Skip to main content
. 2021 Mar 10;31(8):5498–5506. doi: 10.1007/s00330-020-07668-x

Fig. 1.

Fig. 1

Training and application of the deep learning reconstruction (DLR) algorithm. During training, the convolutional neural network (CNN) is trained to replicate an MBIR image with the reconstruction settings set to maximise image quality, given an input Hybrid-IR image (above). In the reconstruction process, the trained network is applied to Hybrid-IR images (below)