(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: To present a patient-specific correspondence model to track lung tumors during radiation therapy using surface displacement as a surrogate signal.
Methods: A CIRS Xsight® Lung Tracking Phantom kit was used to model the respiratory motion. A 2.5 cm diameter spherical target moves linearly and rotationally in the phantom. The iso-center was placed at the center of the sphere. A large skin-like silicone pad was placed on the phantom to cover the surrogate platform and connect it to the phantom. While a 4DCT scan was acquired from the phantom, displacement of two points on the silicone pad were recorded as surrogate signals. Using the C-RAD Sentinel system, the primary signal point was placed on the surrogate panel and the secondary point was placed on the part connecting the platform to the phantom. The primary signal was used to reconstruct 4DCT images into 10 respiratory phases. The data of internal motion was calculated by performing deformable image registration (DIR) algorithm on the 4DCT images with end-exhalation (EE) phase as the reference. Principal component analysis (PCA) was used to fit the correspondence model to the internal data and surrogate signals. In the treatment room the primary and secondary signals were recorded by C-RAD CatalystHD and placed into the model as an input. The model’s output was the motion of the iso-center along X, Y and Z direction.
Results: There was submillimeter agreement between the actual and calculated target motions, with differences of 0.08 mm along X (RL), 0.51 mm along Y (AP) and 0.33 mm along Z (SI). The direction of the motion matches the true direction of the target motion along all three axes.
Conclusion: Application of the correspondence model designed in this study is a non-invasive and dose free technique to track the location of the tumor during radiation therapy.
Keywords
Target Localization, Respiration, Modeling
Taxonomy
IM/TH- Image Analysis (Single Modality or Multi-Modality): Image registration
Contact Email
