2231 - Optimizing Gantry Rotation Time for Deviceless 4DCT-Derived Internal Target Volumes (ITV)

Purpose/Objective(s):

The step-and-shoot technique employed in the Deviceless 4DCT system minimizes interpolation and couch motion artifacts but introduces a pronounced stairstep effect that resembles motion-related artifacts. This study compared the accuracy of Deviceless 4DCT-derived ITV with that calculated from a mathematical motion model for various CT gantry rotation times. The goal was to identify the optimal gantry rotation time for thoracoabdominal imaging and assess the impact of the stairstep effect on ITV delineation accuracy.

Materials/Methods:

A custom 4D motion phantom with a 3-cm spherical target was constructed on a battery-powered nonmetallic platform (1.2 cm vertical amplitude, 6.8 s period). The target traced a 3D arc trajectory. The theoretical ITV was calculated using a mathematical model. Deviceless 4DCT scans (10-phase MIPs) were acquired at six rotation times (0.5–1.0 s). Intensity-based ITVs were auto-segmented using 50% HU thresholding. Deviations from the theoretical ITV were quantified by percentage ITV difference (?%ITV), Equivalent Sphere Diameter (ESD), and Edge Response Width (ERW). ANOVA and Pearson’s correlation (r) were applied to evaluate the impact of gantry rotation times on ITV accuracy.

Results:

The theoretical ITV was calculated as 24.77 cm³ (ESD = 3.62 cm). Intensity-based ITVs from Deviceless 4DCT across rotation times (0.5–1.0 s) were compared to the ground truth using one-sample t-tests. At 0.5 s rotation time, the 4DCT ITV (24.05 ± 0.44 cm³, ?%ITV = -2.9 ± 1.3%) showed no significant deviation from the theoretical value (p = 0.106). Longer rotation times resulted in progressive underestimation: 0.6 s (23.01 ± 1.03 cm³, ?%ITV = -7.1 ± 5.1%, p = 0.067), 0.7 s (22.12 ± 2.92 cm³, ?%ITV = -10.7 ± 11.8%, p = 0.167), 0.8 s (22.40 ± 0.75 cm³, ?%ITV = -9.6 ± 3.0%, p = 0.008), 0.9 s (21.92 ± 1.02 cm³, ?%ITV = -11.5 ± 4.1%, p = 0.011), and 1.0 s (20.62 ± 0.70 cm³, ?%ITV = -16.8 ± 2.8%, p < 0.001). ANOVA confirmed significant differences across rotation times (p < 0.001), with post-hoc tests identifying 0.8–1.0 s as outliers (p < 0.05). For geometrical congruence, 0.5 s rotation achieved the closest ESD agreement (3.58 ± 0.02 cm vs. theoretical 3.62 cm, ?%ESD = -0.99%), outperforming others. Edge Response Width (ERW) analysis revealed degraded spatial resolution with longer rotations: AP/SI/LL ERW at 0.5 s (0.31/0.28/0.29 cm) increased to 0.72/0.56/0.54 cm at 1.0 s (r = 0.94–0.98, p < 0.001). The stairstep effect had no measurable impact on ITV accuracy.

Conclusion:

The 0.5 s rotation time provided the most accurate ITV delineation, showing non-significant deviation from the mathematical model (-2.9%) and superior geometrical fidelity (?%ESD = -0.99%, ERW < 0.3 cm). Prolonged rotation times (>0.7 s) introduced significant ITV underestimation (-9.6% to -16.8%) and edge blurring. The strong correlation (r > 0.9) between rotation time and ERW highlights the necessity of rapid imaging for effective motion management.