Room: Exhibit Hall | Forum 1
Purpose: 4D moving blocker (â€œ4D-MBâ€?) cone-beam computed tomography (CBCT) has been proposed to reconstruct 4D images for tumor motion modeling without considering scatter effect previously. In this work, we investigate the performance of 4D-MB for simultaneous dose reduction and scatter correction (SC).
Methods: A moving blocker is placed between the X-ray tube and the subject for partial X-ray fluence blocking at each projection view, where the blocked region provides the information for SC. There is no strict requirement on the blocker moving speed or style as long as the blocker position is evenly distributed for each respiratory phase. In 4D-MB, each respiratory phase suffers both fewer views and partial projection data problems. To address this severe missing data problem, we adopt the simultaneous motion estimation and image reconstruction (SMEIR) method to jointly reconstruct each phase image using all projection data. The scatter-contaminated projection data of 4D-MB with a ratio of 1/3 of the blocked region to the unblocked region (i.e. ~1/3 dose reduction) was simulated using the NCAT phantom and a scatter kernel.
Results: Four reconstruction methods are compared: 1) 3D phase-by-phase total variation (TV) reconstruction without SC; 2) 4D SMEIR without SC; 3) 3D TV with SC; and 4) 4D SMEIR with SC (â€œMB-SMEIRâ€?). MB-SMEIR achieved the best structural similarity (SSIM) index and root mean square error (RMSE) values of 0.86 and 59 HU among all four methods, representing an improvement of 28% on SSIM and 34% on RMSE over the second-ranked 3D TV with SC. Regarding tumor motion recovery, MB-SMEIR also outperforms other methods with a maximum deviation of 3.01 mm and a mean deviation of 1.48 mm.
Conclusion: The proposed 4D-MB CBCT method can achieve both dose reduction and scatter correction simultaneously. With advanced motion-compensated reconstruction, it can produce high quality images and faithful tumor motion tracking.