Post-Treatment Target Motion Analysis Using Offset-Dependent Three-Dimensional Reconstruction of on-Treatment Kv Images
Abstract
Purpose
This study aims to quantitatively evaluate intrafraction target motion through post-treatment analysis based on offset-dependent three-dimensional reconstruction of on-treatment kV images.
Methods
Forty on-treatment kV projection images of a prostate phantom were acquired using a Varian TrueBeam system, with intrafraction motion simulated in sixteen projections using two-dimensional and three-dimensional translational shifts (10 mm in each coordinate). Target motion was estimated using an iterative offset-dependent three-dimensional reconstruction framework in which the reconstruction origin was optimized. Three-dimensional volumes were reconstructed with systematic offset updates of 1 mm and subsequently 0.1 mm along the x, y, and z axes, and RMSE and MS-SSIM were used as objective functions to localize the target displacement in a post-treatment analysis.
Results
The proposed metric-driven, offset-dependent three-dimensional motion evaluation framework successfully localized intrafraction target motion in a post-treatment setting. For two-dimensional shifts, no-shift and shifted projections were clearly separated, with no-shift projections yielding higher MS-SSIM values on the order of 0.3–0.35 and lower RMSE values on the order of . In the three-dimensional shift scenario, no-shift projections exhibited the highest MS-SSIM, whereas projections affected by out-of-plane motion showed reduced MS-SSIM (approximately 0.15–0.25) and increased RMSE (approximately ). In both metrics, motion-related responses were well separated from background levels, enabling reliable discrimination between no-shift and intrafraction motion states.
Conclusion
This study demonstrates that metric-driven, offset-dependent three-dimensional reconstruction of on-treatment kV images enables reliable post-treatment estimation of intrafraction target motion, including out-of-plane displacement, without requiring additional imaging hardware.