Alignment of the CBCT with the reference CT is called image registration (IR). The parameters for utilizing the automated Elekta XVI IR software for IGRT of the prostate still remain to be defined. In this study, we compare several automated XVI IR parameters to manual registration to identify the optimal automated IR technique for the prostate gland. 280 prostate IRs were conducted as follows: 210 automated, and 70 manual IR were performed using 70 CBCT scans of seven patients. The three arms of the automated registrations were: (i) extended FOI/Bone + grey scale (double IR); (ii) limited FOI/GS (single IR); and (iii) extended FOI/GS (single IR). Automated IRs were compared to manual IRs; x, y, z shifts, failures, and errors recorded for off-line analysis. Based on the most successful parameters, a departmental protocol was developed and 432 automated IR were performed (on 20 patients) for analysis. Automated IR were classified as: Successful, failed, error, or unregistered. In arm 1, the rate of successful, failed, error, and unregistered IR were 52.8%, 1.5%, 8.6%, 37.1%, respectively, arm 2: 90% successful, 10% failed, arm 3: 100% successful. Using the arm 3 parameters for the 432 automated IRs, the incidence of unregistered scans was 0%, rescanning was required in 1% of treatments, and the time for performing the auto IR was < 5.5 minutes. We found that extended FOI + single (GS) IR results in shifts comparable to manual IR using automated XVI software. We experienced multiple unsuccessful registrations with the other methods. We conclude that when utilizing the Elekta XVI automated IR software, the extended FOI/single IR results in successful registrations most often. In addition, it is currently effectively used in our clinical practice.

Key words: IGRT; Prostate; Cone beam CT; Automated image registration; Setup correction; ELEKTA, and X-ray volumetric imaging (XVI).