Abstarct: Medical images produced by X-ray of the organs containing mextal implants are often followed by corruption. These corruptions result in dark and bright streaking artifacts around the mextal implants that obscures the biological structures information of studying areas. In this thesis, a method have been proposed that can significantly reduce the mextallic artifacts in computed tomography images. Since the computed tomography images obtained as continues slice cuts of an area, the mextallic artifacts are not the same in the regions. Hence, existing correct images information can be used to reduce corrupted images. For mextallic artifact reduction in this thesis, the Radon transfer which is a line integration in various direction is used to transfer the images to sinogram domain. Afterward various sinograms decomposed to illumination-reflectance components using a homomorphic filter. Our studies shows that mextallic artifacts mostly appears in illumination component of the sinogram. In this research two methods are proposed to reduce the noise in illumination component of corrupted sinogram. In the first method, the illumination component improved using the taken knowledge from previous sinograms in a hierarchical process. In the second method, the illumination component of corrupted sinogram improved without using previous sinograms information with the linear interpolation method and the reflectance component of the sinogram. The result obtained from two proposed methods shows that the noise is significantly improved with maintaining the structures of biological tissues information