一种改进的基于单尺度Retinex的LDCT医学影像增强算法

In order to improve the intensity and contrast qualities of CT (Computed Tomography) images and reduce the image noise in clinic application, this paper proposed a modified low-dose CT (LDCT) images contrast enhancement algorithm based on Single-Scale Retinex (SSR). Different from the classical Retinex methods estimate the illumination image in a complex way, the proposed algorithm estimates the illumination image approximately, and then does the reflectance image in a nonlinear way to amplify and emphasize these details. Experimental results show that the proposed method avoids more complex implementations and yields better performance of low-dose chest CT image enhancement over the conventional histogram equalization, the Frankle-McCann Retinex algorithm and McCann99 algorithm.