A method for denoising an image. The method includes training a fully convolutional neural network (FCN) and generating a reconstructed image by applying the FCN on the image. Training the FCN includes generating an nth training image of a plurality of training images, initializing the FCN with a plurality of initial weights, and repeating a first iterative process. The nth training image includes a training array In. The first iterative process includes extracting an nth denoised training image from the FCN, generating a plurality of updated weights, and replacing the plurality of initial weights with the plurality of updated weights. The nth denoised training image may be extracted by applying the FCN on the nth training image. In an exemplary embodiment, the nth denoised training image may include a denoised array În. The plurality of updated weights is generated by minimizing a loss function including Σn=1N|In−În|.