Multiple approaches to use deep learning for image restoration have recently\nbeen proposed. Training such approaches requires well registered pairs of high\nand low quality images. While this is easily achievable for many imaging\nmodalities, e.g. fluorescence light microscopy, for others it is not.\nCryo-transmission electron microscopy (cryo-TEM) could profoundly benefit from\nimproved denoising methods, unfortunately it is one of the latter. Here we show\nhow recent advances in network training for image restoration tasks, i.e.\ndenoising, can be applied to cryo-TEM data. We describe our proposed method and\nshow how it can be applied to single cryo-TEM projections and whole\ncryo-tomographic image volumes. Our proposed restoration method dramatically\nincreases contrast in cryo-TEM images, which improves the interpretability of\nthe acquired data. Furthermore we show that automated downstream processing on\nrestored image data, demonstrated on a dense segmentation task, leads to\nimproved results.\n