Session Item

Following 4D-CT imaging of lung and liver lesions, SBRT planning data often contains 4D-CT artifacts that counteract plan quality and clinical outcome. 4D-CT artifacts, in particular interpolation artifacts, mainly result from suboptimal imaging protocols and violation of the data sufficiency condition (DSC) at extended breathing periods. In this regard, we analyse inter- and intra-patient variability of breathing data and correlate 4D-CT artifacts in SBRT planning data with local control (LC) of lung and liver lesions.

The study includes 297 patients treated between 2012-2020 with SBRT. For each treatment, 10-phase 4D-CT images were acquired using Siemens Definition AS (0.09 pitch, 0.5s gantry rotation) for ITV definition. Patient breathing was recorded with Varian RPM during 4D-CT acquisition (297 curves; 90s beam-on time) and during dose delivery in 5 fractions for a subset of 204/297 patients (1020 curves). Breathing data was processed by automated peak detection and splitting the mean period at 5.25s. For breathing periods above 5.25s, the 4D-CT protocol is known to result in interpolation artifacts due to DSC violation. For correlation of LC and 4D-CT artifacts, a mixed effects Cox proportional hazards model was applied for a subset of 62/297 patients with 102 lung/liver lesions treated in 67 sessions. Fixed effects were risk factors of interest potentially influencing LC: artifact score (cf. criteria defined in legend of Fig. 1 left); patient breathing variability; and treatment-related covariates (metastatic site, GTV volume, BED enclosing ITV, treatment fractionation, chemotherapy prior SBRT).

Interpolation artifacts due to extended breathing periods (>5.25s) during 4D-CT imaging were observed in 70/297 (24%) patients. In the patient subset with breathing data of 4D-CT imaging and dose delivery, 30% of these patients showed a constant mean breath period above 5.25s. LC was observed in 17/102 (17%) lesions (patient subset 62/297). Significant univariable factors for LC were artifact score (severe CT artifacts vs. few CT artifacts; HR 8.22; 95%-CI 2.04-33.18) and median-split mean patient breathing period (>4.8s vs. ≤4.8s; HR 3.6; 95%-CI 1.2-10.8). Hazard ratio of DSC-split mean patient breathing period (>5.25s vs. ≤5.25s) was slightly lower (HR 2.3; 95%-CI 0.4-13.8). Corresponding Kaplan-Meier curves are presented in Fig. 1. Following multivariable analysis, artifact score remained as dominating factor, although statistically not significant (HR 10.3; 95%-CI 0.6-184.2).

The results support the assumption that artifact-affected SBRT planning data due to suboptimal 4D-CT imaging impacts SBRT plan quality and clinical outcome. Imaging protocol-related limitations associated with extended breathing periods can and should be overcome by patient breathing-adjusted 4D-CT imaging protocols.