以運動軌跡預測及追蹤控制方法進行肝腫瘤高能超音波熱治療

Abstract

高能聚焦超音波是一種非侵入式治療腫瘤的新方法，主要是利用超音波在腫瘤部位聚焦提高溫度燒灼惡性細胞以達到治療的目的。其中我們可經由電子式的聚焦藉由調整相位陣列換能器的相位與振幅來改變聚焦點的位置，且由於聚焦超音波所形成的焦點相當小，因此可以精確地加熱於腫瘤部位。 當利用聚焦超音波做治療時，我們必須準確地將焦點聚焦在所要治療的患部以避免對正常的組織造成傷害。由於人體內一些內臟器官，如肝臟、肺臟，它們會因呼吸而受橫膈膜的擠壓在一個三度空間中運動，此一運動會造成聚焦超音波無法準確持續聚焦於患部上而造成治療效率的降低，更嚴重甚至會傷害到正常的組織造成不可預期的後果。 在本論文中，我們利用伺服馬達帶動目標物作非規則往復運動模擬內臟因呼吸所形成的運動軌跡，再以CCD攝影機模仿MRI抓取影像進而得知目標物的所在位置，為了克服在影像處理系統方面處理影像所造成的延遲，我們採用類神經網路來預測目標物的下一個和下下一個取樣時間點位置。追蹤方面以雷射筆的光點替代聚焦點追蹤目標物，加熱方面利用內插重建取樣點之間的軌跡，達到持續追蹤加熱的目標。如此預測演算法可以克服系統延遲的問題，內插重建軌跡能減少聚焦點與目標物之間距離的誤差，使得超音波能更精確的聚焦在患部上。

High intensity focus ultrasound is a non-invasive method to treat the tumor. A phased array transducer has the ability to change focus position or produce multi-focus by adjusting the amplitude and phase of each element on the surface of transducer. The characteristic of phased array can be utilized to kill the tumor that has reciprocation due to breathing within human body. Image guidance in hyperthermia offers the potential for deliver thermal dosage to a moving tumor precisely. Because MRI image provides not only the temperature varying but also other information, combining MRI with HIFU for treating tumor is the main purpose. In this study, CCD image is used to imitate MRI image. A platform is controlled by servo-motor to simulate the organ motion due to breathing. However, tracking a target via image processing takes a small period of time including of image acquisition, image processing and delay of communications. To overcome the whole system time delay, it is necessary to develop a predictive algorithm. Neural network is used to predict the future trend of tumor to overcome the system time delay. And interpolation is used to reconstruct trajectory between sampling points to let the heating can be continuous.