Comparison of image registration performed with MV cone beam CT and CT on rails and Syngo™ Adaptive Targeting software

Abstract

Background Minimization of geometric errors in treatment delivery is essential in modern conformal and intensity-modulated techniques. Aim In this paper two Siemens systems, MVision megavoltage cone beam CT, and CTVision (CT on rails), are compared. Material and Methods The reproducibility and uncertainty of the image registration procedure performed with Adaptive Targeting (AT) software were evaluated. Both systems were evaluated by means of simulating the clinical situation with an anthropomorphic phantom in three anatomical sites: head & neck, thorax and pelvis. Results The results for two methods of image registration, manual and automatic, were evaluated separately. The manual procedure was used by two users, more and less experienced. Conclusions The MVision system and CTVision and the Therapist Adaptive software ensure image registration with the uncertainty of about 2.0 mm (2 standard deviations). In the case of the automatic registration method better reproducibility of image registration was obtained for MVision. For CTVision the necessity of manual identification of the machine isocentre made the registration less reproducible. In the case of MVision, the automatic method was more reproducible than the manual one (smaller dispersion of results). In the case of CTVision, similar results were obtained for both registration methods. In the case of manual registration slightly better reproducibility for CT data acquired at 2 mm slice thickness and 2 mm slice separation than for data acquired at 5 mm slice thickness and 5 mm slice separation were obtained. Similar results of manual registration performed by more and less experienced users were obtained.