SU-E-J-223: A Method to Remove the Streaking Artifacts Caused by Calypso Couch Table Add-On in the MV Portal Images

Abstract

Purpose: Calypso EM tracking is widely used to setup the prostate patient and to track prostate motion while daily MV portal images are used as the secondary method to verify patient setup. The severe streaking artifacts, caused by the Calypso couch table add-on, make the MV portal images almost impossible to review. The purpose of this work is to develop a novel method to reduce the streaking artifacts so that the portal images are usable to verify patient setup. Methods: We analyzed the corrupted portal images and identified that most streaking artifacts are stable horizontally but has more dramatic change vertically. Based on signal processing theory, we developed an artifact reduction algorithm 1) to filter out the low frequency signal along the horizontal direction using a ideal low-pass filter in the Fourier domain, and 2) to smooth the intensity variation in the vertical direction and to reduce the remaining high frequency noises by using 2D Gaussian low-pass filter with a small window size. Results: We have successfully applied the proposed algorithm on the corrupted MV portal images in both PA and lateral directions from multiple Calypso prostate patients. The results show that the algorithm is very effective to reduce the artifacts introduced by the Calypso couch table add-on. Patient anatomical structures are much more clearly after processing to support reviews of patient setup. Conclusion: In this work an effective method is proposed to reduce artifacts caused by the Calypso table in MV portal images while preserving the key components of original image. The method could assist the verification of Calypso guided patient setup using portal images.