Geometric Calibration of Novel Cone-Beam CT Developed For Dedicated Stereotactic Radiotherapy Unit

Abstract

A novel Cone-Beam CT system has been under development for a dedicated stereotactic radiosurgery/radiotherapy unit (Gamma Knife PerfeXion) to be equipped with image guidance. The stability and the reproducibility of the imaging system were investigated using the full geometric calibration. The imaging system consists of a rotational arm (Parker system) holding X-ray tube (Dunlee PX 1483), flat panel detector (Varian 4030 CB) and in-house system control and reconstruction software (Pilot 1.0). The arm travels in the vertical direction through the linear guide system (Parker 406xr) from its home position to the imaging position by 1250 mm. The accuracy and precision of the vertical motion is measured using dial gauge each time before geometric calibration. Geometric calibrations were repeated over several days with 1) the fixed calibration phantom position and 2) repositioning the imaging system in order to measure its stability and reproducibility. The calibration phantom consists of 16 ball bearings precisely located in two circular trajectories in a cylindrical plastic phantom. All the geometric parameters including source position, detector position and detector tilt angles with respect to gantry angle were determined. The calibration algorithm previously developed and verified at the cone-beam CT systems on optical bench and clinical linear accelerator was applied. Positioning accuracy of the vertical linear guide measured with dial gauge is better than 0.02mm (average absolute error). More than 200 projection images of calibration phantom were acquired each time over 210 degrees of gantry rotation for geometric calibration. Piercing points was reliably reproduced and its position was always in radius of 0.2mm or half pixel radius. In the current design, the reproducibility of vertical position, V, is limited to the precision of mechanical pin structure which allows three different X-ray tube positions. Gantry angles are noticed to have the maximum error of 0.2 degrees. Detector tilt angles are all within 0.5 degrees. Considering the uncertainty of the calibration algorithm of 0.4o (roll and pitch) and 0.01o (yaw) in detector tilt angles, the yaw is found to be the only angle that is significantly tilted. Adjusting this angle can impact the image quality. Source and detector trajectories show interesting behavior of gantry tilt and wobbling motion although the deviation from the ideal trajectory is less than 0.2 mm, which has negligible impact on image quality. Geometric calibration showed excellent stability and reproducibility of the prototype imaging system. More extensive work on quality assurance including full geometric calibration will be done after the imaging system is installed on the treatment unit.