Comparative evaluation of 4DCT and 4DCBCT for motion and volume measurement accuracy in a dynamic phantom

Effective motion management is essential for accurate treatment planning, target localization and dose delivery in lung stereotactic body radiotherapy (SBRT). Lung tumors typically move 5-30 mm in superior-inferior (SI) direction. The internal target volume (ITV) may be expanded up to 5 mm for day-to-day setup variations. Four-dimensional cone beam computed tomography (4DCBCT), with a rotation scan time of 120 seconds, has the potential to improve on-line target localization; however, its accuracy in target motion and volume measurements should be compared to four-dimensional computed tomography (4DCT) with a scan time of < 60 seconds.

To compare the accuracy of 4DCT and 4DCBCT in measuring target motion and volume in a phantom programmed with both regular and patient-derived irregular respiratory waveforms.

A thorax phantom with a known target was imaged using 4DCT and 4DCBCT under controlled sinusoidal and irregular breathing patterns. 4DCBCT images were reconstructed using both Basic (i.e. FDK) and Advance (i.e. modified Mckinnon-Bates) algorithms where the Advance algorithm subtracts re-projection of prior image (using the entire un-binned projections) from phase sorted projections to reduce streak artifacts. Motion amplitudes were calculated from target centroids across respiratory phases, as well as from 3DCBCT-based motion blur profiles. Target volumes were assessed in average intensity projection (AIP) and maximum intensity projection (MIP), and in the reconstructed phases. Measured amplitudes were compared against programmed values, while the target volumes were compared in AIP, MIP and the phases between 4DCT and 4DCBCT.

For regular motion, 4DCT demonstrated the smallest deviations from the programmed amplitudes (SI, Left-right (LR) and Anterior-posterior (AP) 95% confidence intervals (CI): -0.8 to 0.1 mm, -0.3 to 0.2 mm and -0.5 to 1.1 mm respectively). In regular motion, 4DCBCT Advance algorithm slightly underestimated the SI motion by 1 mm (SI CI: -1.5 to -0.4 mm, LR CI -1.4 to 1.2 mm, AP CI -1.2 to 0.6 mm) while the Basic algorithm provided slightly larger variability (SI CI: -1.1 to 0.7 mm, LR CI -1.1 to 0.6 mm, AP CI -0.9 to 1.3 mm). 3DCBCT-based estimates showed higher variability across all motion directions (SI CI -1.7 to 1.6 mm, LR CI -1.9 to 1.6 mm, AP CI -1.6 to 1.0 mm). For irregular motion, both 4DCT and 4DCBCT Advance yielded differences from average programmed amplitude with no consistent superiority. The measured target volume showed strong agreement between 4DCT and 4DCBCT. The target volume expansion due to within-phase blurring showed a very similar trend between 4DCT and 4DCBCT Advance.

4DCBCT within clinical workflow provides comparable motion and volume measurement accuracy to 4DCT. 3DCBCT estimated motion amplitude showed large differences from the programmed amplitude. 4DCBCT offers explicit and confident motion resolution, and phase-dependent target volume at the time of treatment. These findings support the feasibility of integration of 4DCBCT, pending clinical validation, for image guidance in lung SBRT, especially where motion consistency between simulation and treatment is uncertain. Patient studies are needed to capture the influence of patient-specific breathing irregularities such as hysteresis, baseline drift or anatomical deformation.