pith. sign in

arxiv: 2606.24561 · v1 · pith:LPQ44HIVnew · submitted 2026-06-23 · 💻 cs.CV

Quantum CT via Dynamic Interval Encoding and Prior-Balanced QUBO Reconstruction

Ao Wang , Yikuang Yuluo , Yujie Liu , Shuangyang Zhong , Yuwen Zhang , Zihao Wang , Fenglin Liu , Andreas Maier
show 2 more authors
Haijun Yu Yixing Huang
This is my paper

Pith reviewed 2026-06-26 00:36 UTC · model grok-4.3

classification 💻 cs.CV
keywords quantum computed tomographyQUBOdynamic interval encodingsparse-view CTlimited-angle CTquantum annealingimage reconstructionprior balancing
0
0 comments X

The pith

Dynamic local gray-level encoding in QUBO enables more faithful grayscale CT reconstruction from sparse and limited-angle views.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper presents a QUBO formulation for quantum CT that replaces fixed global bit-plane encodings with dynamic interval encoding, restricting binary variables to local gray-level intervals around the current pixel estimate in each refinement round. A boundary-hit-guided rule decides whether to expand the search or refine locally, while a prior-balancing step weights projection data consistency against an edge-preserving quadratic term before the QUBO is formed. Sparse-view and limited-angle fan-beam experiments demonstrate that the resulting reconstructions recover both structural details and intensity distributions more accurately than analytic, iterative, variational, and representation-based baselines. The authors attribute the gains primarily to improved gray-level expressivity and more stable data-prior coupling rather than to the choice of solver. The formulation is shown to run on a hybrid quantum-classical backend.

Core claim

The paper establishes that encoding active pixels only inside adaptive local gray-level intervals, steered by boundary-hit updates, and balancing the data-fidelity term with an edge-preserving prior before constructing the QUBO produces grayscale CT images whose structures and gray-level values match ground truth more closely than those obtained from analytic, iterative, variational, or learned baselines under sparse-view and limited-angle fan-beam conditions.

What carries the argument

Dynamic interval encoding, which restricts each pixel's binary variables to a local gray-level window around the running estimate and uses boundary hits to toggle between expansion and refinement inside a prior-balanced QUBO.

If this is right

  • The method recovers both structures and gray-level distributions more faithfully than the tested analytic, iterative, variational, and representation-based methods.
  • The formulation executes on hardware-backed hybrid quantum-classical solvers such as the D-Wave BQM.
  • Ablation and expressivity analyses indicate that the performance lift stems mainly from the local encoding strategy and the balanced data-prior coupling.
  • The approach remains executable on current hybrid solvers without requiring fully fault-tolerant quantum hardware.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The variable budget per pixel can be kept smaller than global encodings while still supporting high gray-level precision, potentially lowering the qubit or coupling requirements for larger images.
  • The same local-interval mechanism may transfer to other inverse problems that currently suffer from fixed-precision quantization error in binary optimization formulations.
  • Further scaling could be tested by replacing the hybrid solver with newer quantum annealing or gate-model backends to measure wall-clock or solution-quality changes.

Load-bearing premise

The boundary-hit update rule and prior-balancing step produce stable convergence without introducing quantization or optimization artifacts that offset the reported fidelity gains.

What would settle it

A controlled experiment on the same sparse-view and limited-angle fan-beam phantoms in which the proposed reconstructions exhibit lower structural similarity or intensity accuracy metrics, or visible new artifacts, relative to the evaluated baselines.

Figures

Figures reproduced from arXiv: 2606.24561 by Andreas Maier, Ao Wang, Fenglin Liu, Haijun Yu, Shuangyang Zhong, Yikuang Yuluo, Yixing Huang, Yujie Liu, Yuwen Zhang, Zihao Wang.

Figure 1
Figure 1. Figure 1: Overview of the proposed framework: (a) dynamic interval encoding versus fixed global encoding under a 2-bit budget, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Reference images 512×512 and their 32×32 versions used to generate projections and compute metrics: Shepp– Logan (a,e), American Association of Physicists in Medicine (AAPM) CT slices (b,f) and (c,g), and Kaggle head CT (d,h). dataset [29], and one head CT slice from the Kaggle dataset Computed Tomography (CT) of the Brain [30]. The first row of [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Sparse-view reconstruction results. For each test image, the top row shows reconstructions and the bottom row shows [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Limited-angle reconstruction on the phantom with 18 [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Fixed global encoding versus dynamic interval encod [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Component ablation. Columns show no prior, no [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: Convergence curves for different rinit values. Interme￾diate half-widths converge faster and yield better final PSNR, SSIM, and NRMSE [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Larger-image sparse-view (20 views) reconstruction [PITH_FULL_IMAGE:figures/full_fig_p009_10.png] view at source ↗
read the original abstract

Quadratic unconstrained binary optimization (QUBO)-based quantum computed tomography (CT) casts reconstruction as a binary quadratic problem for quantum annealing and hybrid quantum--classical solvers. For grayscale CT, however, image encoding is constrained by the binary-variable budget: fixed global bit-plane encodings increase QUBO size and coupling complexity as gray-level precision improves, whereas low-bit encodings introduce quantization error. We propose a QUBO-based grayscale CT reconstruction framework that combines dynamic interval encoding with prior-balanced optimization. Each refinement round encodes active pixels only within local gray-level intervals around the current estimate, and a boundary-hit-guided update rule adaptively switches between search expansion and local refinement. To improve optimization stability, the method balances projection-domain data consistency and an edge-preserving quadratic prior before forming the final QUBO. Sparse-view and limited-angle fan-beam CT experiments show that the proposed method recovers structures and gray-level distributions more faithfully than the evaluated analytic, iterative, variational, and representation-based baselines. Expressivity analysis and ablation studies further indicate that the improvement mainly arises from effective gray-level representation through dynamic local encoding and more stable data-fidelity--prior coupling. Experiments on the D-Wave hybrid binary quadratic model (BQM) solver further demonstrate that the formulation is executable on a hardware-backed hybrid quantum--classical backend.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The manuscript proposes a QUBO-based framework for grayscale CT reconstruction that uses dynamic interval encoding (encoding active pixels only within local gray-level intervals around the current estimate) together with a boundary-hit-guided update rule and prior-balanced optimization (balancing projection-domain data consistency against an edge-preserving quadratic prior). It claims that this yields more faithful recovery of structures and gray-level distributions than analytic, iterative, variational, and representation-based baselines on sparse-view and limited-angle fan-beam CT experiments, attributes the gains primarily to the encoding and coupling via expressivity analysis and ablations, and demonstrates executability on the D-Wave hybrid BQM solver.

Significance. If the empirical claims and attribution to the proposed mechanisms hold, the work would offer a concrete route to mitigate the binary-variable budget constraint in QUBO formulations of grayscale inverse problems, potentially enabling practical hybrid quantum-classical CT pipelines. The explicit ablation studies and hardware-backed demonstration constitute strengths that could support reproducibility and adoption.

major comments (2)
  1. [Method (boundary-hit-guided update rule and prior-balancing)] The boundary-hit-guided update rule is presented as an adaptive heuristic that switches between search expansion and local refinement, yet no derivation from first principles, convergence analysis, or explicit check against new quantization/optimization artifacts is supplied (see the method description of the update rule and prior-balancing step). This is load-bearing for the central claim that observed gains arise from encoding stability and data-fidelity–prior coupling rather than implicit regularization or solver-specific effects.
  2. [Abstract / Experiments] The abstract asserts qualitative superiority in structure and gray-level recovery on fan-beam experiments but supplies no quantitative metrics, error bars, dataset sizes, or numerical ablation results; without these the central empirical claim cannot be verified at the level required for a journal in the field.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We respond to each major comment below, indicating planned revisions where appropriate.

read point-by-point responses

  1. Referee: [Method (boundary-hit-guided update rule and prior-balancing)] The boundary-hit-guided update rule is presented as an adaptive heuristic that switches between search expansion and local refinement, yet no derivation from first principles, convergence analysis, or explicit check against new quantization/optimization artifacts is supplied (see the method description of the update rule and prior-balancing step). This is load-bearing for the central claim that observed gains arise from encoding stability and data-fidelity–prior coupling rather than implicit regularization or solver-specific effects.

    Authors: We acknowledge that the boundary-hit-guided update rule is presented as a heuristic without a formal derivation from first principles or convergence analysis in the current manuscript. The attribution of gains to the encoding and prior-balancing mechanisms is grounded in the expressivity analysis and ablation studies. To address the concern, we will expand the revised manuscript with additional discussion of the heuristic rationale and an explicit examination of potential quantization and optimization artifacts. revision: partial

  2. Referee: [Abstract / Experiments] The abstract asserts qualitative superiority in structure and gray-level recovery on fan-beam experiments but supplies no quantitative metrics, error bars, dataset sizes, or numerical ablation results; without these the central empirical claim cannot be verified at the level required for a journal in the field.

    Authors: We agree that the abstract would be strengthened by the inclusion of quantitative metrics. In the revision we will update the abstract to report key numerical results from the experiments, including average PSNR and SSIM values with error bars, dataset sizes, and references to the ablation studies. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected in the proposed QUBO CT framework

full rationale

The paper presents an algorithmic framework for QUBO-based grayscale CT reconstruction that combines dynamic interval encoding, a boundary-hit-guided update rule, and prior balancing. Performance claims are supported by sparse-view and limited-angle experiments plus ablation studies, without any derivation chain that reduces by construction to fitted inputs, self-definitions, or load-bearing self-citations. The method builds on standard QUBO and CT priors as external foundations, and no step equates a claimed result to its own inputs via renaming, ansatz smuggling, or uniqueness theorems imported from the authors' prior work.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies insufficient detail to enumerate free parameters or invented entities; the method appears to rest on standard QUBO formulation and conventional edge-preserving priors.

pith-pipeline@v0.9.1-grok · 5795 in / 1108 out tokens · 15381 ms · 2026-06-26T00:36:14.884095+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

33 extracted references

  1. [1]

    Quantum annealing in the transverse Ising model,

    T. Kadowaki and H. Nishimori, “Quantum annealing in the transverse Ising model,”Physical Review E, vol. 58, no. 5, pp. 5355–5363, 1998

    1998

  2. [2]

    Ising formulations of many NP problems,

    A. Lucas, “Ising formulations of many NP problems,”Frontiers in Physics, vol. 2, p. 5, 2014

    2014

  3. [3]

    Exploring the limitations of hybrid adiabatic quantum computing for emission tomography reconstruction,

    M. A. Nau, A. H. Vija, W. Gohn, M. P. Reymann, and A. K. Maier, “Exploring the limitations of hybrid adiabatic quantum computing for emission tomography reconstruction,”Journal of Imaging, vol. 9, no. 10, p. 221, 2023

    2023

  4. [4]

    Quantum annealing-based computed tomography using vari- ational approach for a real-number image reconstruction,

    A. Haga, “Quantum annealing-based computed tomography using vari- ational approach for a real-number image reconstruction,”Physics in Medicine & Biology, vol. 69, no. 4, p. 04NT02, 2024

    2024

  5. [5]

    A highly accurate quantum optimization algorithm for CT image reconstruction based on sinogram patterns,

    K. Jun, “A highly accurate quantum optimization algorithm for CT image reconstruction based on sinogram patterns,”Scientific Reports, vol. 13, p. 14407, 2023

    2023

  6. [6]

    Utilizing quantum annealing in computed tomography image reconstruction,

    K. Dremel, D. Prjamkov, M. Firsching, M. Weule, T. Lang, A. Papadaki, S. Kasperl, M. Blaimer, and T. O. J. Fuchs, “Utilizing quantum annealing in computed tomography image reconstruction,”IEEE Transactions on Quantum Engineering, vol. 6, p. 3100810, 2025

    2025

  7. [7]

    Quantum optimization algorithms for CT image segmentation from X-ray data,

    K. Jun and H. Lee, “Quantum optimization algorithms for CT image segmentation from X-ray data,”Scientific Reports, vol. 15, p. 20649, 2025

    2025

  8. [8]

    A. C. Kak and M. Slaney,Principles of Computerized Tomographic Imaging. New York, NY , USA: IEEE Press, 1988

    1988

  9. [9]

    Practical cone-beam algorithm,

    L. A. Feldkamp, L. C. Davis, and J. W. Kress, “Practical cone-beam algorithm,”Journal of the Optical Society of America A, vol. 1, no. 6, pp. 612–619, 1984

    1984

  10. [10]

    Algebraic reconstruction techniques (ART) for three-dimensional electron microscopy and X-ray photography,

    R. Gordon, R. Bender, and G. T. Herman, “Algebraic reconstruction techniques (ART) for three-dimensional electron microscopy and X-ray photography,”Journal of Theoretical Biology, vol. 29, no. 3, pp. 471– 481, 1970

    1970

  11. [11]

    Simultaneous algebraic reconstruction technique (SART): A superior implementation of the ART algorithm,

    A. H. Andersen and A. C. Kak, “Simultaneous algebraic reconstruction technique (SART): A superior implementation of the ART algorithm,” Ultrasonic Imaging, vol. 6, no. 1, pp. 81–94, 1984

    1984

  12. [12]

    Methods of conjugate gradients for solving linear systems,

    M. R. Hestenes and E. Stiefel, “Methods of conjugate gradients for solving linear systems,”Journal of Research of the National Bureau of Standards, vol. 49, no. 6, pp. 409–436, 1952

    1952

  13. [13]

    Accelerated projection methods for com- puting pseudoinverse solutions of systems of linear equations,

    A. Bjorck and T. Elfving, “Accelerated projection methods for com- puting pseudoinverse solutions of systems of linear equations,”BIT Numerical Mathematics, vol. 19, no. 2, pp. 145–163, 1979

    1979

  14. [14]

    Nonlinear total variation based noise removal algorithms,

    L. I. Rudin, S. Osher, and E. Fatemi, “Nonlinear total variation based noise removal algorithms,”Physica D: Nonlinear Phenomena, vol. 60, no. 1–4, pp. 259–268, 1992

    1992

  15. [15]

    Compressed sensing,

    D. L. Donoho, “Compressed sensing,”IEEE Transactions on Informa- tion Theory, vol. 52, no. 4, pp. 1289–1306, 2006

    2006

  16. [16]

    Image reconstruction in circular cone-beam computed tomography by constrained, total-variation minimization,

    E. Y . Sidky and X. Pan, “Image reconstruction in circular cone-beam computed tomography by constrained, total-variation minimization,” Physics in Medicine & Biology, vol. 53, no. 17, pp. 4777–4807, 2008

    2008

  17. [17]

    A first-order primal-dual algorithm for convex problems with applications to imaging,

    A. Chambolle and T. Pock, “A first-order primal-dual algorithm for convex problems with applications to imaging,”Journal of Mathematical Imaging and Vision, vol. 40, no. 1, pp. 120–145, 2011

    2011

  18. [18]

    Deep con- volutional neural network for inverse problems in imaging,

    K. H. Jin, M. T. McCann, E. Froustey, and M. Unser, “Deep con- volutional neural network for inverse problems in imaging,”IEEE Transactions on Image Processing, vol. 26, no. 9, pp. 4509–4522, 2017

    2017

  19. [19]

    Framing U-Net via deep convolutional framelets: Application to sparse-view CT,

    Y . Han and J. C. Ye, “Framing U-Net via deep convolutional framelets: Application to sparse-view CT,”IEEE Transactions on Medical Imaging, vol. 37, no. 6, pp. 1418–1429, 2018

    2018

  20. [20]

    Learned primal-dual reconstruction,

    J. Adler and O. ¨Oktem, “Learned primal-dual reconstruction,”IEEE Transactions on Medical Imaging, vol. 37, no. 6, pp. 1322–1332, 2018

    2018

  21. [21]

    Deep learning computed tomography: Learning projection-domain weights from image domain in limited angle problems,

    T. W ¨urfl, M. Hoffmann, V . Christlein, K. Breininger, Y . Huang, M. Unberath, and A. K. Maier, “Deep learning computed tomography: Learning projection-domain weights from image domain in limited angle problems,”IEEE Transactions on Medical Imaging, vol. 37, no. 6, pp. 1454–1463, 2018

    2018

  22. [22]

    LEARN: Learned experts’ assessment- based reconstruction network for sparse-data CT,

    H. Chen, Y . Zhang, Y . Chen, J. Zhang, W. Zhang, H. Sun, Y . Lv, P. Liao, J. Zhou, and G. Wang, “LEARN: Learned experts’ assessment- based reconstruction network for sparse-data CT,”IEEE Transactions on Medical Imaging, vol. 37, no. 6, pp. 1333–1347, 2018

    2018

  23. [23]

    Deep image prior,

    D. Ulyanov, A. Vedaldi, and V . Lempitsky, “Deep image prior,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 9446–9454

    2018

  24. [24]

    Unsupervised self-prior embedding neural representation for iterative sparse-view CT reconstruction,

    X. Tian, L. Chen, Q. Wu, C. Du, J. Shi, H. Wei, and Y . Zhang, “Unsupervised self-prior embedding neural representation for iterative sparse-view CT reconstruction,”Proceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 7, pp. 7383–7391, 2025

    2025

  25. [25]

    3DGR-CT: Sparse- view CT reconstruction with a 3D gaussian representation,

    Y . Li, X. Fu, H. Li, S. Zhao, R. Jin, and S. K. Zhou, “3DGR-CT: Sparse- view CT reconstruction with a 3D gaussian representation,”Medical Image Analysis, vol. 103, p. 103585, 2025

    2025

  26. [26]

    Discretized gaussian representation for tomographic reconstruction,

    S. Wu, Y . Lu, Y . Guo, W. Ji, S. Huang, F. Yang, S. Sirejiding, Q. He, J. Tong, Y . Ji, Y . Ding, and H. Lu, “Discretized gaussian representation for tomographic reconstruction,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2025, pp. 25 073–25 082

    2025

  27. [27]

    Quantum medical imaging algorithms,

    B. T. Kiani, A. Villanyi, and S. Lloyd, “Quantum medical imaging algorithms,” 2020

    2020

  28. [28]

    Quantum compressed sensing to- mographic reconstruction algorithm,

    A. Ryou, K. Kim, and K. Jun, “Quantum compressed sensing to- mographic reconstruction algorithm,”IEEE Transactions on Quantum Engineering, 2026, early Access

    2026

  29. [29]

    Low-dose CT image and projection dataset,

    T. R. Moen, B. Chen, D. R. Holmes III, X. Duan, Z. Yu, L. Yu, S. Leng, J. G. Fletcher, and C. H. McCollough, “Low-dose CT image and projection dataset,”Medical Physics, vol. 48, no. 2, pp. 902–911, 2021

    2021

  30. [30]

    Computed tomography (CT) of the brain,

    TrainingDataPro, “Computed tomography (CT) of the brain,” Kaggle dataset, 2023, accessed: 2026-06-21. [Online]. Available: https: //www.kaggle.com/datasets/trainingdatapro/computed-tomography-ct-o f-the-brain

    2023

  31. [31]

    Fast and flexible X-ray tomography using the ASTRA toolbox,

    W. van Aarle, W. J. Palenstijn, J. De Beenhouwer, T. Altantzis, S. Bals, K. J. Batenburg, and J. Sijbers, “Fast and flexible X-ray tomography using the ASTRA toolbox,”Optics Express, vol. 24, no. 22, pp. 25 129– 25 147, 2016

    2016

  32. [32]

    D-Wave Hybrid Solver Service: An overview,

    D-Wave Systems Inc., “D-Wave Hybrid Solver Service: An overview,” D-Wave Systems Inc., Tech. Rep. 14-1039A-B, 2020

    2020

  33. [33]

    Image quality assessment: From error visibility to structural similarity,

    Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: From error visibility to structural similarity,”IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 600–612, 2004

    2004