Cone-beam-CT guided radiation therapy: A model for on-line application

Background and Purpose: This paper presents efficient and generalized processes for the clinical application of on-line X-ray volumetric cone-beam CT imaging (XVI) to improve the accuracy of patient set-up in radiation therapy. XVI image-guided therapy is illustrated by application to two contrasting sites, intra-cranial radiosurgery and prostate radiation therapy, with very different characteristics regarding organ motion, treatment precision, and imaging conditions. Patients and methods: On-line set-up errors are determined in a two-step process. First the XVI data is registered to the planning data by matching the machine-isocenter with the planning-isocenter, respectively. The machine isocenter is defined in the XVI data during the reconstruction. The planning-isocenter is defined during the planning process in the planning CT data. Set-up errors are then determined from a second registration to remove residual displacements. The accuracy of the entire procedure for on-line set-up error correction was investigated in precision radiosurgery phantom studies. Results: The phantom studies showed that sub-pixel size set-up errors (down to 0.5 mm) can be correctly determined and implemented in the radiosurgery environment. XVI is demonstrated to provide quality skull detail enabling precise skull based on-line alignment in radiosurgery. A 'local XVI' technique was found to give encouraging soft-tissue detail in the high-scatter pelvic environment, enabling on-line soft-tissue based set-up for prostate treatment. The two-step process for determination of set-up errors was found to be efficient and effective when implemented with a dedicated six panel interface enabling simultaneous visualization on the XVI and planning CT data sets. Conclusions: XVI has potential to significantly improve the accuracy of radiation treatments. Present image quality is highly encouraging and can enable bony and soft-tissue patient set-up error determination and correction. As with all image guided treatment techniques the development of efficient procedures to utilize on-line data are of paramount importance.