SU‐C‐BRA‐01: 4D Cone‐Beam CT Acquisition Using Respiratory Phase Predication Technique

Abstract

Purpose: Respiration‐correlated CBCT reduces respiratory motion artifacts at the cost of increased view‐aliasing artifacts due to the gaps in projection data that are sorted to other phases. Respiratory‐gated acquisitions can cover the full scanning range but is inefficient (∼5 minutes per phase). In this study, we developed a 4D CBCT acquisition method using respiratory phase predication technique that can potentially acquire full projection data for each phase within a clinically acceptable (∼10 minutes or less) time. Methods: The proposed 4D CBCT continuously scans the target until sufficient number of projections is acquired for all phases. The respiratory phase predication technique consists of the following iterative steps: (1) training using respiratory signal for the first few (e.g., five) breathing cycles; (2) initial optimization of the gantry rotation speed; (3) adaptive control of the rotation speed and image acquisition during the scan (4) sorting the projection images into different phases. The method was validated using the 4D XCAT digital lung phantom with the motion driven by a sinusoid wave and a real patient respiration signal acquired with RPM system. CBCT images were reconstructed for each phase using the simultaneous arithmetic reconstruction technique and compared with the mixed‐phase reconstruction. Results: Sufficient projection images (an angular interval less than 1 degree) were acquired for all 10 phases after 10 gantry rotations. The acquisition time was estimated from 10–12 minutes assuming each gantry rotation took ∼1 minute. The CBCT images reconstructed from the mixed‐phase projections were blurred in comparison to those for individual phases. Conclusion: A respiratory prediction technique for 4D CBCT has been developed and validated with real patient breathing pattern. We plan to further reduce the number of gantry rotations and therefore the scanning time by improving the control of adaptive image acquisition and taking advantage of data redundancy as in 4D CT.