Purpose: In our conventional image registration workflow, 4D-CBCT was directly registered to reference HCT using a dual registration approach in the Elekta XVI software. In this study, we proposed a new HCT-CBCT auto-registration strategy using previously registered CBCT as the reference image.
Methods: From a previous CBCT session, the average 4D-CBCT was selected as CBCT0 and the registration vector of HCT-CBCT0 from clinician’s registration was recorded as F0. In the new CBCT session, auto-registration was first performed between the new average 4D-CBCT (CBCT1) and CBCT0, and the registration vector of CBCT0-CBCT1 was recorded as F1. During this auto-registration step, the reference point in CBCT0 was shifted from its image center by F0, while the reference point in CBCT1 was kept at its original position. This operation would eliminate the uncertainty associated with rotation vectors. Combining F0 and F1 mathematically, the final registration matrix of HCT-CBCT1 was derived and 6 registration parameters were solved. The final registration parameters were compared with clinician’s results (ground-truth) to evaluate its accuracy using a test dataset consisted of 32 challenging registration cases.
Results: The uncertainty of the new auto-registration method was -0.1±0.5, -0.0±1.0 and -0.2±0.7 mm in 3 translational directions (lateral, longitudinal and vertical, respectively) and 0.0±1.5, 0.5±1.7 and 0.5±0.8 degree for rotations. Two patients (6.25%) had translational uncertainty >2mm (max=2.5mm) and both occurred in the longitudinal direction. Meanwhile, the uncertainty of the conventional HCT-CBCT1 auto-registration was -0.4±2.6, -0.2±7.4, -1.4±3.6 mm for translations and -0.3±1.2, 0.1±1.5 and 0.0±1.1 degree for rotations with eleven patients (34.4%) had translation uncertainty >2mm (max=30mm) in at least one direction. Accuracy in translation was significantly improved with the new method while rotation accuracy stayed in the same order.
Conclusion: By utilizing prior registration knowledge, the proposed auto-registration method provides a quick and reliable starting solution for online HCT-CBCT registration.