WE-G-134-01: A Novel Method to Correct Scatter and Metal Artifacts in Kilovoltage CBCT Using Megavoltage Projections

Purpose: To improve quantitative accuracy of CT numbers in kV‐CBCT, we developed a method to correct kV projections using MV images acquired during CBCT acquisition. This method exploits inherent characteristics of MV image signal, such as relatively lower scatter fraction and lower sensitivity to elemental composition, to both correct for scatter and suppress metal artifacts in kV‐CBCT. Methods: To correct for scatter in kV‐CBCT, two water‐based mapping functions were developed to correct beam hardening in MV projections, and to transform them to kV‐equivalent MV projections in Radon space. At a given gantry angle, the difference in log‐attenuation values between kV‐equivalent MV and raw kV projections served as the estimate of scatter distribution. Smoothed scatter distributions were subtracted from the raw kV projections to obtain corrected kV projections. In addition, the difference between kV‐equivalent MV and scatter‐corrected kV projections were used to identify image pixels that exhibited nonlinearities due to metal objects in raw KV projections. Such pixels were replaced by the corresponding pixel values in the kV‐equivalent MV projections. Standard filter‐backprojection algorithm was utilized to reconstruct the CBCT image from corrected KV projections. Results: Varian TrueBeam linac operated in research mode was utilized to acquire kV‐MV projections. In a torso phantom image, shading artifacts due to scatter was reduced from 300 HU to less than 100 HU after corrections. Two steel rods placed on the surface of the phantom lead to more than 600 HU image streaks in the uncorrected kV‐CBCT image. After metal artifact corrections, variation in CT numbers was reduced below 100 HU. Conclusion: Scatter and metal artifacts can be significantly reduced in kV‐CBCT using our approach. Improved quantitative accuracy of CT numbers may enable CBCT image based dose calculations in radiation therapy, and be particularly useful to reduce range uncertainties in particle therapy for patients with metallic implants.