Multi-Branch Deep Residual Convolutional Neural Networks for Enhancing the Spatial Resolution of 3D Computed Tomography Images. — ICTMS Committee Dep of Applied Maths Australian National University
  1. Schedule
  2. Theme 2: Automated methods for volume segmentation/analysis, e.g., machine learning
  3. Multi-Branch Deep Residual Convolutional Neural Networks for Enhancing the Spatial Resolution of 3D Computed Tomography Images.

Multi-Branch Deep Residual Convolutional Neural Networks for Enhancing the Spatial Resolution of 3D Computed Tomography Images. (127)

kaicong sun 1 , sven simon 1
  1. University of Stuttgart, Stuttgart, BADEN-WüRTTEMBERG, Germany

In this paper, we propose a multi-branch deep residual neural network X-EDSR based on the state-of-the-art network EDSR to enhance the spatial resolution of 2D CT slices of 3D CT images. In particular,X-EDSR-2, X-EDSR-4, X-EDSR-8 with 2, 4 and 8 branches, respectively, reduce the parameters of networksto 76.9%, 65.4% and 59.5% of EDSR but maintaining the performance in terms of PSNR and SSIM

Full Paper