A Novel Variation on Non Local Means Algorithm for Denoising of Contrast-Enhanced MR Images

Dynamic Contrast-enhanced magnetic resonance Imaging yields information about the Hemodynamic, properties of tissues and extracellular leakage space. The task is further complicated by the presence of hardware-induced noise, geometric distortions, and intensity non-uniformities (bias field), as well as motion artifacts resulting from patient movement during image acquisition.  This paper presents a new algorithm for denoising dynamic contrast-enhanced (DCE) MR images, called dynamic nonlocal means. It is a variation of the nonlocal means (NLM) algorithm. DNLM exploits the redundancy of information in the temporal sequence of images. We also perform the qualitative and quantitative results which suggest that the DNLM algorithm is more effective in attenuating noise.