The Algorithms Study Of Beam-hardening Correction In CT

Beam-hardening is caused by polychromatic energy spectrum and energy dependency attenuation coefficient. When the polychromatic rays cut through the object, the lower energy X-photon absorption as the photoelectric effect was first made through X-ray high-energy component material proportion of the increase reflected higher average energy, so with the transmitted ray thickness becomes more penetrating, the spectrum distribution of the peak moves to higher energy, which is beam hardening effect. It will lead to a uniform density in the reconstruction of the measured object slice CT images showed different brightness, the image pixel value distribution is the brightest among the black edges "cup" shape, that is, "cup" artifacts.The present research on beam hardening correction of great progress has been made to achieve a certain effect of beam hardening correction, but still not perfect, still needs further study.This paper has two improvements on traditional method of linearization correction: First, the traditional method of the equivalent polynomial correction method for finding the attenuation coefficient has been improved, The method of close to the origin of several groups of data lines fitting, fitting the slope selected to replace the traditional methods of fitting curves in the zeros of derivatives, in order to fit the slope as the equivalent attenuation coefficient of monochromatic, effectively reduced error. Second,the fitting function has been further improved, To better meet the relevant requirements of beam hardening curve, exponential function instead of the traditional method of polynomial functions, experimental results show that the correction results is better with exponential function.This paper also studied the hardening correction based on statistical iterative method. And build the math model according to the continuity of energy spectrum and the number of photons detected obeys Poisson distribution. Then make parameter estimation with the statistical method on the basis of traditional reconstruction algorithm. Experimental results show that the method effectively improves the image quality.