Yue Guan1, Yudu Li2,3, Xi Peng4, Yao Li1, Yiping P. Du1, and Zhi-Pei Liang2
1Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, China, 2Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 3Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 4Mayo Clinic, Rochester, MN, United States
1Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, China, 2Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 3Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 4Mayo Clinic, Rochester, MN, United States
This
paper presents a novel learning-based method for efficient selection of optimal
regularization parameters for constrained reconstruction. The proposed method
will enhance the speed, effectiveness and practical utility of constrained
reconstruction.
Figure
1. An
illustration of image quality manifolds for (a) SSIM, (b) data fidelity, and
(c) norm of regularization function. As can be seen, all image quality metrics,
as a function of
λ and ρ,
reside in a low-dimensional manifold and thus learnable from training data.
Figure
2. Comparison
of reconstruction results obtained from L-curve and the proposed method for
image deblurring. Note the reconstruction error was significantly reduced by
the proposed method, demonstrating its effectiveness in learning the optimal
regularization parameters.