3134 - Optimizing 4DCT Phase Sampling for Small Internal Target Volume Delineation in Lung Stereotactic Body Radiotherapy

Purpose/Objective(s): 4DCT is essential for internal target volume (ITV) delineation in lung SBRT. The conventional 10-phase binning protocol (0%-90%) represents a discretized approximation of continuous respiratory motion, potentially leading to ITV underestimation for small targets (<2 cm) where volumetric errors substantially impact dose uncertainty. While 20-phase 4DCT is technically available on major scanner platforms, it remains clinically underutilized due to limited evidence of benefit. This phantom study systematically compared the geometric accuracy of 10-phase versus 20-phase 4DCT for small target ITV delineation against a mathematically defined ground truth

Materials/Methods: A custom motion lung phantom was built with a spherical nylon target (F = 1 cm, ? ˜ 1.14 g/cm³) embedded in lung-equivalent foam (? ˜ 0.32 g/cm³), closely mimicking clinical tumor-to-lung contrast. The phantom underwent 4DCT scanning on a 16-slice CT Simulator with standardized parameters. Five independent scans were reconstructed into both 10-phase and 20-phase bins. ITVs were auto-delineated using a standardized 50% iso-intensity threshold on maximum intensity projection (MIP) images. Ground truth ITV (ITV_GT) was calculated mathematically from target geometry and motion parameters. Statistical analysis employed the Wilcoxon signed-rank test (a=0.05).

Results: The 10-phase protocol consistently underestimated target volume, with a mean percentage difference of -5.61% (range: -4.4% to -7.4%) relative to ITV_GT. In contrast, the 20-phase protocol demonstrated excellent agreement, with a mean percentage difference of +0.92% (range: -1.1% to +2.7%). The median absolute volumetric deviation for ITV20 (0.015 cm³) was substantially smaller than for ITV10 (0.073 cm³). Remarkably, the 20-phase protocol yielded smaller deviations than the 10-phase protocol in 100% (5/5) of paired samples. The Wilcoxon signed-rank test revealed a strong trend toward superior accuracy for the 20-phase protocol (p = 0.0625), with the test statistic confirming the perfect directional consistency. While not reaching conventional statistical significance at a=0.05, likely due to limited sample size, the observed effect size and consistency strongly support the clinical benefit of increased temporal sampling.

Conclusion: The 20-phase 4DCT protocol provides substantially improved geometric accuracy for small target ITV delineation compared to the standard 10-phase approach, reducing mean volumetric error from -5.6% (underestimation) to +0.9% (negligible overestimation). The consistent underestimation observed with 10-phase sampling poses a potential risk for geographic miss and tumor underdosage in high dose SBRT. These findings support the clinical adoption of 20-phase 4DCT for small lung targets (<2 cm), where the modest increase in data processing is outweighed by the critical need for precision. A confirmatory study with larger sample size is warranted to achieve formal statistical significance.