Weighted comprehensive score evaluation of CBCT image guided positioning accuracy in lung cancer radiation treatment

Abstract

Background and Objective: Cone-beam computed tomography (CBCT) has been widely used in radiation treatment positioning as so called image guided radiation therapy (IGRT) technique. It is impossible to correct all local position error by doing global IGRT positioning since deformation always happens in the patient and the positioning shifts relies on the image registration technique used in the image guidance. This study evaluated and analyzed the regional accuracy of CBCT guided positioning used different image registration strategy with a comprehensive scoring method to provide a reasonable guidance for clinic application of cancer IGRT. Materials and Methods: Planning CT and CBCT images acquired during the first radiation treatment positioning of 15 lung cancer cases were studied, registered with different image registration algorithms (bony and gray registration) and different registration area (target-only, ipsilateral structure and body).The CBCT target volume (GTVCBCT) coverage by planning target volume (PTVCT), the dice similarity coefficient (DSC) between the planning CT target volume (GTVCT) and GTVCBCT, the volume of organ at risk on planning CT (OARCT) and on CBCT (OARCBCT), and the position deviation of the GTV geometric center were compared. A weighted comprehensive scoring function was used to evaluate the positioning correction. The comprehensive registration factors (CRF) was defined as following, while the CRF values 100 meant all specific structures registered 100%.(formula presented) Results: For the tested lung cases, the bony registration had less accuracy in almost all evaluated specifications, especially when registered the target only. The coverage ratio of PTVCT to GTVCBCT of bony and gray registration in the three groups of registering target only, ipsilateral structure, body were 66%±35% and 97%±8% (P=0.005), 98%±5% and 99%±2%(P=0.034), 98%±4% and 98%±4% (P=0.478), respectively. Using gray registration to register the ipsilateral structure had the best result, with the DSC of GTV and OARs of 0.86±0.10 (GTV), 0.71±0.10 (Esophagus), 0.76±0.10 (Spinal cord), 0.89±0.05 (Heart), and the deviation of GTV center of 2.5mm±1.6mm, meanwhile the CRF was 55.28±40.59. Conclusions: CBCT guided positioning is able to ensure that the GTV covered by PTV, which is the first goal of IGRT, when suitable image registration strategy is applied. Using gray registration and registering the ipsilateral structure is recommend for CBCT guided lung cancer radiation treatment positioning. The weighted comprehensive scoring is helpful for evaluating the image registration of IGRT.