TY  - CONF
AU  - Dedmari, Muneer Ahmad
AU  - Conjeti, Sailesh
AU  - Estrada Leon, Edgar Santiago
AU  - Ehses, Philipp
AU  - Stöcker, Tony
AU  - Reuter, Martin
TI  - Complex Fully Convolutional Neural Networks for MR Image Reconstruction
M1  - DZNE-2020-00874
SN  - 978-3-030-00128-5 (print)
PY  - 2018
AB  - Undersampling the k-space data is widely adopted for acceleration of Magnetic Resonance Imaging (MRI). Current deep learning based approaches for supervised learning of MRI image reconstruction employ real-valued operations and representations by treating complex valued k-space/spatial-space as real values. In this paper, we propose complex dense fully convolutional neural network (CDFNet) for learning to de-alias the reconstruction artifacts within undersampled MRI images. We fashioned a densely-connected fully convolutional block tailored for complex-valued inputs by introducing dedicated layers such as complex convolution, batch normalization, non-linearities etc. CDFNet leverages the inherently complex-valued nature of input k-space and learns richer representations. We demonstrate improved perceptual quality and recovery of anatomical structures through CDFNet in contrast to its real-valued counterparts.
T2  - Machine Learning for Medical Image Reconstruction
CY  - 16 Sep 2018 - 16 Sep 2018, Granada (Spain)
Y2  - 16 Sep 2018 - 16 Sep 2018
M2  - Granada, Spain
LB  - PUB:(DE-HGF)3 ; PUB:(DE-HGF)6
DO  - DOI:10.1007/978-3-030-00129-2_4
UR  - https://pub.dzne.de/record/145540
ER  -