Abstract
放射治療手術中，雖然降低放射劑量會降低發生在病患身上的副作用，相 反的，降低放射計量卻會減低腫瘤的控制率，而增加腫瘤復發的可能性。 因此正確的腫瘤位置，在放射治療中就顯得非常重要了，因為有了正確的 腫瘤位置區域和範圍，我們就可以控制使腫瘤增加劑量，同時減低正常組 織的曝露劑量。我們發展出一種能從斷層掃描影像資料中，藉由電腦輔助 ，來決定於模擬造影片上，腫瘤區域位置的演算法。此演算法改進了一般 傳統的最小均方差運算，而將數據經過排名和量化的處理，即使用量化排 名環最小均方差運算子，來搜尋出斷層掃描投影影像與模擬照影片之間最 佳的對位中心與最佳的旋轉角度。然後，將先前由放射治療醫師在斷層掃 描圖中圈出的腫瘤區域以點光源角錐光投影在透明片與模擬照影片重疊在 一起，如此在模擬照影片上的腫瘤位置也就被正確的標示出來了。此電腦 輔助技術提供一個較高效率，副作用傷害較小，且較精確的放射治療技術 。結構安排如下，第一章，簡介，描述整個問題並介紹斷層掃描和模擬照 影片。第二章，介紹量化排名環最小均方差演算法以適應存在兩種影像上 的非線性轉換關係。第三章描述對位演算法的操作步驟與方法，並使用第 二章定義的運算來做影像對位的工作。第四章展示實驗結果與資料。第五 章結果與討論。 In radiotheraphy, decreasing the total treatment dosage will reduce the side effect on a patient. On the contrary, decreasing the total treatment dosage lowers the tumor control probability of the patients. Accurate tumor localization for radiotherapy treatment appears to be very important because it allows to increase the dosage on the tumor and reduce simultaneously the dosage on the normal tissues. We develop computer techniques to determine the relative positions of the patient under radiotherapy treatment and under computed tomography scan and to find the tumor location and area on simulation film from the computed tomography(CT). The simulation film is an X-ray film of the simulator for determining radiation area before real treatment on the linear accelerator. This approach uses a quantized rank circular minimum mean square error(QRCMMSE) operator to find the optimal matching center between CT projection image and simulation film. The location and area of tumor are consequently determined by projecting the CT slices with the tumor contours, marked by the oncologist previously, onto a transparency to be overlaid with the simulation film. This computer techniques can also revise the rotation error if existing. Experimental results show that the algorithm are quite accurate in performing the desired task. This technique could be need to give more effective, less harmful and more efficient radiation therapy treatment.