIFU) ablation is being increasingly developed and high potential as a non-invasive treatment option for liver cancer. Displacement of liver tumor occurs due to the patient’s respiratory, which can be predicted based on a respiratory external signal, e.g., chest displacement or air flow of spirometer. Therefore, it is potential to treat liver tumor synchronously. This project is aimed to develop an ultrasound assisted robotic HIU ablation system with respiratory movement compensation. According to the synchronous measurements of displacement of hepatic vessel and chest displacement of three human subjects, an experimental mechanism is designed to simulate the relation between liver tumor movement and chest displacement due to respiratory. Ultrasound scan of the tumor phantom is done in six different respiratory statuses. The sequential sonograms are used to reconstruct the three-dimensional tumor model and compute its location. Then, the trajectory of tumor movement is synthesized by using cubic polynomials and the correlation between the laser measurement value and tumor movement is determined. After the registration among the robotic manipulator, optic tracker and tumor model has been completed, the robot manipulator can automatically control the HIFU focus point to track the trajectory of tumor targets and to ablate the tumor targets synchronizing with respiration. The experiments are separated into static and dynamic ablations of tumor phantom. The results of static ablation experiments show that the average distance error in single-point ablation was about 1mm and the cross-cut plane of tumor phantom can be ablated completely. As to dynamic ablation experiments with time-delay compensation, the results show that the average distance errors of single-point tracking and trajectory curve fitting are 0.65±0.31mm and 1.72±1.26mm respectively. The average distance errors of cross-cut plane tracking and trajectory curve fitting are 0.85±0.39mm and 3.04±1.24mm respectively. The large distance error of trajectory curve fitting is because only six scanning data of tumor phantom is used to synthesize the trajectory
