Medical image denoising using convolutional denoising autoencoders

Image denoising is important in medical image analysis. Different algorithms have been proposed in last three decades with varying denoising performances. More recently deep learning methods have shown great promise in image denoising and have outperformed all conventional methods. These methods are however limited for requirement of large training sample size and high computational costs. In this paper we show that denoising autoencoders constructed using convolutional layers can be used for efficient denoising of medical images with small training sample size and heterogeneous images can be combined to boost sample size for increased denoising performance. Simplest of networks can reconstruct images with corruption levels so high that noise and signal are not differentiable to human eye.