Pith. sign in

REVIEW 4 major objections 6 minor 32 references

Layer Separation: Adjustable Joint Space Width Images Synthesis in Conventional Radiography

T0 review · 4 major / 6 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims a network can separate finger-joint X-rays into soft tissue, upper bone, and lower bone layers, then shift and recombine the bones to synthesize images with adjustable, ground-truth-labeled joint space width.

desk verdict Useful synthetic radiograph pipeline, but unvalidated layer separation and a circular downstream evaluation keep the headline claims from holding up. read the letter →

arxiv 2502.01972 v1 pith:KHWVALTA submitted 2025-02-04 eess.IV cs.AIcs.CVcs.LG

Haolin Wang , Yafei Ou , Prasoon Ambalathankandy , Gen Ota , Pengyu Dai , Masayuki Ikebe , Kenji Suzuki , Tamotsu Kamishima This is my paper · ORCID
classification eess.IVcs.AIcs.CVcs.LG
keywords rheumatoidarthritisjointspacewidthlayerseparationsyntheticimagegenerationconventionalradiographydataaugmentationgroundtruthannotationdeeplearninginmedicalimaging
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Joint space width (JSW) in hand X-rays is a key measure of rheumatoid arthritis progression, but the datasets used to train automated JSW models are small, imbalanced, and expensive to annotate. This paper claims that a single conventional radiograph can be decomposed into three anatomically meaningful layers—soft tissue, upper bone, and lower bone—and that shifting the bone layers and recombining them according to X-ray attenuation yields realistic synthetic images whose JSW is known exactly from the shift amount. The proposed Layer Separation Network (LSN) makes this decomposition learnable without ground-truth layer images by combining reconstruction loss, segmentation supervision on shifted images, and a soft-tissue discriminator that suppresses bone shadows. The paper reports that synthetic images closely resemble real radiographs, and that pre-training on them improves downstream joint space narrowing progression, JSW regression, and SvdH-like scoring tasks while reducing reliance on annotated real images.

What carries the argument

The load-bearing object is the Layer Separation Network (LSN) together with the reconstruction function $f_r(L) = 1 - \prod_{i=0}^{n}(1 - L_i)$, which encodes the X-ray assumption that overlapping tissues attenuate exponentially and hence combine multiplicatively. The generator (TransUNet) produces layer images from the joint image and bone masks; the segmentation-based supervision network (U-Net) sees randomly shifted reconstructions and is matched against shifted masks; the soft-tissue discrimination network penalizes residual bone shadows in the soft-tissue layer. Random shifting $f_s$ applies rigid transforms (translation $x_i$, $y_i$, rotation $\theta_i$) to bone layers, and the synthesis formula $J^* = f_r(f_s(L, t^*))$ turns the shift parameters into adjustable JSW labels. A two-stage training scheme with pseudo-images supplies layer-level supervision absent from real data.

What would settle it

Use a physical phantom of bone- and soft-tissue-equivalent materials at known separations: take a real radiograph at one spacing, let LSN separate it, shift the bone layers to another spacing, and reconstruct; then compare the synthetic image with an actual radiograph of the phantom at that spacing. If the synthetic and real images differ by more than noise in regions of bone-soft-tissue interfaces, the layer model or the rigid-shift synthesis assumption is wrong.

Watch

Extended reading notes

Core claim

The central discovery, on the paper's own terms, is that layer separation of conventional finger-joint radiographs is achievable without any ground-truth layer images. The generation network outputs three layer images $L_0$, $L_1$, $L_2$ and reconstructs the radiograph as $R = 1 - \prod_{i=0}^{2}(1 - L_i)$, the standard exponential-attenuation model for overlapping tissues. A segmentation network supervises the decomposition by segmenting shifted reconstructions against shifted bone masks, while a discrimination network drives bone shadows out of the soft-tissue layer. Because the layers separate cleanly, joint space width can be adjusted by rigidly translating and rotating the bone layers, and the shift parameters themselves provide ground-truth labels for the synthesized JSW. The paper's evidence is that reconstruction error is low (overall MSE 2.19e-4, SSIM 0.9502, PSNR 36.66), expert radiologic technologists identify real versus synthetic images at about 71% accuracy, and downstream models improve with synthetic pre-training.

Load-bearing premise

The whole pipeline rests on the assumption that a finger-joint X-ray is exactly the multiplicative combination of exactly three independent layers—soft tissue, upper bone, lower bone—and that rigidly shifting the bones and recombining with the same formula reproduces what a real radiograph at that joint spacing would look like.

Editorial extensions

If this is right

  • A single annotated radiograph can be expanded into many synthetic images with a controlled, balanced distribution of JSW values by sampling shift parameters.
  • Each synthetic image carries a ground-truth JSW label derived directly from the applied bone-layer shift, removing the need for manual joint-space annotation.
  • Synthetic pre-training improves downstream joint space narrowing progression quantification, JSW regression, and SvdH-like score classification, and stabilizes models trained on as little as 5% of the real annotated data.
  • The layer decomposition itself provides a foundation for eliminating bone-overlap artifacts, not only for synthesis but for quantification on real images.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, the same multiplicative layer model could transfer to other joints such as the knee or hip where JSW matters, provided the number of tissue layers matches the anatomy and the attenuation assumptions still hold.
  • The synthesis operation defines a differentiable mapping from a shift parameter $t$ to an image; one could invert it to estimate JSW from a real image by finding the shift that minimizes reconstruction distance, turning the synthesizer into a measurement tool.
  • The paper validates realism on a small expert panel of five readers and 100 images, so a larger blinded multi-reader study across imaging conditions would be the natural next check of how far the realism claim extends.
  • If the layer separation is anatomically valid, the same network could clean bone shadows from soft tissue in other projection radiographs, enabling artifact correction beyond finger joints.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The manuscript proposes Layer Separation Networks (LSN) to decompose conventional radiographs of finger joints into three layers -- soft tissue, upper bone, and lower bone -- and then synthesize adjustable joint-space-width (JSW) images by rigidly shifting the bone layers and recombining them with the soft tissue layer via the multiplicative reconstruction formula in Eq. (4). The authors claim that LSN achieves accurate layer separation, that the synthesized images are realistic enough to pass a visual Turing test, and that pre-training on the synthetic data significantly improves downstream JSN progression quantification, JSW quantification, and SvdH-like scoring. The experiments report reconstruction metrics (MSE, SSIM, PSNR, FID), an ablation study, a visual Turing test with five radiological technologists, and downstream-task comparisons with and without synthetic-data pre-training.

Significance. If the layer separation and the adjustable-JSW synthesis were validated independently, the framework would be practically valuable for RA CAD: it addresses data imbalance and annotation scarcity, provides automatic labels through shift parameters, and could serve as a data-augmentation tool for downstream models. Strengths of the paper include a clearly stated image-formation model, a two-stage training strategy with pseudo-image supervision, and the promise of released code and data. However, the central claim of anatomically accurate layer separation is not directly verified -- Section 3.2 explicitly states that no ground truth for layers exists -- and the downstream evaluation is weakened by the fact that both the real-image JSW annotations and the synthetic-image labels are generated by the same layer-separation concept. The paper therefore needs additional independent validation before its main claims can be accepted.

major comments (4)
  1. [§3.2, Eq. (4)] The paper states in §3.2 that, due to the absence of ground truth for layer images, evaluation was conducted exclusively on reconstructed images and real images. This is a load-bearing limitation because Eq. (4), R = 1 - Π_i (1 - L_i), is satisfied by infinitely many triples of layer images; reconstruction fidelity alone cannot certify that the network's decomposition corresponds to true soft tissue, upper bone, and lower bone anatomy. The shifted-segmentation and soft-tissue-discrimination losses constrain the solution, but they do not establish anatomical correctness. Please validate the layers directly using the pseudo-image stage, where bone-layer ground truth exists, or using synthetic phantoms with known attenuation layers, and report layer-level error metrics; otherwise the Abstract's claim that LSN can 'accurately separate' the layers should be withdrawn or substantially softened.
  2. [§3.1 and §3.5] The real-image JSW annotations were created by a manual method 'based on the layer separation' (§3.1), which is the same conceptual decomposition implemented by LSN, and the synthetic-image ground truth is generated from the same LSN shift parameters. The downstream improvements in Table 4 could therefore reflect the model learning LSN's own systematic decomposition rather than an independently defined joint-space width. Please validate the downstream predictions against an independent reference, such as radiologist JSW measurements made without layer-alignment assumptions, known phantom geometry, or an external public dataset with established annotations. Without such a check, the claim that synthetic data 'significantly enhanced the performance in downstream tasks' remains self-referential.
  3. [§3.3, Table 2] Table 2 shows that the full LSN has worse reconstruction metrics than the baseline configuration (MSE 2.19e-4 vs 0.76e-4, SSIM 95.02 vs 97.98, PSNR 36.66 vs 41.22, FID 3.03 vs 1.30). The text explains this as a trade-off for removing bone shadows in the soft tissue layer, but no quantitative bone-shadow metric is reported. Please provide an objective measure of bone-shadow contamination, for example on pseudo-image ground truth or via a structured expert rating, and show that the full model is significantly better on that measure while quantifying the reconstruction trade-off. As written, the quantitative evidence does not support the qualitative claim that the full model produces more clinically acceptable layers.
  4. [§3.4 and §3.5, Tables 3 and 4] The visual Turing test in Table 3 reports an average accuracy of 0.71 from five raters, with no confidence intervals, no chance-level comparison, and no inter-rater agreement statistic; for a forced-choice real/fake task this is only moderate evidence of realism. Similarly, Table 4 reports downstream improvements without error bars or significance tests for most metrics (for example, JSW MSE 8.3166 vs 4.6437 and SvdH accuracy 0.8628 vs 0.8954). Please add repeated-seed experiments with confidence intervals and appropriate statistical tests, or explicitly frame the results as preliminary, before using 'significantly enhanced' in the Abstract.
minor comments (6)
  1. [Eq. (8) and Eq. (12)] Eq. (8) defines LS using ND(J), but the segmentation-based supervision network NS is described as the network in question; similarly Eq. (12) uses D instead of NS. These appear to be typos and should be corrected for reproducibility.
  2. [Table 2] The first row of Table 2 has no checkmarks, making the baseline configuration ambiguous, and the rows with checkmarks do not clearly indicate which combination of NS, ND, T1, and fs is being evaluated. Please format the table with explicit column entries or a legend for every configuration.
  3. [Eqs. (10) and (11)] The notation in Eqs. (10) and (11) is unclear: the definition of B, the meaning of the dot product in Rb = fr(L) · Mb, and the relationship between Mb and M′b should be spelled out, including whether the operations are element-wise and whether the masks are binary or soft.
  4. [§3.1] The sentence 'The dataset contains 430 MCP joints for 1,594 joint images' is ambiguous about whether multiple images correspond to the same joint and how the joint-level split prevents data leakage; please clarify the patient/joint/image hierarchy.
  5. [Eq. (5)] The transformation matrix in Eq. (5) uses xi, yi, and θi but does not specify the coordinate convention or whether θi is in degrees or radians; please state the convention and the origin used for rotation.
  6. [Figures 3 and 4] Figures 3 and 4 are captioned as 'ablation study' results, but one appears to illustrate the main method outputs; the captions should be revised to describe exactly which configuration each figure shows.

Circularity Check

3 steps flagged · score 4.0 of 10

Partial circularity: the synthetic JSW ground truth is defined by the same layer-shift parameters used to render the images, and the real-image JSW annotations are themselves based on the same layer-separation concept, so the downstream improvement is partly self-consistent rather than independently validated.

  1. self definitional [Section 2.3, Eq. 13]
    "A substantial dataset of synthetic images with varying JSW can be created from a single input image, as illustrated in Eq.13, where t∗ denotes the shifting parameters. Combined with the original annotations (e.g., JSW and SvdH), the shifting parameters can be used to produce GT of this synthetic image."

    Equation 13 defines the synthetic image as J* = fr(fs(L, t*)), and the ground truth is then produced from the same shifting parameters t*. Thus the synthetic image's JSW label is not an independent measurement; it is exactly the control parameter used to construct the image. This is a legitimate synthetic-data construction, but presenting it as 'ground truth generation' means downstream evaluation on these images checks consistency with the construction, not anatomical truth.

  2. self definitional [Section 3.1, Joint Image Dataset]
    "For downstream tasks, the JSW of each image was annotated using the method developed based on the layer separation, under the guidance of experienced radiologists. Specifically, the method enabled manual alignment of the upper and lower bones by adjusting their positions to align the boundaries of the joint contact surfaces, where the JSW value was defined as zero. Thus, the JSW was subsequently calculated as the difference between the displacements of the bone layers."

    The real-image JSW labels are defined by manually displacing bone layers until the joint surfaces align, i.e., JSW equals a layer displacement. LSN's synthetic adjustable-JSW images are generated by exactly this operation, shifting bone layers and recomposing them with the soft tissue layer. Therefore the downstream JSW quantification evaluation measures agreement with a label definition that shares the same layer-shift model as the synthetic pre-training data. The reported improvement may partly reflect this definitional alignment rather than independent anatomical accuracy.

1 more flagged steps
  1. self citation load bearing [Section 2.2, Pseudo Images and Two-stage training]
    "In order to improve the stability and accuracy of the network, we create pseudo-images ˜J for two-stage training, with overlapped regions based on non-overlapped images by modifying the image processing described in BLS-GAN (Wang et al., 2024)."

    The first-stage training obtains upper and lower bone ground truth from pseudo-images created by the authors' prior BLS-GAN framework, which is the same layer-separation approach this paper extends. The paper later concedes that there is no independent ground truth for the layer images (Section 3.2), so this self-citation is load-bearing: the claimed accuracy of LSN's layer separation is trained and assessed within the same family of constructions rather than against an external gold standard.

full rationale

The paper is not fully circular: it reports reconstruction quality against real radiographs (MSE, SSIM, PSNR, FID), a visual Turing test with experienced radiological technologists, and downstream improvements on real images, all of which provide some independent checks of realism. The core synthesis equation (Eq. 4, R = 1 - product(1 - L_i)) is an imaging-model assumption, not a result derived from itself. However, three steps introduce partial circularity. First, the synthetic adjustable-JSW ground truth is defined by the same shift parameters used to render the image, so the label is a construction rather than an independent truth. Second, the real-image JSW annotations are made by a manual layer-alignment method that is explicitly 'based on the layer separation', meaning the same layer-shift concept underlies both the synthetic pre-training data and the evaluation labels; downstream gains are therefore partly a consistency check with that concept. Third, the pseudo-image supervision in the first training stage comes from the authors' self-cited BLS-GAN, and no independent layer ground truth exists, making that self-citation load-bearing. These issues do not collapse the paper's central contribution to a pure tautology, but they do mean that the 'accurate layer separation' and 'downstream improvement' claims are weaker than stated. Accordingly, the circularity score is 4 rather than 0-2.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a physical layer-decomposition model (Eq. 4), the availability of bone masks, and pseudo-image training borrowed from the self-cited BLS-GAN. The free parameters are loss weights and the unstated random shifting range, all chosen by hand; no new physical entities are introduced.

free parameters (3)
  • Loss weights (α, β, γ) in Eq. 7 = 0.6, 0.3, 0.1
    Set from 'experimental experience' (Sec 2.2); these weights determine the balance between reconstruction fidelity, segmentation supervision, and soft tissue discrimination, directly shaping the layer separation output.
  • Two-stage loss weights (α', β', γ', δ and α'', β'', δ') in Eq. 11 and Eq. 12 = α'=0.5, β'=0.2, γ'=0.2, δ=0.1; α''=1, β''=0.4, δ'=0.4
    Hand-picked for pseudo-image stage; no sensitivity analysis or independent validation is provided.
  • Random shifting parameter range (translation xi, yi and rotation θi) = not specified
    The shifting function fs (Eq. 6) uses a 'predefined range' for synthesis (Sec 2.3), but the actual range is not reported; this controls the JSW distribution and overlap in synthetic images.
assumptions (5)
  • domain assumption A conventional radiograph of a finger joint is formed only through the overlap of upper bone, lower bone, and soft tissue textures following specific principles (Sec 2.2).
    This decomposition ignores scatter, beam hardening, patient motion, and other physical effects; it is necessary for the layer separation to be well-defined.
  • domain assumption The reconstruction function R = 1 - Π_i (1 - L_i) (Eq. 4) exactly models X-ray attenuation as multiplicative exponential decay.
    Standard physics textbook model (Bushberg and Boone, 2011), but an idealization; the paper offers no validation that this invertible product form captures real radiographs.
  • domain assumption Rigidly shifting and rotating bone layers is sufficient to synthesize realistic adjustable-JSW images (Eq. 6 and Eq. 13).
    Assumes soft tissue stays unchanged while bones move, i.e., no tissue deformation, no change in scatter, and no alignment artifacts; unvalidated.
  • domain assumption The pseudo-image construction from BLS-GAN (Eq. 10) is a valid proxy for real radiographs during first-stage training.
    Pseudo-images are created by splicing shifted bone regions with original soft tissue; this is taken from a self-cited paper and not independently verified here.
  • domain assumption Bone masks M are available for every input image (Eq. 1).
    The method requires annotated bone masks at train and test time; the dataset (from BLS-GAN) provides them, but this limits applicability to datasets with masks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Layer Separation: Adjustable Joint Space Width Images Synthesis in Conventional Radiography." pith.science (2026). https://pith.science/paper/KHWVALTA

@misc{pith2026250201972,
  author       = {Pith},
  title        = {Pith review of: Layer Separation: Adjustable Joint Space Width Images Synthesis in Conventional Radiography},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KHWVALTA}},
  note         = {Machine review of arXiv:2502.01972}
}
read the original abstract

Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by joint inflammation and progressive structural damage. Joint space width (JSW) is a critical indicator in conventional radiography for evaluating disease progression, which has become a prominent research topic in computer-aided diagnostic (CAD) systems. However, deep learning-based radiological CAD systems for JSW analysis face significant challenges in data quality, including data imbalance, limited variety, and annotation difficulties. This work introduced a challenging image synthesis scenario and proposed Layer Separation Networks (LSN) to accurately separate the soft tissue layer, the upper bone layer, and the lower bone layer in conventional radiographs of finger joints. Using these layers, the adjustable JSW images can be synthesized to address data quality challenges and achieve ground truth (GT) generation. Experimental results demonstrated that LSN-based synthetic images closely resemble real radiographs, and significantly enhanced the performance in downstream tasks. The code and dataset will be available.

Figures

Figures reproduced from arXiv: 2502.01972 by the authors.

Figure 1
Figure 1. The adjustable JSW synthetic images are generated by producing layer images, following random shifting of the bone layers, and reconstruction with soft tissue layer. Original Images: imbalanced distribution of JSW, limited semantic variety, and diffi￾culty in manual annotation. Synthetic Data: balanced distribution, enhanced semantic variety, and generative ground truth (GT) anno￾tations. in joint space width (JSW) … view at source ↗
Figure 2
Figure 2. Layer Separation: it generates the layer images of upper and lower bones and soft tissues based on a single image. The LSN consists of five main components: a generation network NG, a supervision network NS , a discrimination network ND, a random shifting function fs and a reconstruction function fr. The generation process is performed as follows: (i) The NG processes the original joint images and the corresponding … view at source ↗
Figure 3
Figure 3. Visualization results of our ablation study. (A) Real Joint image; (B) Shifted Reconstruction Joint Image; (C) Upper Bone Layer; (D) Lower bone layer; (E) Soft Tissue Layer; (F) MSE Spectrum (Reconstruction Joint Image v.s. A) [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Original and Synthetic images in Visual Turing Test [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Reduced annotations for the downstream models. datasets), and for synthetic data, we reduced the original real data used to create synthetic data and amplified it into an equal amount of synthetic data. As illustrated in [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 26 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    A., Mahafza, W., and Al-Kadi, O

    Abu-Srhan, A., Almallahi, I., Abushariah, M. A., Mahafza, W., and Al-Kadi, O. S. Paired-unpaired unsupervised attention guided gan with transfer learning for bidirectional brain mr-ct synthesis. Computers in Biology and Medicine, 136: 0 104763, 2021

    work page 2021

  3. [3]

    T., and Joseph Raj, A

    Ahalya, R., Umapathy, S., Krishnan, P. T., and Joseph Raj, A. N. Automated evaluation of rheumatoid arthritis from hand radiographs using machine learning and deep learning techniques. Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine, 236 0 (8): 0 1238--1249, 2022

    work page 2022

  4. [4]

    On the usability of synthetic data for improving the robustness of deep learning-based segmentation of cardiac magnetic resonance images

    Al Khalil, Y., Amirrajab, S., Lorenz, C., Weese, J., Pluim, J., and Breeuwer, M. On the usability of synthetic data for improving the robustness of deep learning-based segmentation of cardiac magnetic resonance images. Medical Image Analysis, 84: 0 102688, 2023

    work page 2023

  5. [5]

    and Smolen, J

    Aletaha, D. and Smolen, J. S. Diagnosis and management of rheumatoid arthritis: a review. Jama, 320 0 (13): 0 1360--1372, 2018

    work page 2018

  6. [6]

    Bushberg, J. T. and Boone, J. M. The essential physics of medical imaging. Lippincott Williams & Wilkins, 2011

    work page 2011

  7. [7]

    and Draxler, R

    Chai, T. and Draxler, R. R. Root mean square error (rmse) or mean absolute error (mae)?--arguments against avoiding rmse in the literature. Geoscientific model development, 7 0 (3): 0 1247--1250, 2014

    work page 2014

  8. [8]

    L., and Zhou, Y

    Chen, J., Lu, Y., Yu, Q., Luo, X., Adeli, E., Wang, Y., Lu, L., Yuille, A. L., and Zhou, Y. Transunet: Transformers make strong encoders for medical image segmentation. arXiv preprint arXiv:2102.04306, 2021

    arXiv 2021

Show all 32 references
  1. [9]

    Towards generalizable tumor synthesis

    Chen, Q., Chen, X., Song, H., Xiong, Z., Yuille, A., Wei, C., and Zhou, Z. Towards generalizable tumor synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 11147--11158, 2024

    2024

  2. [10]

    P., Brooks, R., En, S., Zucker, E., Pareek, A., Lungren, M

    Cohen, J. P., Brooks, R., En, S., Zucker, E., Pareek, A., Lungren, M. P., and Chaudhari, A. Gifsplanation via latent shift: a simple autoencoder approach to counterfactual generation for chest x-rays. In Medical Imaging with Deep Learning, pp.\ 74--104. PMLR, 2021

    2021

  3. [11]

    Sasamim: Synthetic anatomical semantics-aware masked image modeling for colon tumor segmentation in non-contrast abdominal computed tomography

    Dai, P., Ou, Y., Yang, Y., Liu, D., Hashimoto, M., Jinzaki, M., Miyake, M., and Suzuki, K. Sasamim: Synthetic anatomical semantics-aware masked image modeling for colon tumor segmentation in non-contrast abdominal computed tomography. In International Conference on Medical Ima...

    2024

  4. [12]

    Tutorial on variational autoencoders

    Doersch, C. Tutorial on variational autoencoders. arXiv preprint arXiv:1606.05908, 2016

    2016 arXiv

  5. [13]

    Generative adversarial networks

    Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., and Bengio, Y. Generative adversarial networks. Communications of the ACM, 63 0 (11): 0 139--144, 2020

    2020

  6. [14]

    Development and validation of a deep-learning model for scoring of radiographic finger joint destruction in rheumatoid arthritis

    Hirano, T., Nishide, M., Nonaka, N., Seita, J., Ebina, K., Sakurada, K., and Kumanogoh, A. Development and validation of a deep-learning model for scoring of radiographic finger joint destruction in rheumatoid arthritis. Rheumatology advances in practice, 3 0 (2): 0 rkz047, 2019

    2019

  7. [15]

    Denoising diffusion probabilistic models

    Ho, J., Jain, A., and Abbeel, P. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 0 6840--6851, 2020

    2020

  8. [16]

    and Abrahams, R

    Huda, W. and Abrahams, R. B. Radiographic techniques, contrast, and noise in x-ray imaging. American Journal of Roentgenology, 204 0 (2): 0 W126--W131, 2015

    2015

  9. [17]

    Jaipuria, N., Zhang, X., Bhasin, R., Arafa, M., Chakravarty, P., Shrivastava, S., Manglani, S., and Murali, V. N. Deflating dataset bias using synthetic data augmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pp.\ 772-...

    2020

  10. [18]

    T., Han, T., Haarburger, C., Schulze-Hagen, M., Schad, P., Engelhardt, S., Baessler, B., Foersch, S., et al

    Khader, F., Mueller-Franzes, G., Arasteh, S. T., Han, T., Haarburger, C., Schulze-Hagen, M., Schad, P., Engelhardt, S., Baessler, B., Foersch, S., et al. Medical diffusion: denoising diffusion probabilistic models for 3d medical image generation. arXiv preprint arXiv:2211.03364, 2022

    2022 arXiv

  11. [19]

    M., Puglisi, C

    Kingsmore, K. M., Puglisi, C. E., Grammer, A. C., and Lipsky, P. E. An introduction to machine learning and analysis of its use in rheumatic diseases. Nature Reviews Rheumatology, 17 0 (12): 0 710--730, 2021

    2021

  12. [20]

    R., Wu, J., Mastrodicasa, D., Lutz, A., Chung, M., Koszek, W

    Koetzier, L. R., Wu, J., Mastrodicasa, D., Lutz, A., Chung, M., Koszek, W. A., Pratap, J., Chaudhari, A. S., Rajpurkar, P., Lungren, M. P., et al. Generating synthetic data for medical imaging. Radiology, 312 0 (3): 0 e232471, 2024

    2024

  13. [21]

    Automatic quantification of joint space narrowing and erosions in rheumatoid arthritis

    Langs, G., Peloschek, P., Bischof, H., and Kainberger, F. Automatic quantification of joint space narrowing and erosions in rheumatoid arthritis. IEEE transactions on medical imaging, 28 0 (1): 0 151--164, 2008

    2008

  14. [22]

    A unified conditional disentanglement framework for multimodal brain mr image translation

    Liu, X., Xing, F., El Fakhri, G., and Woo, J. A unified conditional disentanglement framework for multimodal brain mr image translation. In 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI), pp.\ 10--14. IEEE, 2021

    2021

  15. [23]

    A sub-pixel accurate quantification of joint space narrowing progression in rheumatoid arthritis

    Ou, Y., Ambalathankandy, P., Furuya, R., Kawada, S., Zeng, T., An, Y., Kamishima, T., Tamura, K., and Ikebe, M. A sub-pixel accurate quantification of joint space narrowing progression in rheumatoid arthritis. IEEE Journal of Biomedical and Health Informatics, 27 0 (1): 0 53--64, 2023

    2023

  16. [24]

    Synthetic data for deep learning in computer vision & medical imaging: A means to reduce data bias

    Paproki, A., Salvado, O., and Fookes, C. Synthetic data for deep learning in computer vision & medical imaging: A means to reduce data bias. ACM Computing Surveys, 2024

    2024

  17. [25]

    E., Heyne, J.-P., Petrovitch, A., Eidner, T., Mentzel, H.-J., Wolf, G., Hein, G., and Kaiser, W

    Pfeil, A., B \"o ttcher, J., Seidl, B. E., Heyne, J.-P., Petrovitch, A., Eidner, T., Mentzel, H.-J., Wolf, G., Hein, G., and Kaiser, W. A. Computer-aided joint space analysis of the metacarpal-phalangeal and proximal-interphalangeal finger joint: normative age-related and gend...

    2007

  18. [26]

    Fully automated joint space width measurement and digital x-ray radiogrammetry in early ra

    Platten, M., Kisten, Y., K \"a lvesten, J., Arnaud, L., Forslind, K., and van Vollenhoven, R. Fully automated joint space width measurement and digital x-ray radiogrammetry in early ra. RMD open, 3 0 (1): 0 e000369, 2017

    2017

  19. [27]

    C., Staring, M., Reijnierse, M., and van der Helm-van Mil, A

    Stoel, B. C., Staring, M., Reijnierse, M., and van der Helm-van Mil, A. H. Deep learning in rheumatological image interpretation. Nature Reviews Rheumatology, 20 0 (3): 0 182--195, 2024

    2024

  20. [28]

    How to read radiographs according to the sharp/van der heijde method

    Van der Heijde, D. How to read radiographs according to the sharp/van der heijde method. The Journal of rheumatology, 27 0 (1): 0 261--263, 2000

    2000

  21. [29]

    Reliability and sensitivity to change of a simplification of the sharp/van der heijde radiological assessment in rheumatoid arthritis

    Van der Heijde, D., Dankert, T., Nieman, F., Rau, R., and Boers, M. Reliability and sensitivity to change of a simplification of the sharp/van der heijde radiological assessment in rheumatoid arthritis. Rheumatology, 38 0 (10): 0 941--947, 1999

    1999

  22. [30]

    A deep registration method for accurate quantification of joint space narrowing progression in rheumatoid arthritis

    Wang, H., Ou, Y., Fang, W., Ambalathankandy, P., Goto, N., Ota, G., Okino, T., Fukae, J., Sutherland, K., Ikebe, M., et al. A deep registration method for accurate quantification of joint space narrowing progression in rheumatoid arthritis. Computerized Medical Imaging and Gra...

    2023

  23. [31]

    Bls-gan: A deep layer separation framework for eliminating bone overlap in conventional radiographs

    Wang, H., Ou, Y., Ambalathankandy, P., Ota, G., Dai, P., Ikebe, M., Suzuki, K., and Kamishima, T. Bls-gan: A deep layer separation framework for eliminating bone overlap in conventional radiographs. arXiv preprint arXiv:2409.07304, 2024

    2024 arXiv

  24. [32]

    Unified focal loss: Generalising dice and cross entropy-based losses to handle class imbalanced medical image segmentation

    Yeung, M., Sala, E., Sch \"o nlieb, C.-B., and Rundo, L. Unified focal loss: Generalising dice and cross entropy-based losses to handle class imbalanced medical image segmentation. Computerized Medical Imaging and Graphics, 95: 0 102026, 2022

    2022

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.