Quantification of Radiotherapy CT image for Non-Small Cell Lung Cancer: Development of Longitudinal Registration Algorithm and Correlation Analysis of Lung Parenchyma Change and Planning Radiation Dose Distribution

肺癌是世界上最主要的癌症死亡原因之一，在台灣也同樣不分性別地高居所有癌症死亡原因第一位，其中在男性死亡率更高達48.9％。早期肺癌最佳的治療方式為外科手術切除，然而適合手術治療之早期肺癌病患僅佔總體約15%，大多數之肺癌被發現時已進入中晚期，不再適合僅以外科手術方式治療，因此，化學治療、放射線治療及化學治療合併放射線治療在針對中晚期肺癌治療扮演相當重要的角色。在病患接受放射線治療過程中，高能放射線除了對腫瘤進行治療，同時也會傷害鄰近的肺部正常組織導致併發症，包含急性期產生放射性肺炎及晚期發生肺纖維化等症狀，造成病人肺功能下降且影響病人治療後之生活品質。 本研究提出一用於評估放射治療後肺實質變化之多時間點對位演算法，並且針對肺實質變化與劑量強度以及時間之關聯性進行探討。在所提出之對位演算法中，透過肺區以外鄰近肺臟的解剖結構，克服過往大範圍之肺實質改變導致對位困難之問題。於所提之演算法中以脊椎作為剛性對位的參考對象；以鄰近肺區的所有骨骼組織表面，包含胸骨柄、肋骨、脊椎等，以及氣管壁與肺部下緣表面，共三個肺區周圍主要的肺區周圍解剖特徵作為非剛性對位中描述不同時間點時呼吸飽滿度所造成的肺部擴張差異之依據，並對此三個解剖特徵進行特徵點群參考點採樣後進行連貫性群點對位。 使用本研究發展之多時間點對位演算法對治療前與治療後之肺臟進行對位後，探討肺實質變化與劑量強度以及時間之關聯性，得到了在22 Gy 以上的劑量區間在3~7個月時便有顯著的肺實質變化的結論。Lung cancer is one of the leading cause of cancer deaths worldwide, including Taiwan, across gender. In Taiwan, the male mortality rate of lung cancer is as high as 48.9%. Surgery is the best and effective method in the early stage of the lung cancer. However, only 15% of the diagnosed patients are suitable for early-stage surgery. Surgery only is no longer considered for those being diagnosed at the middle to the late stage. Treatment for patients with middle-stage and late-stage may involve chemotherapy, radiotherapy, or a concurrent chemoradiation therapy. Radiation therapy, which is one of the primary therapeutic approaches for non-small cell lung cancer, is a treatment that uses high-energy rays or particles that destroy lung cancer cells. Radiation-induced lung damage (RILD) is a severe complication of radiotherapy in lung cancer patients that presents as a progressive pulmonary injury affecting prognosis and quality of life in patients. In this study a longitudinal registration algorithm is proposed for evaluating the lung parenchyma change after radiotherapy and the correlation to the given radiation strength and distribution of dosage. The proposed registration algorithm overcomes the large parenchyma change which makes the registration much harder by using anatomical structures around the lung, including using spine for the reference set of rigid registration step; using three anatomical structures: bone structures surface, including sternal, rib and spine, airway wall and surface of lower lung to describe the longitudinal difference of breath holding degree. Reference points are sampled from these three anatomical feature structures for the further step of coherent point set registration. Registered by the proposed longitudinal registration algorithm developed by this study, the correlation of regional dose distribution with longitudinal parenchyma change has been evaluated and obvious parenchyma change in the region of radiation dosage above 22 Gy and in 3~7 month is observed