Room: AAPM ePoster Library
Purpose: The respiratory phase shift between lung tumor and external surrogate marker is important for accurate motion-tracking radiotherapy. We developed a method to evaluate the tumor motions and the phase shifts quickly.
Methods: We analyzed the four-dimensional (4D) CT images of 31 patients who received stereotactic body radiotherapy to lung cancer. The images were acquired using an Aquilion ONE, a 320-slice area detector CT. The CT acquires dynamic 4D-CT within the 16 cm scan length without external respiration signals. The images were acquired every 0.5 s, and the acquisition time was 10 s for most cases. Coronal and sagittal planes passing through the center of the tumor were extracted from all volumes to generate the cine images. The motion vectors between continuous two images were analyzed using pyramidal Lucas-Kanade method, an optical flow estimation algorithm. A tumor contour was delineated on one frame, and the contour was shifted and copied onto other frames using the motion vectors inside the contour. The lateral tumor motion was analyzed on the coronal cine images, and the sagittal cine images were generated again to analyze the three-dimensional tumor motion. On the sagittal cine images generated on the mid-plane, we also analyzed the respiratory motion of the sternal bone representing external marker motions. The respiratory phase at the end-inspiration was defined as 50%, and the phases at other frames were calculated by interpolation.
Results: At the end-inspiration phase of the sternal bone, the mean ±SD (range) of the respiratory phase shifts of the tumors were -10.6% ±15.3% (-46.7% - 13.1%) and 2.3% ±15.3% (-50.0% - 21.4%) in anteroposterior and superior-inferior directions, respectively.
Conclusion: The method developed in this study calculated the 3D tumor motion and the respiratory phase shifts in approximately 2 minutes. Large variations were observed in the phase shifts especially in AP direction.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by JSPS KAKENHI Grant number JP17K15802.