TY - GEN
T1 - Deep Mr Image Super-Resolution Using Structural Priors
AU - Cherukuri, Venkateswararao
AU - Guo, Tiantong
AU - Schiff, Steven J.
AU - Monga, Vishal
N1 - Publisher Copyright:
© 2018 IEEE.
PY - 2018/8/29
Y1 - 2018/8/29
N2 - High resolution magnetic resonance (MR) images are desired for accurate diagnostics. In practice, image resolution is restricted by factors like hardware, cost and processing constraints. Recently, deep learning methods have been shown to produce compelling state of the art results for image super-resolution. Paying particular attention to desired hi-resolution MR image structure, we propose a new regularized network that exploits image priors, namely a low-rank structure and a sharpness prior to enhance deep MR image superresolution. Our contributions are then incorporating these priors in an analytically tractable fashion in the learning of a convolutional neural network (CNN) that accomplishes the super-resolution task. This is particularly challenging for the low rank prior, since the rank is not a differentiable function of the image matrix (and hence the network parameters), an issue we address by pursuing differentiable approximations of the rank. Sharpness is emphasized by the variance of the Laplacian which we show can be implemented by a fixedJeedback layer at the output of the network. Experiments performed on two publicly available MR brain image databases exhibit promising results particularly when training imagery is limited.
AB - High resolution magnetic resonance (MR) images are desired for accurate diagnostics. In practice, image resolution is restricted by factors like hardware, cost and processing constraints. Recently, deep learning methods have been shown to produce compelling state of the art results for image super-resolution. Paying particular attention to desired hi-resolution MR image structure, we propose a new regularized network that exploits image priors, namely a low-rank structure and a sharpness prior to enhance deep MR image superresolution. Our contributions are then incorporating these priors in an analytically tractable fashion in the learning of a convolutional neural network (CNN) that accomplishes the super-resolution task. This is particularly challenging for the low rank prior, since the rank is not a differentiable function of the image matrix (and hence the network parameters), an issue we address by pursuing differentiable approximations of the rank. Sharpness is emphasized by the variance of the Laplacian which we show can be implemented by a fixedJeedback layer at the output of the network. Experiments performed on two publicly available MR brain image databases exhibit promising results particularly when training imagery is limited.
UR - http://www.scopus.com/inward/record.url?scp=85062897175&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85062897175&partnerID=8YFLogxK
U2 - 10.1109/ICIP.2018.8451496
DO - 10.1109/ICIP.2018.8451496
M3 - Conference contribution
C2 - 30930696
AN - SCOPUS:85062897175
T3 - Proceedings - International Conference on Image Processing, ICIP
SP - 410
EP - 414
BT - 2018 IEEE International Conference on Image Processing, ICIP 2018 - Proceedings
PB - IEEE Computer Society
T2 - 25th IEEE International Conference on Image Processing, ICIP 2018
Y2 - 7 October 2018 through 10 October 2018
ER -