Solving Inverse Problems in Compressive Imaging with Score-Based Generative Models

Chong, Zhen Yuen; Zhao, Yaping; Wang, Zhongrui; Lam, Edmund Y.

doi:10.1109/dsaa60987.2023.10302579

Cited by 8 publications

(8 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, generative model has drawn considerable attention in the field of MRI. [20][21][22][23][24][25] Leveraging distribution prior, generative models exhibit promise in enhancing acceleration while retaining intricate details. The data distribution can be learned implicitly by directly representing the sampling process like generative adversarial network (GAN) 26 or explicitly by representing the probability density/mass like Bayesian networks 27 and score-based generative model 28 that estimates the gradients of the data distribution.…”

Section: Introductionmentioning

confidence: 99%

Integrating data distribution prior via Langevin dynamics for end‐to‐end MR reconstruction

Cheng,

Cui,

Zhu

et al. 2024

Magnetic Resonance in Med

View full text Add to dashboard Cite

PurposeTo develop a novel deep learning‐based method inheriting the advantages of data distribution prior and end‐to‐end training for accelerating MRI.MethodsLangevin dynamics is used to formulate image reconstruction with data distribution before facilitate image reconstruction. The data distribution prior is learned implicitly through the end‐to‐end adversarial training to mitigate the hyper‐parameter selection and shorten the testing time compared to traditional probabilistic reconstruction. By seamlessly integrating the deep equilibrium model, the iteration of Langevin dynamics culminates in convergence to a fix‐point, ensuring the stability of the learned distribution.ResultsThe feasibility of the proposed method is evaluated on the brain and knee datasets. Retrospective results with uniform and random masks show that the proposed method demonstrates superior performance both quantitatively and qualitatively than the state‐of‐the‐art.ConclusionThe proposed method incorporating Langevin dynamics with end‐to‐end adversarial training facilitates efficient and robust reconstruction for MRI. Empirical evaluations conducted on brain and knee datasets compellingly demonstrate the superior performance of the proposed method in terms of artifact removing and detail preserving.

show abstract

Section: Introductionmentioning

confidence: 99%

Integrating data distribution prior via Langevin dynamics for end‐to‐end MR reconstruction

Cheng,

Cui,

Zhu

et al. 2024

Magnetic Resonance in Med

View full text Add to dashboard Cite

show abstract

“…Diffusion-based image generative models are increasingly popular for synthesis of medical imaging. Importantly, diffusion models can be relatively easily extended to enforce data consistency during the sampling process [25], [26], [27].…”

mentioning

confidence: 99%

“…Diffusion image generation models (Fig. 1.A) approximate the score term with a neural network 𝛻 𝜇 𝑡 𝑙𝑜𝑔 𝑝 𝑡 (𝜇 𝑡 ) ≈ 𝑠 𝜃 (𝜇 𝑡 , 𝑡) [25], [26]. The parameters (𝜃) of the score-matching network 𝑠 𝜃 are obtained by training an unsupervised synthesis using a dataset assumed to be representative of the underlying noise-free image distribution 𝑝(𝜇 0 ).…”

mentioning

confidence: 99%

“…This can be achieved by replacing the unconditional score in Eq. ( 2) with the conditional score 𝛻 𝜇 𝑡 𝑙𝑜𝑔 𝑝(𝜇 𝑡 |𝑦) [26].…”

mentioning

confidence: 99%

“…The conditional score can be approximated with a deep CNN conditioned on the projection data, 𝛻 𝜇 𝑡 𝑙𝑜𝑔 𝑝(𝜇 𝑡 |𝑦) = 𝑠 𝜃 (𝜇 𝑡 , 𝑡, 𝑦), which must be trained in a fully supervised fashion. This formulation has shown promising performance [29], [30] but it faces challenges common to supervised training, including diminished performance in out-of-domain scenarios and inexplicit enforcement of data consistency during sampling [26], [27]. Here, we investigate an alternative based on diffusion posterior sampling (DPS) for x-ray reconstruction involving an approximation of the conditional score that applies a likelihood data-consistency correction term to the unconditional synthesis score function [27], [31].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Stationary x-ray tomography for hemorrhagic stroke imaging: sampling and resolution properties

Montes

McSkimming

Zbijewski

et al. 2022

7th International Conference on Image Formation in X-Ray Computed Tomography

View full text Add to dashboard Cite

Hemorrhagic stroke accounts for up to 20% of all stroke cases, and requires a treatment pathway drastically different to ischemic stroke. Prompt triage is therefore crucial and often only attainable with neuroimaging for intracranial hemorrhage (ICH) evaluation, for which MDCT is the frontline modality. Availability of ICH dedicated imaging in the pre-hospital setting, with portable CT systems, would facilitate early ICH diagnosis. However, current CT or cone-beam CT (CBCT) approaches often use conventional x-ray sources mounted on a rotating gantry limiting their minimum weight and footprint. Recent advances on cold-cathode, compact x-ray sources, based on carbon nanotube (CNT) technology, enable the development of ultra-compact designs based on source-array arrangements on stationary configurations. However, such geometrical arrangements show limited angular sampling, and sparse, non-stationary, volume sampling.In this work we present first investigation of geometric configuration and effects of 3D sampling pertinent to the task of ICH detection on an ultra-portable stationary CBCT for ICH imaging. The baseline configuration included 31 CNT sources on a curved array illuminating a curved panel detector (871 mm length), on a compact geometrical configuration (SDD = 690 mm). Metrics of sampling completeness, sampling density, and MTF shape and band-width integral were explored for configurations varying in source angular span (30°-170°), source array and detector curvature radius (250 mm to flat), use of 2D matrix source arrangements, and multi-acquisition protocols. The results show that sufficient sampling and resolution can be achieved with a combination of moderate curvature (~450 mm radius) of the source array and detectors, with better sampling properties for approximately matched curvature radii (up to 30% BWI improvement). Improved image quality was demonstrated with configurations featuring matrix source arrangements in combination with multi-acquisition protocols (around a 6% of improvement in sampling completeness).

show abstract

Ai4tech: X-AI enabling X-Tech with human-like, generative, decentralized, humanoid and metaverse AI

Cao

2024

Int J Data Sci Anal

View full text Add to dashboard Cite

The X-AI age is characterized by the synergy of diverse forms of intelligence-human, natural, social, and artificial (AI). It represents an ecosystem of comprehensive AI paradigms and the flourishing development of deep learning, large language models, and generative AI applications. X-AI reignites the debate on fundamental AI questions: What is AI? What constitutes machine intelligence? What are the implications when AI integrates with business and technology in the new era of artificial general intelligence and human-like AI? This article aims to inspire critical thinking, debate, and discussion on this new era of AI meeting technology, shaping AI4Tech through X-AI enabling X-Tech. We explore the ecosystems of X-AI and X-Tech and the general and domain-specific AI4Tech areas. X-AI enabling X-Tech nurtures a new age of smart businesses and intelligent technologies. The synergy between conventional, human-like, generative, decentralized, humanoid, and metaverse AI unlocks the potential to overcome previous limitations, impossibilities, unknowns, and dreams of AI and technology.

show abstract

Solving Inverse Problems in Compressive Imaging with Score-Based Generative Models

Cited by 8 publications

References 20 publications

Integrating data distribution prior via Langevin dynamics for end‐to‐end MR reconstruction

Integrating data distribution prior via Langevin dynamics for end‐to‐end MR reconstruction

Stationary x-ray tomography for hemorrhagic stroke imaging: sampling and resolution properties

Ai4tech: X-AI enabling X-Tech with human-like, generative, decentralized, humanoid and metaverse AI

Contact Info

Product

Resources

About