Research On Digitalize X-ray Imaging Nonuniformity Correction Methods

Computed X-ray (CR) image has its own intrinsic characteristics other than normal video image. The nonuniformity caused by "double fiber bundle" greatly damages the quality of CR image besides small useful signal and large noise interference (inherent noise, Poisson noise), and goes against clinical diagnosis. The performance of the double fiber bundle, the key component directly affects the CR system, how to correct the nonuniformity of CR image effectively is the key point of improving the quality of CR image and also a difficult issue to be resolved in computed X-ray medical image field.For this purpose, correction arithmetic design and scanning hardware structure design are developed in this paper to remove the nonuniformity of CR image.In the correction arithmetic aspect: the linear and non-linear correction arithmetics usually adapted at home and abroad are discussed.Linear correction arithmetic: when the light transmission coefficient of fiber bundle is approximately a constant, the nonuniformity can be corrected through linear correction method. The experimental results show, the light transmission coefficient of fiber bundle is non-linear and irregular, the linear correction will bring large error.Non-linear correction arithmetic: before using non-linear correction arithmetic, we should know its response function and perform the non-linear correction. Because the response function of fiber bundle is usually irregular, it is difficult to choose a suitable curve model. It would be out of fit if the curve model were unsuitable, producing larger error. For these reasons, normal correction arithmetic is unsuitable for nonuniformity of computed X-ray image.A subsection linear correction arithmetic based on multipoint calibration is presented in this paper with regard to the insufficiency of normal correction arithmetic used in the correction of X-ray image nonuniformity.The principle of the subsection linear correction arithmetic based on multipoint calibration is that the response curve of fiber bundle is divided into several sections, in each of which linear correction is used which, to approximating nonlinear response byusing subsection linear function. The calibration multipoints should be close, the linear segment after dividing is nearly of nonlinear response, the gray level after correction of each segment is compute using linear correction arithmetic.The subsection linear correction arithmetic based on multipoint calibration is proved to be a effective arithmetic used in computer X-ray image non uniformity correction, the nonuniformity before correction is 13%, it becomes 2.6% after correction by this arithmetic, which meets the demand of nonuniformity less than 5%, the quality of image after correction also meets the demand of clinic diagnose. The presented arithmetic has been applied to innovative project "CR" successfully, and produced satisfactory results. ?Scanning hardware structure: a "new type ScanHead" design plan is presented to avoid the nonuniformity of CR image caused by fiber bundle completely. That is to design new type ScanHead system. Two new ScanHeads are presented, one is called "line scanner"; other is called "point scanner", and the structure design methods are given.The new scanner structure not only eliminates the nonuniformity of CR, but also provides scientific reference for developing new type CR ScanHead system, having theoretic significance and applicable value.