Awesome Computed Tomography EN 在 GitHub 上加星标

自监督与免训练方法

自监督与免训练方法是指在没有成对干净真值的情况下重建或去噪 CT 图像。它们利用测量数据本身的结构,比如重复或拆分数据中相互独立的噪声、与实测正弦图的一致性、对变换的不变性,或者未经训练的网络本身的隐式偏置。

它们之所以重要,是因为 CT 中干净的成对训练数据很稀缺:不能为了研究让患者分别在全剂量和低剂量下各扫一次,植入物扫描也不存在无金属版本,工业和同步辐射数据也很少附带参考图像。

主要方法族有:只用含噪数据训练的 Noise2Noise 类去噪;拟合单次扫描的深度图像先验及其他免训练网络;表示体数据并拟合实测投影的逐扫描神经场;等变成像损失;以及监督与无监督混合的方案。它们避免了训练集偏差,但测试时通常更慢,并需要谨慎的提前停止或正则化。

在可检索的目录中浏览全部 35 篇「自监督与免训练方法」论文

入门推荐

如果你刚接触这个主题,可以先读这几篇;由维护者挑选。

论文

共 35 篇,按时间倒序列出最新的 30 篇。

论文发表出处年份链接
3D Field of Junctions: A Noise-Robust, Training-Free Structural Prior for Volumetric Inverse ProblemsECCV2026paper · code
Deep Pseudo-Proximal Map: A Self-Supervised Data-Fitting Agent for Iterative Reconstruction预印本2026paper
Regularizing INR with Diffusion Prior for Self-Supervised 3D Reconstruction of Neutron Computed Tomography DataICASSP2026paper
Equivariance2Inverse: A Practical Self-Supervised CT Reconstruction Method Benchmarked on Real, Limited-Angle, and Blurred DataTCI2026paper
From Sparse X-rays to 3D CT: Training-Free Reconstruction with Diffusion PriorsDGM4MICCAI2026paper · code
HorusEye: A self-supervised foundation model for generalizable X-ray tomography restorationNat Comput Sci2026paper · code
MedGMAE: Gaussian Masked Autoencoders for Medical Volumetric Representation LearningICLR2026paper · code
Zero-shot CT Super-Resolution using Diffusion-based 2D Projection Priors and Signed 3D GaussiansMICCAI2026paper
NGPS: Structure-Preserving Self-Supervised Denoising via Neighbor-Guided Patch SamplingECCV2026paper · code
Noise2Match: Self-Supervised Low-Dose CT Denoising via Adaptive Inter-Slice Matching预印本2026paper
Unsupervised Metal Artifact Reduction in Dental CBCT using Fine-tuned Cycle-Consistent Adversarial NetworksDigital2026paper
An unsupervised sparse-view CT reconstruction framework using combination of iterative deep image prior and ADMMMed Phys2025paper
Limited-angle SPECT image reconstruction using deep image priorPMB2025paper
Unsupervised Self-Prior Embedding Neural Representation for Sparse-View CT ReconstructionAAAI2025paper · code
Zero-Shot Low-dose CT Denoising via Sinogram Flicking预印本2025paper
Unsupervised CT Metal Artifact Reduction by Plugging Diffusion Priors in Dual DomainsTMI2024paper · code
RBP-DIP: Residual back projection with deep image prior for ill-posed CT reconstructionNeural Netw.2024paper
PINER: Prior-Informed Implicit Neural Representation Learning for Test-Time Adaptation in Sparse-View CT ReconstructionWACV2023paper · code
Unsupervised Polychromatic Neural Representation for CT Metal Artifact ReductionNeurIPS2023paper · code
SemiMAR: Semi-Supervised Learning for CT Metal Artifact ReductionJBHI2023paper · code
Stabilizing deep tomographic reconstruction - Part A. Hybrid framework and experimental resultsPatterns2022paper
Stabilizing deep tomographic reconstruction - Part B. Convergence analysis and adversarial attacksPatterns2022paper
An Educated Warm Start for Deep Image Prior-Based Micro CT ReconstructionTCI2022paper
An Unsupervised Reconstruction Method For Low-Dose CT Using Deep Generative Regularization PriorBSPC2022paper · code
A Dataset-free Deep Learning Method for Low-Dose CT Image ReconstructionInverse Problems2022paper
Noise Characteristics Modeled Unsupervised Network for Robust CT Image ReconstructionTMI2022paper
Sparse-view and limited-angle CT reconstruction with untrained networks and deep image priorCMPB2022paper
CoIL: Coordinate-Based Internal Learning for Tomographic ImagingTCI2021paper
IntraTomo: Self-supervised Learning-based Tomography via Sinogram Synthesis and PredictionICCV2021paper · code
Self-Supervised Training For Low-Dose CT ReconstructionISBI2021paper

在可检索的目录中浏览全部 35 篇「自监督与免训练方法」论文

数据集、基准与工具库

目录中任务或标签与本主题相符的条目。

指南

涵盖本主题的指南。

其他主题