Residual U-Net Convolutional Neural Network Architecture for Low-Dose CT Denoising

Mattias P. Heinrich 1 , Maik Stille 2 ,  and Thorsten M. Buzug 2
  • 1 Institute of Medical Informatics, University of Lübeck,, Lübeck, Germany
  • 2 Institute of Medical Engineering, University of Lübeck,, Lübeck, Germany

Abstract

Low-dose CT has received increasing attention in the recent years and is considered a promising method to reduce the risk of cancer in patients. However, the reduction of the dosage leads to quantum noise in the raw data, which is carried on in the reconstructed images. Two different multilayer convolutional neural network (CNN) architectures for the denoising of CT images are investigated. ResFCN is based on a fully-convolutional network that consists of three blocks of 5×5 convolutions filters and a ResUNet that is trained with 10 convolutional blocks that are arranged in a multi-scale fashion. Both architectures feature a residual connection of the input image to ease learning. Training images are based on realistic simulations by using the XCAT phantom. The ResUNet approach shows the most promising results with a peak signal to noise ratio of 44.00 compared to ResFCN with 41.79.

If the inline PDF is not rendering correctly, you can download the PDF file here.

OPEN ACCESS

Journal + Issues

Current Directions in Biomedical Engineering is an open access journal and closely related to the journal Biomedical Engineering - Biomedizinische Technik. CDBME is a forum for the exchange of knowledge in the fields of biomedical engineering, medical information technology and biotechnology/bioengineering for medicine and addresses engineers, natural scientists, and clinicians working in research, industry, or clinical practice.

Search