Organizations: University of Bordeaux, CNRS, INRIA, Bordeaux INP, IMB, UMR 5251, F-33405, Talence, France · AIMOKA: Inria, AP-HP, Université Sorbonne Paris Nord, F-35042, Rennes, France
Variational reconstruction provides a rigorous framework for cone-beam computed tomography (CBCT) by combining an explicit acquisition model with image priors. However, iterative optimization remains computationally demanding, particularly under severe noise and sparse-view conditions. Although learning-based methods achieve high reconstruction quality, they often replace the original variational formulation with learned optimization schemes or nonlinear equilibrium models. We propose a neural network-guided warm-start framework that accelerates variational reconstruction while preserving its original formulation. Projection data are first reconstructed using the Feldkamp-Davis-Kress (FDK) algorithm, and the resulting image is refined by an image-domain neural network. The network output is used solely to initialize the subsequent model-based optimization, leaving the forward model, data-fidelity term, regularization, and optimization scheme unchanged. A pre-regularization step further improves compatibility with the variational priors. Experiments on pelvic CT images from the SynthRAD2025 challenge show a more than threefold reduction in reconstruction time while maintaining comparable image quality.
Figures & tables
Figure 1: Typical evolution of the variational energy (a) and SSIM (b) as a function of the number of iterations during variational refinement, for the three initialization strategies considered, with ρ=50 .
Weak regularization ( ρ=10 )
Strong regularization ( ρ=50 )
WS strategy
FDK-WS
NN-WS
Reg-NN-WS
FDK
NN-WS
Reg-NN-WS
N0.95[#]
153
10
3
62
11
2
N0.99[#]
829
167
148
285
108
78
T0.95 [s]
305
20
13
125
23
10
T0.99 [s]
1643
331
300
566
214
163
Table 1: Summary of the number of iterations and computational time required to reach the prescribed SSIM thresholds of 0.95 and 0.99, averaged across the 10 held-out test volumes, for the three initialization strategies and the two values of the sparsity parameter ρ∈{10,50} in Eq. ( 2 ). The cells are color-coded according to their relative values within each group corresponding to a given ρ : green, yellow, and orange indicate the lowest, intermediate, and highest values, respectively.
Figure 2: Typical example of a 2D pelvic image in the coronal plane: (a) reference solution; (b) conventional FDK reconstruction; and (c) approximate solution obtained using the regularized NN warm-start strategy at the iteration at which the reconstruction reached a prescribed SSIM threshold of 0.99, with ρ=10 .
Image reconstruction recovers images, volumes, or scenes from incomplete, noisy, indirect, or compressed measurements using physical models and learned priors.
Cone-beam computed tomography (CBCT) is widely used for image-guided radiotherapy, but reducing the number of projections to lower dose and scan time makes reconstruction increasingly ill-posed. We present EUReCA (End-to-end Unified REconstruction across Clinical Acquisitions), a geometry-aware feed-forward model for variable-view clinical linac CBCT. For a given acquisition trajectory, EUReCA accepts 8-100 projections with a single set of weights, without view-count-specific retraining; the same architecture and training recipe extend unchanged to other clinical trajectories. It combines learned projection features with an in-frame Feldkamp-Davis-Kress (FDK) reconstruction. The model encodes each projection with a shared multiscale backbone, back-projects features into multiresolution 3D grids using the acquisition geometry, aggregates variable numbers of views at fixed computational cost, and refines the volume with a global 3D transformer and multilevel decoder. On simulated data, EUReCA achieved 29.9-33.2 dB PSNR with 10-100 views and outperformed all applicable feed-forward baselines; it also exceeded iterative optimization-based methods while running orders of magnitude faster. On 15 measured patient CBCT scans evaluated against vendor reconstructions, EUReCA achieved the best performance at every tested view count. These results suggest that geometry-aware, measurement-anchored feed-forward reconstruction can provide a practical and flexible foundation for variable-view linac CBCT.
Jiening Zhu, Chengzhu Zhang, Jason Fan +8
The authors are with the Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY 10065 USA
A long-term goal in CT imaging is to achieve fast and accurate 3D reconstruction from sparse-view projections, thereby reducing radiation exposure, lowering system cost, and enabling timely imaging in clinical workflows. Recent feed-forward approaches have shown strong potential toward this overarching goal, yet their results still suffer from artifacts and loss of fine details. In this work, we introduce Iterative Latent Volumes (ILV), a feed-forward framework that integrates data-driven priors with classical iterative reconstruction principles to overcome key limitations of prior feed-forward models in sparse-view CBCT reconstruction. At its core, ILV constructs an explicit 3D latent volume that is repeatedly updated by conditioning on multi-view X-ray features and the learned anatomical prior, enabling the recovery of fine structural details beyond the reach of prior feed-forward models. In addition, we develop and incorporate several key architectural components, including an X-ray feature volume, group cross-attention, efficient self-attention, and view-wise feature aggregation, that efficiently realize its core latent volume refinement concept. Extensive experiments on a large-scale dataset of approximately 14,000 CT volumes demonstrate that ILV significantly outperforms existing feed-forward and optimization-based methods in both reconstruction quality and speed. These results show that ILV enables fast and accurate sparse-view CBCT reconstruction suitable for clinical use. The project page is available at: https://sngryonglee.github.io/ILV/.
Generative models, particularly Diffusion Models (DM), have shown strong potential for Computed Tomography (CT) reconstruction serving as expressive priors for solving ill-posed inverse problems. However, diffusion-based reconstruction relies on Stochastic Differential Equations (SDEs) for forward diffusion and reverse denoising, where such stochasticity can interfere with repeated data consistency corrections in CT reconstruction. Since CT reconstruction is often time-critical in clinical and interventional scenarios, improving reconstruction efficiency is essential. In contrast, Flow Matching (FM) models sampling as a deterministic Ordinary Differential Equation (ODE), yielding smooth trajectories without stochastic noise injection. This deterministic formulation is naturally compatible with repeated data consistency operations. Furthermore, we observe that FM-predicted velocity fields exhibit strong correlations across adjacent steps. Motivated by this, we propose an FM-based CT reconstruction framework (FMCT) and an efficient variant (EFMCT) that reuses previously predicted velocity fields over consecutive steps to substantially reduce the number of Neural network Function Evaluations (NFEs), thereby improving inference efficiency. We provide theoretical analysis showing that the error introduced by velocity reuse is bounded when combined with data consistency operations. Extensive experiments demonstrate that FMCT/EFMCT achieve competitive reconstruction quality while significantly improving computational efficiency compared with diffusion-based methods. The codebase is open-sourced at https://github.com/EFMCT/EFMCT.
Jiayang Shi, Lincen Yang, Zhong Li +3
Centrum Wiskunde en Informatica · LIACS, Leiden University · Great Bay University +1