Curvelet Domain De-tagging of Tagged Cardiac MR Images

This paper presents the de-tagging of the cardiac magnetic resonance images (MRI) in the curvelet domain. The second generation discrete curvelet transform captures the directional activities of an image as well as the directional high intensity peak of the magnitude spectrum in their subbands effectively. Hence, the curvelet transform is used to identify the high directional peak corresponds to the tag patterns in the magnitude spectrum and suppress the same using the curvelet coefficients. The de-tag method presented here undergoes three steps in the curvelet domain: (1) since, the fine scale subband of the curvelet decomposition capture the tag lines the fine scale isotropic wavelet subband coefficients are suppressed and at the initial stage. (2) Identifying the subbands which capture the tag patterns using the directional subband coefficients. (3) Filter out the coefficients of the identified subbands. The proposed method show the better results compared to the several existing de-tag methods.