SU‐GG‐J‐114: Intrafraction Set‐Up Variability in SBRT Patients Monitored by IR Optical Localizer and KV X‐Ray Images

Abstract

Purpose: To investigate intrafraction setup variability in high precision stereotactic body radiotherapy (SBRT) by means of integrated infra‐red optical localization and stereoscopic kV X‐ray imaging. Method and Materials: We analyzed data coming from 24 patients treated with SBRT. Patient setup was realized through the ExacTrac X‐ray 6D system (BrainLAB, Germany), consisting of 2 infra‐red TV cameras for external fiducial localization and 2 couples of X‐ray source‐detectors for image registration. Before irradiation, patients were pre‐aligned relying on optical marker localization. Subsequently, patient position was refined through the automatic matching of X‐ray images to digitally reconstructed radiographs (DRR, 3 mm CT slice thickeness), providing 6 corrective parameters that were automatically applied using a robotic couch. The 3D position of the external configuration of control points was detected and stored before (PREmk) and after (POSTmk) the irradiation fraction. A final set of stereoscopic images was acquired and registered to DRR at the end of the treatment, relying on bony anatomy, thus obtaining 6 verification parameters (VP_post). Patient intrafraction motion was evaluated by comparing PREmk with POSTmk (optical measurement). Moreover VP_post parameters were applied to POSTmk thus defining a new configuration of control points (POSTmk‐Xray) that was compared to PREmk (X‐ray measurement). Results: According to the optical measurement, intrafraction motion measured on external fiducials was 0.61±0.53 mm (median±quartile). The X‐ray measurement overestimated intrafraction motion up to 2.05±1.8 mm. The latero‐lateral direction resulted more sensitive to residual errors in X‐ray measurement (0.34±2.88 mm, mean±std). Conclusion: Intrafraction motion, as monitored by the optical system, is not significant. When relying on X‐ray imaging registration, larger motion may be measured because of inherent limitations in DRR cranio‐caudal resolution and digital X‐ray image noise. Such results lead to further questions in assessing the accuracy of X‐ray based setup when integration with optical patient localization is not taken into account.