Implementation of respiratory-correlated cone-beam CT on Varian linac systems

• Main Author: Cropp, Robert James
• Language: English
• Published: University of British Columbia 2011
• Online Access: http://hdl.handle.net/2429/37220

Respiratory-correlated (4D) X-ray CT scans produce a set of images corresponding to different phases of a patient's breathing cycle. In external beam radiotherapy, information about a tumor's motion due to respiration can be used to optimize a treatment plan, provided the patient can be accurately aligned for treatment. Cone-beam CT (CBCT) systems are becoming widespread on treatment linac units and are used to aid in alignment. This thesis describes the implementation of respiratory-correlated cone-beam CT scans on two types of Varian units: iX and TrueBeam. Procedures for 4D CBCT scans on each type have been developed and used to image a moving phantom. The respiratory phase of the motion is recorded with the Varian Real-time Position Management (RPM) system, which uses optical tracking. To improve image quality, the gantry rotation speed is reduced below the default value of 6°/s: this reduces streak artifacts. Each projection image from the scan is assigned to one of ten phase bins according to its respiratory phase value. A 3D image is reconstructed for each phase bin with software developed for this project, which uses conventional Feldkamp-Davis-Kress filtered backprojection. Four 4D scans of a periodically moving phantom have been taken, with different gantry speeds and mAs values. To evaluate the effect of these scan parameters on image quality, and demonstrate a potential application of 4D CBCT, a procedure for automated tumor trajectory measurement has been developed. The measurement uses image registration between phase images, with either a rigid translation or a B-spline deformation algorithm. In the highest-quality images, the displacements of an insert in the phantom are measured within 1 mm of the correct values by both algorithms. In lower-quality images the translation algorithm is more robust. The two algorithms give similar results when applied to 4D CT images of actual lung cancer patients.