An Educated Warm Start for Deep Image Prior-Based Micro CT Reconstruction

  title={An Educated Warm Start for Deep Image Prior-Based Micro CT Reconstruction},
  author={Riccardo Barbano and Johannes Leuschner and Maximilian Schmidt and Alexander Denker and Andreas Hauptmann and Peter Maass and Bangti Jin},
  journal={IEEE Transactions on Computational Imaging},
Deep image prior (DIP) was recently introduced as an effective unsupervised approach for image restoration tasks. DIP represents the image to be recovered as the output of a deep convolutional neural network, and learns the network's parameters such that the model output matches the corrupted observation. Despite its impressive reconstructive properties, the approach is slow when compared to supervisedly learned, or traditional reconstruction techniques. To address the computational challenge… 

Uncertainty Estimation for Computed Tomography with a Linearised Deep Image Prior

This paper develops a method, termed as the linearised deep image prior (DIP), to estimate the uncertainty associated with reconstructions produced by the DIP with total variation regularisation (TV), and endow it with conjugate Gaussian-linear model type error-bars computed from a local linearisation of the neural network around its optimised parameters.

Unsupervised denoising for sparse multi-spectral computed tomography

This work investigates the suitability of learning-based improvements to the challenging task of obtaining high-quality reconstructions from sparse measurements for a 64-channel PCD-CT and proposes an unsupervised denoising and artefact removal approach.



Warmstart Approach for Accelerating Deep Image Prior Reconstruction in Dynamic Tomography

This work proposes to use an adaptive restart method having the ability to switch between warmand coldstart depending on the amount of inter-frame changes to cope with abrupt changes in the captured time-series.

Unsupervised knowledge-transfer for learned image reconstruction

A novel unsupervised knowledge-transfer paradigm for learned reconstruction within a Bayesian framework that can significantly improve reconstruction quality not only visually, but also quantitatively in terms of PSNR and SSIM, when compared with learned methods trained on the synthetic dataset only.

Deep Image Prior

It is shown that a randomly-initialized neural network can be used as a handcrafted prior with excellent results in standard inverse problems such as denoising, super-resolution, and inpainting.

Computed tomography reconstruction using deep image prior and learned reconstruction methods

The deep image prior approach is introduced in combination with classical regularization and an initial reconstruction to overcome end-to-end learned methods' deficiencies in inverse problems and generalization after training with insufficient data.

PET Image Reconstruction Using Deep Image Prior

Quantification results based on simulation and real data show that the proposed reconstruction framework can outperform Gaussian post-smoothing and anatomically guided reconstructions using the kernel method or the neural-network penalty.

Quantitative Comparison of Deep Learning-Based Image Reconstruction Methods for Low-Dose and Sparse-Angle CT Applications

This work presents the results of a data challenge, bringing together algorithm experts from different institutes to jointly work on quantitative evaluation of several data-driven methods on two large, public datasets during a ten day sprint, and observes that the deep learning-based methods are able to improve the reconstruction quality metrics in both CT applications.

Model Adaptation for Inverse Problems in Imaging

This work proposes two novel procedures that adapt the network to a change in the forward model, even without full knowledge of the change, and shows these simple model adaptation approaches achieve empirical success in a variety of inverse problems, including deblurring, super-resolution, and undersampled image reconstruction in magnetic resonance imaging.

A Bayesian Perspective on the Deep Image Prior

It is shown that the deep image prior is asymptotically equivalent to a stationary Gaussian process prior in the limit as the number of channels in each layer of the network goes to infinity, and derive the corresponding kernel, which informs a Bayesian approach to inference.

PET image denoising using unsupervised deep learning

The proposed unsupervised deep learning framework provides excellent image restoration effects, outperforming the Gaussian, NLM methods, BM4D, and Deep Decoder methods.

Model-Based Learning for Accelerated, Limited-View 3-D Photoacoustic Tomography

A deep neural network is presented that is specifically designed to provide high resolution 3-D images from restricted photoacoustic measurements to represent an iterative scheme and incorporates gradient information of the data fit to compensate for limited view artifacts.