Evaluation of Short-Arc CBCT-Based Image Registration Accuracy in Breast Cancer Radiotherapy
Authors: Md Mokhlesur Rahman
Institution: Gono University
Introduction
Image-guided radiotherapy (IGRT) relies on cone-beam computed tomography (CBCT) to achieve accurate image registration and patient positioning. However, the additional radiation exposure associated with CBCT imaging remains a concern. Short-arc CBCT (SA-CBCT), which uses a reduced number of projection images for image reconstruction, has recently emerged as a potential alternative to conventional CBCT. Although previous research has reported a decline in image quality with reduced projection data, the clinical utility of SA-CBCT and its impact on registration accuracy have not been directly established. This study aimed to assess the registration accuracy of SA-CBCT for tangential breast irradiation using both an anthropomorphic phantom and clinical patient data. Its clinical feasibility was investigated by comparing its performance with that of the conventional 200° CBCT acquisition protocol.
Methods
An anthropomorphic phantom was positioned on the radiotherapy treatment couch, and registration was performed between the planning CT and CBCT images. Couch shifts obtained using conventional CBCT and SA-CBCT were measured and compared. Registration accuracy was quantified using N, defined as the Euclidean norm of the vector difference between the registration shifts obtained from the two imaging methods. The effects of the central and total acquisition angles were evaluated. For the clinical assessment, registration data from 30 consecutive patients receiving whole-breast irradiation were analyzed. CBCT datasets were reconstructed using progressively reduced acquisition angles, ranging from 200° to 120°. The registration discrepancy N was calculated using the same approach as in the phantom experiment.
Results
In the phantom study, reducing the acquisition angle to 120° resulted in registration discrepancies of no more than 0.15 mm for three-axis registration and 0.55 mm for six-axis registration. In the clinical evaluation, the mean registration discrepancies were 0.64 mm for three-axis registration, with 84.7% of cases showing discrepancies within 1 mm, and 0.85 mm for six-axis registration, with 71.3% within 1 mm. Despite the reduced acquisition angle, the body contour and chest wall boundaries remained sufficiently visible for registration purposes.
Conclusion
Short-arc CBCT demonstrated promising registration accuracy for image-guided tangential breast irradiation, even when the acquisition angle was reduced to 120°. The preservation of essential anatomical features and the relatively small registration discrepancies indicate its potential as a clinically feasible alternative to conventional CBCT. Appropriate selection of acquisition parameters is necessary to maintain reliable image registration while potentially reducing imaging radiation exposure.