Magnetic Resonance Imaging MRI is widely used in medical studies. It is a non-invasive technique. There are several noises which affect the performance of MRI. Salt and pepper noise is a kind of pulsing noise which is produced in images of MRI due to their sensors. Modified Non-Local Means MNLM filter is applied to remove this noise. Noise of salt and pepper of colored images of MRI are removed using 2-stages Modified Non-Local Means filter or 1-stage Modified Non-Local Means filter. These filters are compared to remove salt and pepper noise from these noisy colored images. Peak signal-to-noise ratio PSNR and structure similarity SSIM are used to evaluate quality of denoised images. The parameters of 2-stages Modified Non-Local Means are over performed of parameters of 1-stage Modified Non-Local Means filters. In 2-stages Modified Non-Local Means, when variance of salt and pepper noise is 0.01, PSNR of noisy image is 24.09dB. PSNR of denoised image is 35.19dB. SSIM of noisy image is 0.842. SSIM of denoised image is 0.9896. . In 1-stages Modified Non-Local Means, when variance of noise is 0.01, PSNR of noisy image is 24.19dB. PSNR of denoised image is 32.31dB. SSIM of noisy image is 0.844. SSIM of denoised image is 0.958.