Research Of Calibration And Evaluation Algorithms Of The Flat Panel Detector In Dental CT System

With the development of CT technology, Cone Beam CT imaging system which is based on flat panel detector plays an important role in the modern health sector for its high image acquisition, the imaging speed, etc. Flat panel detector as a highly integrated and high-precision imaging component CT system is also faced with many problems and challenges. The FPD output correction is one of the most important factors which affect the quality of the output image. Meanwhile, some imaging properties of FPD also determine whether the image quality is good or not. Based on the above two issues, this paper carried out the research of the FPD output correction and performance evaluation method. Its main contents are:Based on the gain correction of existing methods, this thesis proposes the gain correction method based on threshold and the gain correction method based on histogram. The first mothed calculates the gain correction coefficient from air projection image. Then the threshold value detenninates for the response of each pixel to substituting with the compensation value. The second one is based on the gain coefficient image to fit the histogram and calculates the corrected threshold value, and then determines the inconsistency response of each pixel to compensate. The above method can effectively remove the inconsistently response of the detector and improve the quality of the resconstruction image.In view of the existing bad pixel identification method, this thesis proposes the bad pixel identification method based on histogram and the improved mothed based on the threshold value. The first mothed is accomplished by setting a good step in the range of calculated threshold, and then automatically return to a threshold value to determine the location of bad pixel. The second one is based on the histogram of the mean image and the variance image which based on the air response of each angle projection to fit for a distribution, and then get the coefficients to positioning the location of the bad pixel image.Based on the existing bad pixel compensation method, this thesis proposes the order interpolation method based on templates and the recycling interpolation method based on templates. The first mothed compensated the bad pixel by a specific sequence based on the positioning image. It simplifying the compensation of the different classification of bad pixel and making the process easier and faster. The second one is based on the number of bad pixels interpolating compensation sequentially for all bad pixels until the number reaches zero. It is time to solve the poor accuracy of compensating of the existing method.This thesis is based on the calculations of the imaging properties of flat panel detector and proposes quality assessment method based on the nature of the flat panel detector itself. It makes it by collecting a small amount of data which can objectively evaluate the quality of flat panel detector itself.