IEEE Transactions on Computational Imaging最新文献

{"title":"Data-Driven Multi-keV Virtual Monoenergetic Images Generation From Single-Energy CT Guided by Image-Domain Material Decomposition","authors":"Wenwen Zhang;Zihan Chai;Yantao Niu;Zhijie Zhang;Linxuan Li;Baohua Sun;Junfang Xian;Wei Zhao","doi":"10.1109/TCI.2026.3653309","DOIUrl":"https://doi.org/10.1109/TCI.2026.3653309","url":null,"abstract":"Virtual monoenergetic images (VMIs), reconstructed from dual-energy CT (DECT) by capturing photon attenuation data at two distinct energy levels, can reduce beam-hardening artifacts and provide more quantitatively accurate attenuation measurements. Data-driven deep learning approaches have demonstrated the feasibility of synthesizing VMIs from conventional single-energy CT (SECT) scans. However, the lack of incorporation of physics-related information in such methods compromises their interpretability and robustness. Here we propose a novel hybrid data-driven framework that synergizes convolutional neural networks with physics-based material decomposition derived from DECT principles. This approach directly yields high-quality VMIs across various keV levels from SECT acquisitions. Through rigorous validation on 130 clinical cases spanning diverse anatomical regions and pathological conditions, our method demonstrates significant improvements over conventional purely data-driven approaches, as evidenced by enhanced anatomical visualization and superior performance on quantitative metrics. By eliminating dependence on DECT hardware while maintaining computational efficiency and incorporating physics-guided constraints, our framework leverages the widespread availability of SECT to provide a cost-effective, high-performance solution for diagnostic imaging in routine clinical practice.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"321-333"},"PeriodicalIF":4.8,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146026483","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Scan-Adaptive MRI Undersampling Using Neighbor-Based Optimization (SUNO)","authors":"Siddhant Gautam;Angqi Li;Nicole Seiberlich;Jeffrey A. Fessler;Saiprasad Ravishankar","doi":"10.1109/TCI.2026.3653330","DOIUrl":"https://doi.org/10.1109/TCI.2026.3653330","url":null,"abstract":"Accelerated MRI aims to reduce scan time by acquiring data more efficiently, for example, through optimized pulse sequences or readouts that increase <inline-formula><tex-math>$k$</tex-math></inline-formula>-space coverage per excitation (e.g., echo planar imaging), or by collecting partial <inline-formula><tex-math>$k$</tex-math></inline-formula>-space measurements with advanced reconstruction methods. Acceleration via partial <inline-formula><tex-math>$k$</tex-math></inline-formula>-space acquisition (i.e., undersampling) has received significant attention, particularly with the rise of learning-based reconstruction methods. Recent works have explored population-adaptive sampling patterns learned from groups of patients (or scans), which enhance sampling pattern design by tailoring it to dataset-specific characteristics, rather than relying on generic approaches. Building on this idea, sampling techniques can be further personalized down to the level of individual scans, enabling the capture of subject- or slice-specific details that may be overlooked in population-based designs. To address this challenging problem, we propose a framework for jointly learning scan-adaptive Cartesian undersampling patterns and a corresponding reconstruction model from a training set, enabling more tailored sampling for individual scans. We use an alternating algorithm for learning the sampling patterns and the reconstruction model where we use an iterative coordinate descent (ICD) based offline optimization of scan-adaptive <inline-formula><tex-math>$k$</tex-math></inline-formula>-space sampling patterns for each example in the training set. A nearest neighbor search is then used to select the scan-adaptive sampling pattern at test time from initially acquired low-frequency <inline-formula><tex-math>$k$</tex-math></inline-formula>-space information. We applied the proposed framework (dubbed SUNO) to the fastMRI multi-coil knee and brain datasets, demonstrating improved performance over the currently used undersampling patterns at both <inline-formula><tex-math>$4times$</tex-math></inline-formula> and <inline-formula><tex-math>$8times$</tex-math></inline-formula> acceleration factors in terms of both visual quality and quantitative metrics.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"601-613"},"PeriodicalIF":4.8,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147440687","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Learning Degradation-Aware Diffusion Prior for Hyperspectral Reconstruction From RGB Image","authors":"Jingxiang Yang;Haifeng Xu;Heyuan Yin;Hongyi Liu;Liang Xiao","doi":"10.1109/TCI.2025.3650359","DOIUrl":"https://doi.org/10.1109/TCI.2025.3650359","url":null,"abstract":"Hyperspectral image (HSI) is applicable in many fields due to the ability in discriminating different materials. Collecting HSI usually requires expensive hardware and long period. Reconstructing HSI from RGB image, also called spectral super-resolution (SSR), is an affordable and feasible way for HSI acquisition. Despite the SSR results achieved by existing deep unfolding networks (DUNs), they still face challenges in: 1) recovering the fine-grained and realistic details; 2) suppressing the spectral distortion. Diffusion model has advantages in generating diverse and realistic contents, while its fidelity is limited due to the inherent randomness. In this study, to reconstruct a faithful and realistic HSI, we integrate the diffusion model in DUN, and propose a degradation-aware unrolling diffusion model for SSR (deDiff-SSR). The generative diffusion prior is jointly leveraged with the spectral degradation and deep prior learning. Specifically, we first pre-train a channel attention enhanced denoising diffusion probabilistic model (DDPM), the spectral correlation is exploited for learning the diffusion prior of HSI. To aware the degradation, by optimizing a diffusion and deep priors regularized HSI SSR model, we propose a degradation-aware diffusion sampling method, the spectral degradation is learned to refine each diffusion sampling step. Via unrolling the degradation-aware diffusion sampling steps, we build the deDiff-SSR network. It contains diffusion and deep proximal operators to represent the diffusion and deep priors, respectively. We implement the diffusion proximal operator with one sampling step of the pre-trained DDPM. Moreover, we design a state-space Transformer as the deep proximal operator, the spectral-spatial long-range relationship of HSI can be efficiently captured. The experiments on several indoor and remote sensing datasets demonstrate the effectiveness of deDiff-SSR.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"256-269"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145982258","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Analysis and Correction of Spatial Variation in Complex MEO-Airborne Bistatic SAR Configurations","authors":"Yu Ma;Yong Wang","doi":"10.1109/TCI.2026.3675488","DOIUrl":"https://doi.org/10.1109/TCI.2026.3675488","url":null,"abstract":"The Medium-Earth Orbit(MEO)-airborne bistatic synthetic aperture radar (SAR) system provides advantages such as high transmitter altitude, strong survivability, and high observation flexibility. However, the inherent characteristics of this system result in spatial variation in the slant range history of different scattering points within the observed area. This paper presents the qualitative assessment on the two-dimensional (2-D) spatial variation caused by the system geometric configuration firstly, and the accurate slant range history model is established followed by the proposal of a virtual scattering points fitting method to analyze the impact of 2-D spatial variation in SAR imaging procedure quantitatively. Then, the correction algorithm is proposed to compensate the spatial variation in the range dimension through the block processing and mitigate the azimuth dimension variation by the nonlinear chirp scaling(NCS) method. Finally, the experimental results with simulated and semi-real data validate the generalizability and practicality of the proposed approach.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"702-717"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147606180","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SV-FMT-DualDiff: Dual-Path Conditional Diffusion for Single-View Tomography","authors":"Ruxin Cai;Huijie Wu;Zeyu Liu;Jiacheng Liu;Haoran Guo;Bo Wen;Guanglei Zhang","doi":"10.1109/TCI.2026.3680316","DOIUrl":"https://doi.org/10.1109/TCI.2026.3680316","url":null,"abstract":"Fluorescence Molecular Tomography (FMT) is an essential imaging technique for in vivo 3D biomolecular visualization. However, its application for rapid tomography generally remains unexplored since rapid imaging will inevitably require sparse projections, which makes the inverse problem extremely ill-posed and consequently causes traditional methods to fail. To overcome this significant challenge, this paper presents Single-View Fluorescence Molecular Tomography based on Dual-Path conditional Diffusion (SV-FMT-DualDiff), a novel paradigm that integrates the conditional diffusion denoising prior with physics-consistency for high-fidelity and rapid single-view reconstructions. For effective conditional control, the Tomographic Lifting Representation (TLR) enhances 3D volumetric conditioning information extracted from 2D projection features. Besides, a dual-path conditioning mechanism synergistically combines Local Guidance with Multi-feature Fusion (LGMF) and Global Guidance with Discrepancy-Common transformer (GGDC) to guide the denoising process. LGMF enhances local fine-grained detail and temporal awareness, while GGDC aligns noise with conditioning features before interaction to reinforce global correlations. Extensive numerical and in vivo experiments demonstrate that our method outperforms state-of-the-art techniques, delivering exceptional spatial resolution, robustness, and generalization capability. In dual-target reconstruction with a 0.5 mm edge-to-edge distance across different target locations, our method achieves CNR of 25.03 ± 0.74, Dice of 0.88 ± 0.01, LE of 0.026 ± 0.005 cm, and NMSE of 0.23 ± 0.01. This work offers a reliable solution for rapid 3D fluorescent imaging in biomedical research, such as tumor studies and drug development. Furthermore, it shows strong potential for extension to other imaging modalities by addressing severely ill-posed sparse-angle tomography for rapid, high-quality reconstruction.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"747-760"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147696684","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An ISAR Imaging Method for GEO Satellites With Reduced Number of Estimated Parameters","authors":"Shuang Yue;Dong Feng;Yixuan Song;Jian Wang;Chongyi Fan;Xiaotao Huang","doi":"10.1109/TCI.2026.3681440","DOIUrl":"https://doi.org/10.1109/TCI.2026.3681440","url":null,"abstract":"High-resolution inverse synthetic aperture radar (ISAR) imaging of geosynchronous orbit (GEO) satellites with small-to-medium inclination necessitates a long coherent processing interval (CPI) to satisfy the integration angle requirement. However, an extended CPI leads to more complex target motion, thereby increasing the number of motion parameters to be estimated for phase compensation. To overcome this constraint, this paper integrates the prior motion information of GEO satellites and the particle swarm optimization (PSO) algorithm to propose a novel ISAR imaging method. First, the prior motion information of the GEO satellite is derived based on the orbital model, and a parametric global phase compensation function is constructed accordingly. The parameters in this function represent the deviation coefficients introduced by the discrepancy between the satellite's theoretical and actual motion. Subsequently, the PSO algorithm is employed to estimate these parameters. Finally, the spatial-time-variant (STV) phase compensation is implemented based on the obtained parameters, achieving well-focused imaging. Compared with conventional techniques, the proposed method does not require a high-order Taylor series expansion of the motion model, which can effectively reduce the number of parameters to be estimated for phase compensation. Furthermore, by exploiting prior motion information, it enhances the feasibility of neglecting the impact of higher-order phase on imaging, thereby further reducing the required phase compensation order. Processing results of both simulation data and measured data from a scaled satellite model are provided to validate the effectiveness of the proposed method.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"812-826"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147796209","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Feature Decoupling Through Random Sampling for High-Fidelity Flow Matching-Based Artifact Removal","authors":"Menglei Li;Genwei Ma;Yining Zhu;Xing Zhao","doi":"10.1109/TCI.2026.3671418","DOIUrl":"https://doi.org/10.1109/TCI.2026.3671418","url":null,"abstract":"This paper introduces a novel framework for high-fidelity ring artifact removal in computed tomography (CT) by integrating projection-domain localization and image-domain refinement. Traditional methods struggle to balance artifact suppression with structural preservation due to the global nature of ring artifacts originating from detector response inconsistencies. To address this, we propose a dual-domain approach: First, a random sampling strategy in the projection domain decouples global artifacts into localized features. This involves neighborhood-based pixel shuffling with Gaussian-distributed randomness and continuity constraints, effectively disrupting vertical streaks while minimizing discontinuities. The process generates auxiliary projections (maximum, minimum, average) to supplement lost information during shuffling. Second, an image-domain flow matching model refines reconstructions by learning conditional probability paths between artifact-corrupted inputs and clean outputs. Experiments on the AAPM Mayo dataset demonstrate superior performance, achieving a PSNR of 32.952 dB and RMSE of 38.222 HU, outperforming traditional methods (e.g. wavelet-FFT, total variation) and network baselines by 2 dB.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"660-672"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147557508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multivariate Fields of Experts for Convergent Image Reconstruction","authors":"Stanislas Ducotterd;Michael Unser","doi":"10.1109/TCI.2026.3685412","DOIUrl":"https://doi.org/10.1109/TCI.2026.3685412","url":null,"abstract":"We introduce the multivariate fields of experts, a new framework for the learning of image priors. Our model generalizes existing fields of experts methods by incorporating multivariate potential functions constructed via Moreau envelopes of the <inline-formula><tex-math>$ell _infty$</tex-math></inline-formula>-norm. We demonstrate the effectiveness of our proposal across a range of inverse problems that include image denoising, deblurring, compressed-sensing magnetic-resonance imaging, and computed tomography. The proposed approach outperforms comparable univariate models and achieves performance close to that of deep-learning-based regularizers while being significantly faster, requiring fewer parameters, and being trained on substantially fewer data. In addition, our model retains a high level of interpretability due to its structured design. It is supported by theoretical convergence guarantees which ensure reliability in sensitive reconstruction tasks.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"827-838"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11487951","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147828890","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ADOBI: Adaptive Diffusion Bridge for Blind Inverse Problems With Application to MRI Reconstruction","authors":"Yuyang Hu;Albert Peng;Weijie Gan;Ulugbek S. Kamilov","doi":"10.1109/TCI.2026.3678407","DOIUrl":"https://doi.org/10.1109/TCI.2026.3678407","url":null,"abstract":"Diffusion bridges (DB) have emerged as a promising alternative to diffusion models for imaging inverse problems, achieving faster sampling by directly bridging low- and high-quality image distributions. While incorporating measurement consistency has been shown to improve performance, existing DB methods fail to maintain this consistency in blind inverse problems, where the forward model is unknown. To address this limitation, we introduce ADOBI (Adaptive Diffusion Bridge for Inverse Problems), a novel framework that adaptively calibrates the unknown forward model to enforce measurement consistency throughout sampling iterations. Our adaptation strategy allows ADOBI to achieve high-quality parallel magnetic resonance imaging (PMRI) reconstruction in only 5–10 steps. Our numerical results show that ADOBI consistently delivers state-of-the-art performance, and further advances the Pareto frontier for the perception-distortion trade-off.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"776-786"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147736996","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Equivariance2Inverse: A Practical Self-Supervised CT Reconstruction Method Benchmarked on Real, Limited-Angle, and Blurred Data","authors":"Dirk Elias Schut;Adriaan Graas;Robert van Liere;Tristan van Leeuwen","doi":"10.1109/TCI.2026.3684416","DOIUrl":"https://doi.org/10.1109/TCI.2026.3684416","url":null,"abstract":"Deep learning has shown impressive results in reducing noise and artifacts in X-ray computed tomography (CT) reconstruction. Self-supervised CT reconstruction methods are especially appealing for real-world applications because they require no ground truth training examples. However, these methods involve a simplified X-ray physics model during training, which may make inaccurate assumptions, for example, about scintillator blurring, the scanning geometry, or the distribution of the noise. As a result, they can be less robust to real-world imaging circumstances. In this paper, we review the model assumptions of six recent self-supervised CT reconstruction methods. Based on this, we combined concepts of the Robust Equivariant Imaging and Sparse2Inverse methods in a new self-supervised CT reconstruction method called Equivariance2Inverse that is robust to scintillator blurring and limited-angle data. We benchmarked Equivariance2Inverse and the existing methods on the real-world 2DeteCT dataset and on synthetic data with and without scintillator blurring and a limited-angle scanning geometry. The results of our benchmark show that methods that assume that the noise is pixel-wise independent do not perform well on data with scintillator blurring. Moreover, they show that when the distribution of objects is rotationally invariant, this invariance can be used to reduce artifacts in limited-angle reconstructions.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"12 ","pages":"800-811"},"PeriodicalIF":4.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147796100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}