Research On Online Geometric Calibration For Cone Beam CT System

Cone-beam CT(CBCT)system has been widely used in industrial and clinical imaging.Generally,reconstructed algorithms of CBCT systems have strict requirements of geomethric structures,which are called ideal geometric structures.However,in practice,due to manufacturing tolerance and setup errors,actual geometric structures easily deviate to the ideal ones.In this case,geometric artifacts are usually serious for CBCT system,damaging image quality,if the ideal structures are still used for image reconstruction.Recent years,many researchers have proposed a large amount of geometric calibration methods,aiming at eliminating geometric artifacts.Such methods can be divided into two categories,offline geometric calibration by using dedicated calibration phantoms and online geometric calibration by utilizing image content.However,methods falling into this two category are imperfect.In this paper,we proposed a plate-based geometric method with the main idea of allowing the imaging of calibration phantom and the imaging of object to happen at the same time,which can combine the advantage of offline calibrations method with the advantage of online calibration methods.On the one hand,this calibration method use calibration phantom to analytically calculate geometric parameters,which can acquire results in several seconds so this method can be used in clinical environment.On the other hand,due to simultaneous imaging strategy,this method can obtain valid parameters which may fail to be acquired from offline geometric calibration methods when inconsistency appears between pre-scanning and formal scanning.Simulation experiment verified that the accuracy of proposed calibration method was same approximately as the golden calibration method,direct linear transformation(DLT)calibration method.The error of calculated distance geometric parameters were within±1mm wihile the angular geometric parameters were within ±0.04 degree.In addition,both of simulation experiment and actual experiment proved that geometric artifacts of reconstructed images were totally eliminated by presenting plate-based geometric calibration method,compared with reconstructed images with uncalibration.In addition,quantitative indexes such as information entropy,image sharpness and contrast noise ratio of geometric artifacts were also improved after calibration.In addition,considering that the plate-based calibration method also uses calibration phantom,accuracy of method will continue to be affected by the error of fabrication of calibration phantom.Specifically,the actual postion of ball-bearings(BBs)may deviate the desined postion.In this regard,this paper presented a self-calibration method to estimate the real position of BBs on dedicated phantom to get rid of error of calibration phantom.The self-calibration method is not only applicable for the plate phantom used in our online calibration method,it also can be utilized to estimate actual position of BBs for any calibration phantom used in offline calibration method as long as there are BBs fabricated in phantom.Simulation experiment verified accuracy of such self-calibrition method was pretty high with individual coordinate error of BBs within±0.01mm if appropriate parameters were applicated.Actual experiment proved that geometric artifacts were eliminated on reconstructed images after using self-calibration method.In addition,self-calibration method makes the manufacture of the geometric calibration phantom no longer depend on machine tools.Users can fabricate geometric phantoms by themselves,which can save the cost and time cycle of the manufacture of the calibration phantom.More importantly,the self-calibration method only needs to be carried out once in calibration service life of calibration phantom so this method will not increase the difficulty and complexity of the geometric calibration.