An Effective Strategy for Set-up Error Estimation and Reduction in Image-Guided Prostate Radiotherapy

放射劑量提高可提昇攝護腺癌的控制率，但也造成治療副作用增加的危險。安全提高放射劑量首要為具體量化治療誤差並增加治療之精準度。因此本研究主要目的之一為探討攝護腺癌病患接受體外放射治療療程中發生的位移誤差，利用「電子影像驗證設備」及「錐狀射束電腦斷層」等影像導引工具，具體量測各個軸向之位移。根據所量測之結果，設計並優化校正流程以減少誤差及增加治療精準度。 本研究總共收錄85位接受強度調控放射治療之攝護腺癌病患，實驗的第一階段是利用「電子影像驗證設備」進行取像，求得三個不同軸向的位移誤差。第二階段則是利用另一種影像工具：「錐狀射束電腦斷層」取像，其影像與原先定位的電腦斷層影像作比對，量測三個不同軸向的位移誤差，並觀察療程中不同時間點的變化趨勢。最後一個階段，我們設計一校正流程來減少位移誤差，並進行改良優化。此階段依不同時期可分為三個不同校正方式：(1) 流程-A：每週一次（第2-8週的起始日）的重新對位校正，整個療程總共校正次數為7次。(2) 流程-B：在第2、3週的起始日進行重新對位校正，整個療程總共校正次數為2次。(3) 流程-C：在第2、3、6週的起始日進行重新對位校正，整個療程總共校正次數為3次。每次校正的基礎是以校正前一週的平均誤差為基準進行重新對位，估評三種不同校正流程的成效及工作負荷量。 第一階段利用電子影像驗證設備所量測到的誤差在頭腳、前後、左右方向分為別1.9、3.4、及2.5mm，直腸內充氣裝置在三個方向（頭腳、前後、左右）的平均變形程度分別為2.8、2.5、及2.6mm。第二階段則是利用錐狀射束電腦斷層可觀察到三度空間總位移誤差在第1週為最小（3.2mm），之後便逐漸增加到7.3mm（第8週），且有統計學上之差異。實驗的最後階段，流程-A、B、C在三個軸向（頭腳、前後、左右）的平均位移誤差分別為0.4/ 0.4/ 0.3、0.9/ 0.4/ 0.7、及0.2/ 0.3/ 0.3 mm。與校正流程-B相比，流程-A及C可得到較顯著的誤差校正效果，且此變化有統計學上之差異。另外在工作負荷量方面，三組（A、B、C）的在療程中的進行校正所需的取像次數分別為35、10、及15次。 利用錐狀射束電腦斷層進行影像導引放射治療能有效量測及減少位移誤差，最剛開始使用的流程-A方式雖然可得到最佳的誤差校正效果，但卻大幅增加臨床工作量，經由逐步修正後的流程-C校正方式，可得到與流程-A類似的成效且大幅減少工作負荷量。Radiotherapy dose escalation may achieve higher biochemical control rates for prostate cancer. However, a higher radiation dose is associated with a higher risk of radiation toxicity. Improving the accuracy of dose delivery is an important issue in order to safely escalate the radiation dose. Therefore, we conducted an investigation to estimate the set-up errors in prostate radiotherapy using different image-guidance tools. Moreover, we designed a correction protocol and modified it step-by-step to effectively reduce the set-up error and improve treatment accuracy. Eighty-five patients with prostate cancer undergoing intensity-modulated radiation therapy (IMRT) were enrolled. In the first phase of the study, electronic portal images (EPIs) were collected weekly to estimate displacements in 3 orthogonal directions. In the second phase, cone-beam computed tomography (CBCT) images were collected daily and compared with the pretreatment simulation CT to estimate the set-up errors and time-dependent changes over the course of treatment. Patients were then divided into 3 groups: (1) protocol-A: weekly corrections (a total of 7 corrections) of body re-alignment throughout the entire course; (2) protocol-B: weekly corrections in the second and third weeks of treatment (a total of 2 corrections); and (3) protocol-C: weekly corrections in the second, third, and sixth weeks of treatment. The serial set-up errors were compared to evaluate the effectiveness and workload of the 3 correction strategies. With the use of an electronic portal image device (EPID), the mean displacements of the isocenter relative to the bony landmark were determined as 1.9 mm in the superior-inferior (SI), 3.4 mm in the anterior-posterior (AP), and 2.5 mm in the left-right (LR) directions. Balloon deformations of 2.8 mm, 2.5 mm, and 2.6 mm occurred in the SI, AP, and LR directions, respectively. The use of CBCT to estimate the set-up error revealed that the average three-dimensional (3D) displacement increased from 3.2 (first week) to 7.3 mm (eighth week, p = 0.027). Using different correction strategies, the mean displacements for protocols A, B, and C were 0.4/0.4/0.3, 0.9/0.4/0.7, and 0.2/0.3/0.3 mm in the SI, AP, and LR directions, respectively. The effectiveness of reducing set-up errors (3D mean displacements) was more remarkable in groups A (3.4 mm) and C (4.0 mm) compared to group B (5.2 mm, p < 0.05). The numbers of CBCT images required for each correction were 35, 10, and 15 for groups A, B, and C, respectively. Radiotherapy with CBCT was effective for estimating and reducing the set-up errors. Although protocol-A had the least set-up errors, the greater amount of CBCT images required for correction resulted in a notably increased workload. Protocol-C achieved similar effectiveness with an acceptable workload