目的：由於治療技術的進步，可加強腫瘤和正常組織間劑量的差異，提昇治療療效，因此治療計畫對於治療不確定性的敏感度也隨之上升。在各種影響治療不確定性的因素中，目前以器官移動評估的準確性特別受到重視。 材料與方法：本研究是利用模擬攝影中，藉由吞食鋇劑後之透視影像，利用影像前置處理、邊緣偵測、和邊緣定位來自動搜尋食道影像邊緣。 結果：經由本研究所提出自動食道邊緣分割演算法，對於因受腫瘤壓迫而形成的鋇劑聚積處，可描繪出完整的邊緣，並達到以食道邊緣之移動代表腫瘤在治療中的移動。 結論：移動評估的結果，有利於治療計畫中安全邊距的縮小，在腫瘤治療劑量增加時仍能有效控制正常組織所接受之劑量，降低正常組織副作用發生機率及其發生的嚴重程度。

Propose : As the prescribed dose difference between tumor cells and normal tissues increases, treatment planning becomes more sensitive to treatment uncertainties. Among various treatment uncertainties, accurate evaluation of organ motion plays an important role in quality improvement of treatment planning. Materials and Methods : In this study, we describe an automatic segmentation algorithm to extract contrast medium enhanced esophagus in X-ray fluoroscopy. The algorithm is based on image pre-processing, image edge detection, and boundary extraction to identify the esophageal boundary in a fully automatic manner. Results and Conclusion : The extracted esophageal boundary variations between adjacent time frame sequences can be applied to estimate the tumor motions as well as the required safe margins.