Room: Exhibit Hall | Forum 9
Purpose: To develop a novel algorithm for real-time volumetric image generation based on CT image deformation driven by displacement of the internal fiducial markers for motion management in radiation therapy.
Methods: The proposed method utilizes the internal fiducial markers to derive the deformation vector fields (DVFs) during treatment. Since the respiratory motion could be varied and include baseline shift/drift, it is preferred to use the internal landmarks which correlate well with target and surrounding organs. Before the treatment, as the modeling data, the DVFs and the displacement of the fiducial markers are evaluated for each CT data of 4DCT relative to the exhale phase CT as reference volume. Then, the function to derive the DVF from the marker displacement is optimized. During the treatment, the volumetric image can be generated in real-time by applying the DVF estimated from the marker displacement measured by real-time imaging technique to the reference CT image. In this study, we evaluate the reconstruction performance by using 4DCT data including the fiducial markers in right lower lobe. Linear function between the marker displacement and the internal deformation amplitude at each voxel was assumed and optimized to reconstruct the synthetic images.
Results: The synthetic images generated with the actual marker displacement data including large amplitude relative to the modeling data showed reasonable consistency in positional variation of bifurcations and fiducial markers. Absolute difference between the actual CT image of inhale respiratory phase and the synthetic CT image generated by corresponding marker displacement was less than 70 HU in 95% of the voxels in right lower lobe. Computation time to generate the image with volume of 150 mmÂ³ (150 pixels by 150 pixels by 60 slices) was about 70 ms.
Conclusion: We showed the feasibility of the real-time volumetric image generation algorithm using the internal fiducial markers.