Solving ill-posed inverse problems using iterative deep neural networks
Jonas Adler, Ozan Öktem · Inverse Problems · 2017
Abstract We propose a partially learned approach for the solution of ill-posed inverse problems with not necessarily linear forward operators. The method builds on ideas from classical regularisation theory and recent advances in deep learning to perform learning while making use of prior information about the inverse problem encoded in the forward operator, noise model and a regularising functional. The method results in a gradient-like iterative scheme, where the ‘gradient’ component is learned using a convolutional network that includes the gradients of the data discrepancy and regulariser as input in each iteration. We present results of such a partially learned gradient scheme on a non-linear tomographic inversion problem with simulated data from both the Sheep-Logan phantom as well as a head CT. The outcome is compared against filtered backprojection and total variation reconstruction and the proposed method provides a 5.4 dB PSNR improvement over the total variation reconstruction while being significantly faster, giving reconstructions of 512 × 512 pixel images in about 0.4 s using a single graphics processing unit (GPU).
Cited by 529 works. doi:10.1088/1361-6420/aa9581 Open-access PDF
Authors
References
- Deep learning
- On the limited memory BFGS method for large scale optimization
- Deep Convolutional Neural Network for Inverse Problems in Imaging
- Inverse problems: A Bayesian perspective
- Probability Distributions on Banach Spaces
- Iterative Regularization Methods for Nonlinear Ill-Posed Problems
- Fast and flexible X-ray tomography using the ASTRA toolbox
- DART: A Practical Reconstruction Algorithm for Discrete Tomography
- Deep Learning Computed Tomography
- Fast Tomographic Reconstruction From Limited Data Using Artificial Neural Networks