A multi-scale (MS) decomposition method for contrast enhancement of Micro Dose (MD) X-ray images is presented in this paper. First, we get a denoised version of the input exploiting a non-local means filter with adaptable parameters setting that we defined in a former approach. Then, the MS representations of the input and of its de-noised version are combined to obtain an optimal image in terms of preservation of details and noise attenuation. The efficiency of the al-gorithm is demonstrated by quantitative and qualitative assessments on both phantoms and clinical MD images.