Uncertainty Quantification in Deep MRI Reconstruction

@article{Edupuganti2021UncertaintyQI,
  title={Uncertainty Quantification in Deep MRI Reconstruction},
  author={Vineet Edupuganti and Morteza Mardani and Shreyas S. Vasanawala and John M. Pauly},
  journal={IEEE Transactions on Medical Imaging},
  year={2021},
  volume={40},
  pages={239-250}
}
Reliable MRI is crucial for accurate interpretation in therapeutic and diagnostic tasks. However, undersampling during MRI acquisition as well as the overparameterized and non-transparent nature of deep learning (DL) leaves substantial uncertainty about the accuracy of DL reconstruction. With this in mind, this study aims to quantify the uncertainty in image recovery with DL models. To this end, we first leverage variational autoencoders (VAEs) to develop a probabilistic reconstruction scheme… 
K-space refinement in deep learning MR reconstruction via regularizing scan specific SPIRiT-based self consistency
TLDR
This paper proposes a novel refinement method based on SPIRiT (Iterative Self-consistent Parallel Imaging Reconstruction from Arbitrary k-Space) formulation to reduce the k-space errors and enable reconstruction of improved high-frequency image details and textures.
Prospective Deployment of Deep Learning in MRI: A Framework for Important Considerations, Challenges, and Recommendations for Best Practices
TLDR
A framework for important considerations that consist of appropriate model training paradigms, evaluation of model robustness, downstream clinical utility, opportunities for future advances, as well recommendations for best current practices is provided.
High Fidelity Deep Learning-based MRI Reconstruction with Instance-wise Discriminative Feature Matching Loss
TLDR
UFLoss, a patch-based unsupervised learned feature loss, is presented, which allows the training of DL-based reconstruction to obtain more detailed texture, finer features, and sharper edges with higher overall image quality under DL- based reconstruction frameworks.
NPBDREG: A Non-parametric Bayesian Deep-Learning Based Approach for Diffeomorphic Brain MRI Registration
TLDR
The NPBDREG is introduced, a fully non-parametric Bayesian framework for unsupervised DNN-based deformable image registration by combining an Adam optimizer with stochastic gradient Langevin dynamics (SGLD) to characterize the true posterior distribution through posterior sampling.
Robust Compressed Sensing MRI with Deep Generative Priors
TLDR
This paper trains a generative prior on brain scans from the fastMRI dataset, and shows that posterior sampling via Langevin dynamics achieves high quality reconstructions and is robust to changes in the ground-truth distribution and measurement process.
Denoising Score-Matching for Uncertainty Quantification in Inverse Problems
TLDR
This work proposes a generic Bayesian framework forsolving inverse problems, in which the use of deep neural networks are limited to learning a prior distribution on the signals to recover, and adopts recent denoisingscore matching techniques to learn this prior from data.
Sampling possible reconstructions of undersampled acquisitions in MR imaging
TLDR
A low dimensional latent space is introduced and the posterior distribution of the latent vectors given the acquisition data in k-space is modeled, from which one can sample in the latent space and obtain the corresponding images.
Uncertainty-Based Biological Age Estimation of Brain MRI Scans
TLDR
A new framework for organ-specific BA estimation utilizing 3D magnetic resonance image (MRI) scans is proposed and the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer’s patients is demonstrated.
AI-based Reconstruction for Fast MRI - A Systematic Review and Meta-analysis
TLDR
This meta-analysis systematically review the deep learning-based CS techniques for fast MRI, describe key model designs, highlight breakthroughs, and discuss promising directions to assess the pivotal role of deep learning in CS-based acceleration for MRI.
A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
TLDR
This study reviews recent advances in UQ methods used in deep learning and investigates the application of these methods in reinforcement learning (RL), and outlines a few important applications of UZ methods.
...
1
2
...

References

SHOWING 1-10 OF 42 REFERENCES
Uncertainty Quantification in Deep Learning for Safer Neuroimage Enhancement
TLDR
Methods to characterise different components of uncertainty in medical image enhancement problems and demonstrate the ideas using diffusion MRI super-resolution to highlight three key benefits of uncertainty modelling for improving the safety of DL-based image enhancement systems.
DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction
TLDR
This paper provides a deep learning-based strategy for reconstruction of CS-MRI, and bridges a substantial gap between conventional non-learning methods working only on data from a single image, and prior knowledge from large training data sets.
Compressed Sensing MRI Reconstruction with Cyclic Loss in Generative Adversarial Networks
TLDR
A novel deep learning-based generative adversarial model, RefineGAN, for fast and accurate CS-MRI reconstruction, which outperforms the state-of-the-art CS- MRI methods by a large margin in terms of both running time and image quality via evaluation using several open-source MRI databases.
Compressed Sensing MRI Reconstruction Using a Generative Adversarial Network With a Cyclic Loss
TLDR
It is demonstrated that the proposed novel deep learning-based generative adversarial model, RefineGAN, outperforms the state-of-the-art CS-MRI methods by a large margin in terms of both running time and image quality via evaluation using several open-source MRI databases.
Neural Proximal Gradient Descent for Compressive Imaging
TLDR
This work develops a proximal map that works well with real images based on residual networks with recurrent blocks, and significantly outperforms conventional non-recurrent deep ResNets by 2dB SNR and outperforms state-of-the-art compressed-sensing Wavelet-based methods by 4 dB SNR.
Deep Generative Adversarial Neural Networks for Compressive Sensing MRI
TLDR
A novel CS framework that uses generative adversarial networks (GAN) to model the (low-dimensional) manifold of high-quality MR images that retrieves higher quality images with improved fine texture details compared with conventional Wavelet-based and dictionary- learning-based CS schemes as well as with deep-learning-based schemes using pixel-wise training.
Unsupervised Learning with Stein's Unbiased Risk Estimator
TLDR
It is shown that Stein's Unbiased Risk Estimator (SURE) and its generalizations can be used to train convolutional neural networks (CNNs) for a range of image denoising and recovery problems without any ground truth data.
Generating synthetic CTs from magnetic resonance images using generative adversarial networks
TLDR
A GAN model using a single T1-weighted MR image as the input that generates robust, high quality synCTs in seconds is developed and validated and offers strong potential for supporting near real-time MR-only treatment planning in the brain.
Deep residual learning for compressed sensing MRI
TLDR
A novel deep residual learning algorithm to reconstruct MR images from sparsely sampled k-space data is proposed and a deep convolutional neural network is proposed to learn the aliasing artifacts.
Image reconstruction by domain-transform manifold learning
TLDR
A unified framework for image reconstruction—automated transform by manifold approximation (AUTOMAP)—which recasts image reconstruction as a data-driven supervised learning task that allows a mapping between the sensor and the image domain to emerge from an appropriate corpus of training data is presented.
...
1
2
3
4
5
...