Controllable Skin Synthesis via Lesion-Focused Vector Autoregression Model

Jason J. Ong; Jiajun Sun; Lei Zhang; Siyuan Yan; Zhen Yu; Zongyuan Ge

REVIEW 4 major objections 4 minor 28 references

Controllable Skin Synthesis via Lesion-Focused Vector Autoregression Model

T0 review · 4 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read A two-stage model uses lesion masks, lesion type, and quantified lesion measurements to control skin-image synthesis, achieving the best FID among seven lesion types and boosting downstream classifier recall.

desk verdict LF-VAR is a plausible integration of VAR with radiomics measurement conditioning for controllable skin synthesis, but the headline FID improvement over prior SOTA is statistically fragile and the 'lesion-focused' loss in Eq. (2) is actually background-weighting. read the letter →

arxiv 2508.19626 v1 pith:ZQK7RWNF submitted 2025-08-27 cs.CV

Jiajun Sun , Zhen Yu , Siyuan Yan , Jason J. Ong , Zongyuan Ge , Lei Zhang This is my paper

classification cs.CV

keywords skinlesionsynthesiscontrollablegenerationvectorautoregressivetransformerVQ-VAEradiomicsmeasurementHAM10000FréchetInceptionDistance

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

The paper introduces LF-VAR, a controllable skin lesion synthesis model built from a multiscale lesion-focused VQ-VAE and a Visual AutoRegressive (VAR) transformer. The model conditions synthesis on lesion type labels and on quantified lesion measurements—shape, texture, and histogram scores extracted from the lesion region—so that generated images preserve clinically relevant lesion characteristics. The authors report the best average FID score (0.74) across seven lesion types, a 6.3% improvement over prior state of the art, and show that adding synthetic images to training data raises a downstream 7-class classifier's recall from 0.692 to 0.771. The work matters because real skin-image datasets are small, class-imbalanced, and privacy-restricted, making controllable synthetic data an attractive supplement for training medical AI models.

What carries the argument

The central mechanism is the combined conditioning token sequence [S, Fr], where S is a lesion-type class embedding and Fr is a learned embedding of quantified lesion measurements (shape, histogram, texture) extracted from the lesion mask. This token is prepended to the multiscale discrete token maps produced by a lesion-focused VQ-VAE, and a VAR transformer generates images by autoregressive next-scale prediction. A class-average measurement codebook supports inter-class synthesis, while a lesion-focused loss term—enforcing reconstruction fidelity only outside the lesion mask—suppresses background artifacts.

What would settle it

Recompute FID on repeated 4:1 splits of HAM10000 and report per-class 95% confidence intervals; if the average FID confidence intervals of LF-VAR and VAR overlap, the claimed 6.3% improvement is not established.

Watch

Extended reading notes

Core claim

LF-VAR establishes that injecting radiomics-style lesion measurements into a VAR transformer, alongside lesion type embeddings and segmentation masks, yields high-fidelity controllable skin lesion synthesis. The lesion-focused VQ-VAE discretizes images into multiscale token grids while concentrating reconstruction loss on non-lesion regions, reducing background artifacts. Measurements from PyRadiomics are projected through a linear layer, layer normalization, and SiLU activation to form a conditioning token; a codebook of class-average measurements enables inter-class synthesis. On HAM10000, the model achieves an average FID of 0.74, outperforming diffusion, masked generative, and text-promp

Load-bearing premise

The headline FID advantage over the previous best model is 0.05 on average, computed on per-class test sets as small as roughly 24 images, with no confidence intervals or significance tests reported.

Editorial extensions

If this is right

If the FID result holds, LF-VAR is the strongest controllable skin-lesion synthesizer on HAM10000, outperforming diffusion and masked-generation baselines with fewer computational resources.
The downstream recall gain from 0.692 to 0.771 shows that synthetic images from the model can partially compensate for class imbalance in skin-image training sets.
Because synthesis is conditioned on masks and lesion type, the model can generate lesions at specified locations and across categories, including inter-class combinations from the average-measurement codebook.
Cross-dataset FID results on ISIC2017 and Dermofit suggest the conditioning approach transfers to images outside the training distribution.

Reading between the lines

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

The conditioning mechanism is modality-agnostic: radiomics-style measurements exist for CT, MRI, and pathology, so the same token-embedding trick could plausibly control synthesis in those domains.
The class-average measurement codebook offers an interpretable handle: editing a codebook entry should shift generated morphology, enabling controlled synthesis of rare or unseen lesion variants.
Clinical utility needs a human expert study: FID and classifier recall do not guarantee that generated lesions are dermatologically correct, so a clinician-rated realism study is a natural next test.
The balanced augmentation protocol (500 images per class) is one specific recipe; gains likely vary by lesion class and dataset size, and the method should be tested in lower-data regimes.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Quick take: the paper does something genuinely new by injecting PyRadiomics-derived lesion measurements as learned conditional embeddings into a VAR transformer, and by maintaining a class-average measurement codebook for inter-class synthesis. That specific arrangement is not in the prior work they cite, and the evaluation scope—intra-class, inter-class, cross-dataset, and downstream classification—is broader than most synthesis papers attempt. Credit where due: they report a downstream recall gain from 0.692 to 0.771 with balanced synthetic augmentation, which is the most concrete evidence of utility in the paper.

The soft spots are real and concentrated in the quantitative claims. The central FID comparison (0.74 vs 0.79) rests on per-class test sets as small as ~24 images (DF, VASC). FID at that scale is high-variance and biased, and no confidence intervals, bootstrap replicates, or repeated runs are reported. The improvement is driven largely by BCC and DF, two of the smallest classes, so the 6.3% gap is indistinguishable from sampling noise as presented. Table 3's ablation has the same problem, and the non-monotonicity—Baseline + LF + FM (1.05) is worse than Baseline + LF alone (0.76)—is never discussed, which is odd for a component that is supposed to help.

There's also a genuine naming flaw in Eq. (2). The second term multiplies by (1 − M_k), which weights non-lesion (background) regions, because M_k is the lesion mask. Call it 'background consistency' or explain why that is what you want; the current text says it is a lesion-focused loss, and that contradiction weakens the causal story that lesion-focused synthesis drives the gains. This is not fatal to the model—background regularization might well be useful—but it must be fixed.

The downstream task validation is also underbenchmarked. The +11.4% recall improvement is compared only against a weighted random sampler and a baseline classifier, not against other generative augmentation methods or simpler class-balancing strategies. So the practical utility claim is suggestive, not demonstrated.

Who is this for: dermatology AI researchers working on synthetic data augmentation or controllable medical image generation. They will find a useful baseline and a clear design space. The paper deserves a serious referee: the core idea is worth engaging, the experimental breadth is a strength, and the issues are fixable. But the SOTA claim should not be accepted until FID uncertainty is quantified, the loss is clarified, and the downstream comparison is strengthened. I would send it out for review and expect major revision.

Referee Report

4 major / 4 minor

Summary. The paper proposes LF-VAR, a two-stage controllable skin-lesion synthesis model. Stage one is a multiscale VQ-VAE trained with a 'lesion-focused' loss; stage two is a Visual AutoRegressive (VAR) Transformer conditioned on radiomics-style lesion measurement scores extracted from the lesion region and on lesion-type embeddings. The authors claim state-of-the-art average FID of 0.74 on HAM10000, a 6.3% improvement over the previous best VAR (0.79), and they report downstream 7-class recall gains from 0.692 to 0.771 using synthetic data augmentation. Ablation studies and cross-dataset experiments on ISIC2017 and Dermofit are also presented.

Significance. If the headline FID improvement is statistically reliable, the paper makes a useful contribution: it demonstrates that clinically defined lesion measurements can be used as conditioning signals for generative skin-image synthesis, and it provides a concrete downstream classification benefit. The open code and the careful architectural decomposition (VQ-VAE + VAR with explicit conditioning) are strengths. However, the central quantitative claim rests on FID values computed from very small per-class test sets with no confidence intervals or significance testing, so the claimed 6.3% improvement is not currently established. The 'lesion-focused' loss in Eq. (2) also appears to weight background rather than lesion regions, which weakens the causal interpretation of the main contribution.

major comments (4)

[Section 3.2, Table 1] The central claim of best average FID (0.74 vs 0.79) is not statistically supported. FID is computed per class on 4:1 splits of HAM10000; for the smallest classes this yields roughly 24 real test images (DF, original n=118) and 28 (VASC, original n=142). At these sample sizes FID has substantial bias and variance, yet no confidence intervals, bootstrap replicates, or multiple training runs are reported. The average difference of 0.05 FID is within the expected noise of such estimates. The paper should provide bootstrap CIs or repeated-run statistics for Table 1, or otherwise temper the superiority claim.
[Section 2.1, Eq. (2)] The 'lesion-focused' loss term is written as ∑_k ‖(1 − M_k) ⊙ (R_k − R̂_k)‖^2. Since M_k is defined as the lesion mask, (1 − M_k) is zero on the lesion and one on the background, so this term enforces reconstruction in non-masked background regions, not in the lesion. The text says it 'enforcing similarity ... in non-masked areas,' which is consistent with the equation but contradicts the named 'lesion-focused' mechanism and contribution #3. If the intended focus is the lesion, the mask should be M_k (or the definition of M_k reversed). As written, the equation does not support the causal claim that the method's gains come from lesion-focused synthesis.
[Section 3.2, Table 3 and Ablation study] The ablation study reports FID values without any variance or significance information. The claimed gains of 3.7% (Baseline+LF) and 6.3% (Baseline+LF+AM) are per-class averages over the same small test sets discussed above; without confidence intervals, these improvements may be sampling noise. Notably, adding the fixed measurement embedding (FM) degrades performance substantially (average FID 1.05), and the paper does not explain why the adaptive embedding behaves so differently. Statistical validation is needed before the component-wise contribution claims can be accepted.
[Section 3.1, Tasks and metrics] The FID computation protocol is underspecified. The number of generated images per class and the number of real images used as the reference set are not stated. If FID is computed with the small per-class test partitions (e.g., 24 real images for DF), the estimator is known to be biased and high-variance. The paper should state how many synthetic and real images are used for each FID calculation, and ideally use a fixed reference set of adequate size or report Monte Carlo confidence intervals.

minor comments (4)

[Throughout] The paper has several typographical and formatting issues: 'LF-V AR' and 'T able' with stray spaces, inconsistent subscripting (F_q and ε_r), and 'Text-Mask' in Table 1 reads awkwardly. The reference to 'Philipp et al. [27]' should be spelled consistently (Tschandl et al.).
[Section 2.2] The notation for the measurement encoding function is inconsistent: the text says 'Fq = Eq(εr)' but elsewhere uses Ε_q and F_q; please unify the notation. Also clarify what exactly the 'fixed measurement embedding' (FM) in the ablation is: a constant vector across all lesions is ambiguous.
[Figure 3] The t-SNE and feature-distance panels are visually informative but lack axis labels and a clear caption explaining how feature distance is computed. Please add details in the caption or text.
[Section 3.2, Inter-class synthesis] The FID confusion matrix is reported only for selected classes (e.g., 'all lesion categories achieved the best ... when generating VASC'). It would be helpful to show the full 7x7 matrix in the paper rather than only a figure, to permit verification.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the FID result is an empirical evaluation of a conditional generator, not a derivation from its own inputs.

full rationale

This paper does not claim a first-principles derivation; it trains a conditional generator (LF-VAR) and evaluates it with FID/IS on held-out test data. The conditioning signals (lesion-type embedding S, mask M, and radiomic measurement scores extracted from input images) are legitimate inputs for controllable synthesis, not fitted parameters renamed as predictions. The FID reference set is a separate 4:1 held-out test split, and the intra-class task is described as replicating training-set patterns, so no specific equation in the paper makes the reported FID equal to the conditioning input by construction. The main concerns in the surrounding discussion—small per-class test sizes (e.g., ~24 DF and ~28 VASC images), no confidence intervals on FID, and the fact that Eq. (2)'s second term multiplies by (1−M_k) and is therefore misnamed a 'lesion-focused loss'—are statistical-robustness and naming issues, not circularity. There are no load-bearing self-citations or imported uniqueness claims. Under the rule that circularity requires a quotable, specific reduction (e.g., Eq. X = Eq. Y by construction or a fitted parameter renamed as a prediction), none is present here.

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

The central claim depends on undisclosed hyperparameters and on several background assumptions about the data and the FID metric. No new physical entities are introduced.

free parameters (3)

loss weights λP and λG = not reported
Hand-chosen weights in Eq. (2); their values are not given and they balance reconstruction, perceptual, and discriminative terms in the loss.
number of quantization scales K = not reported
Eq. (1) uses K discrete token grids; the exact K is not disclosed in the paper.
radiomics feature vector εr composition = unspecified (shape, histogram, texture subset)
The conditioning signal is said to come from PyRadiomics, but the exact feature set, dimensionality, and normalization are not specified.

assumptions (5)

domain assumption HAM10000 ground-truth lesion masks and class labels are accurate
Masks define the lesion region for measurement extraction and for the (1-M_k) loss; any mask error propagates to the conditioning signal and the loss.
domain assumption PyRadiomics descriptors computed on dermoscopy masks capture clinically relevant lesion morphology
The paper cites radiomics literature (refs 15, 16, 24) but provides no validation that these features are meaningful for dermoscopic lesions beyond their own synthesis use.
standard math The VAR/VQVAE architectures and training procedures from prior work remain valid in this domain
Relies on cited implementations (VAR [25], VQVAE [19]) without re-derivation; reasonable background assumption.
domain assumption FID with pre-trained Inception features is a valid fidelity metric for dermoscopic images
The entire comparison rests on this metric; general-purpose features may not capture dermatological fidelity.
ad hoc to paper The loss in Eq. (2) as written (with (1-M_k)) is the implemented objective
If the text is correct, the 'lesion-focused' term actually constrains background. If the equation is a typo, the implemented objective differs from what is reported.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Controllable Skin Synthesis via Lesion-Focused Vector Autoregression Model." pith.science (2026). https://pith.science/paper/ZQK7RWNF

@misc{pith2026250819626,
  author       = {Pith},
  title        = {Pith review of: Controllable Skin Synthesis via Lesion-Focused Vector Autoregression Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZQK7RWNF}},
  note         = {Machine review of arXiv:2508.19626}
}

read the original abstract

Skin images from real-world clinical practice are often limited, resulting in a shortage of training data for deep-learning models. While many studies have explored skin image synthesis, existing methods often generate low-quality images and lack control over the lesion's location and type. To address these limitations, we present LF-VAR, a model leveraging quantified lesion measurement scores and lesion type labels to guide the clinically relevant and controllable synthesis of skin images. It enables controlled skin synthesis with specific lesion characteristics based on language prompts. We train a multiscale lesion-focused Vector Quantised Variational Auto-Encoder (VQVAE) to encode images into discrete latent representations for structured tokenization. Then, a Visual AutoRegressive (VAR) Transformer trained on tokenized representations facilitates image synthesis. Lesion measurement from the lesion region and types as conditional embeddings are integrated to enhance synthesis fidelity. Our method achieves the best overall FID score (average 0.74) among seven lesion types, improving upon the previous state-of-the-art (SOTA) by 6.3%. The study highlights our controllable skin synthesis model's effectiveness in generating high-fidelity, clinically relevant synthetic skin images. Our framework code is available at https://github.com/echosun1996/LF-VAR.

Figures

Figures reproduced from arXiv: 2508.19626 by the authors.

**Figure 1.** The framework of LF-VAR. A structured VQ-VAE encoded images into K token maps, shared with the VAR, and was trained with lesion-focused loss. Lesion measurement scores were extracted from lesion areas and encoded as Fq, combined with lesion type embedding S, and incorporated into the VAR as a conditional token to enable controlled intra-class skin lesion synthesis. Inter-class synthesis was achieved through the VAR … view at source ↗

**Figure 2.** Comparison of synthesis samples by baseline and our method, with two samples from different lesions of each model. 3 Experiments and Results 3.1 Experiments Setup Dataset. In this study, the HAM10000 [26] dataset was utilized to train the LF-VAR model. The dataset consisted of dermatoscopy images annotated with seven skin lesion categories: actinic keratoses (AKIEC, 327 images), basal cell carcinoma (BCC, 514), beni… view at source ↗

**Figure 3.** Feature analysis and model comparison. (a) Feature distribution across seven skin lesion types. (b) Feature distribution comparison between real and synthetic in melanoma. (c) Synthesis images feature distance and FID comparison among models. IS reflecting improved variation. In this study, we primarily focus on FID as it provides a more reliable measure of fidelity by comparing feature distributions between real an… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Inter-class synthesis comparison and FID matrix. (a) Inter-class synthesis samples for seven lesion types. The first two columns show the original images and masks, followed by synthesis images. (b) FID confusion matrix for inter-class synthesis [PITH_FULL_IMAGE:figu…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

28 extracted references · 14 canonical work pages

[1]

Bai, Y., Geng, X., Mangalam, K., Bar, A., Yuille, A., Darrell, T., Malik, J., Efros, A.A.: Sequential modeling enables scalable learning for large vision models (2023), https://arxiv.org/abs/2312.00785

arXiv 2023
[2]

Color medical image analysis pp

Ballerini, L., Fisher, R.B., Aldridge, B., Rees, J.: A color and texture based hi- erarchical k-nn approach to the classification of non-melanoma skin lesions. Color medical image analysis pp. 63–86 (2013)

work page 2013
[3]

National Science Review 11(12), nwae348 (10 2024)

Chen, M., Mei, S., Fan, J., Wang, M.: Opportunities and challenges of diffusion models for generative ai. National Science Review 11(12), nwae348 (10 2024). https://doi.org/10.1093/nsr/nwae348

doi:10.1093/nsr/nwae348 2024
[4]

https://doi.org/10.48550/arXiv.1710.05006 10 J

Codella, N.C.F., Gutman, D., Celebi, M.E., Helba, B., Marchetti, M.A., Dusza, S.W., Kalloo, A., Liopyris, K., Mishra, N., Kittler, H., Halpern, A.: Skin lesion analysis toward melanoma detection: A challenge at the 2017 international sym- posium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC). https://doi.org/10...

arXiv doi:10.48550/arxiv.1710.05006 2017
[5]

In: 2024 IEEE International Symposium on Biomedical Imaging (ISBI)

Du, S., Wang, X., Lu, Y., Zhou, Y., Zhang, S., Yuille, A., Li, K., Zhou, Z.: Boosting dermatoscopic lesion segmentation via diffusion models with visual and textual prompts. In: 2024 IEEE International Symposium on Biomedical Imaging (ISBI). pp. 1–5 (2024). https://doi.org/10.1109/ISBI56570.2024.10635486

arXiv 2024
[6]

Esser, P., Rombach, R., Ommer, B.: Taming transformers for high-resolution image synthesis (2021), https://arxiv.org/abs/2012.09841

arXiv 2021
[7]

In: 2024 46th Annual International Con- ference of the IEEE Engineering in Medicine and Biology Society (EMBC)

Farooq, M.A., Yao, W., Schukat, M., Little, M.A., Corcoran, P.: Derm-t2im: Har- nessing synthetic skin lesion data via stable diffusion models for enhanced skin disease classification using vit and cnn. In: 2024 46th Annual International Con- ference of the IEEE Engineering in Medicine and Biology Society (EMBC). p. 1–5. IEEE (Jul 2024). https://doi.org/1...

arXiv 2024
[8]

JAMA Dermatology 153(5), 453–457 (05 2017)

Finnane, A., Curiel-Lewandrowski, C., Wimberley, G., Caffery, L., Katragadda, C., Halpern, A., Marghoob, A.A., Malvehy, J., Kittler, H., Hofmann-Wellenhof, R., Abraham, I., Soyer, H.P., behalf of the International Society of Digital Imaging of the Skin (ISDIS) for the International Skin Imaging Collabora- tion (ISIC), O.: Proposed technical guidelines for...

work page arXiv 2017

Show all 28 references

[9]

British Journal of Dermatology 190(6), 789–797 (02 2024)

Gordon, E.R., Trager, M.H., Kontos, D., Weng, C., Geskin, L.J., Dugdale, L.S., Samie, F.H.: Ethical considerations for artificial intelligence in dermatology: a scoping review. British Journal of Dermatology 190(6), 789–797 (02 2024). https://doi.org/10.1093/bjd/ljae040

2024 doi
[10]

Cancer Research77(21), e104–e107 (10 2017)

van Griethuysen, J.J., Fedorov, A., Parmar, C., Hosny, A., Aucoin, N., Narayan, V., Beets-Tan, R.G., Fillion-Robin, J.C., Pieper, S., Aerts, H.J.: Computational radiomics system to decode the radiographic phenotype. Cancer Research77(21), e104–e107 (10 2017). https://doi.org/1...

2017 doi
[11]

https://doi.org/10.48550/arXiv.1605.01397

Gutman, D., Codella, N.C.F., Celebi, E., Helba, B., Marchetti, M., Mishra, N., Halpern, A.: Skin lesion analysis toward melanoma detec- tion: A challenge at the international symposium on biomedical imaging (ISBI) 2016, hosted by the international skin imaging collaboration (I...

2016 doi arXiv
[12]

Heusel,M.,Ramsauer,H.,Unterthiner,T.,Nessler,B.,Hochreiter,S.:Ganstrained by a two time-scale update rule converge to a local nash equilibrium (2018)

2018
[13]

In: Advanced Intelligent Computing Technology and Applications: 20th International Confer- ence, ICIC 2024, Tianjin, China, August 5–8, 2024, Proceedings, Part VIII

Jing, X., Yang, S., Zhou, H., Wang, G., Mao, K.: Skindiff: A novel data synthesis method based on latent diffusion model for skin lesion segmentation. In: Advanced Intelligent Computing Technology and Applications: 20th International Confer- ence, ICIC 2024, Tianjin, China, Au...

2024 doi
[14]

In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion (CVPR)

Li, T., Chang, H., Mishra, S., Zhang, H., Katabi, D., Krishnan, D.: Mage: Masked generative encoder to unify representation learning and image synthesis. In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion (CVPR). pp. 2142–2152 (2023)

2023
[15]

Cancers14(7), 1778 (2022)

Li, Y., Ammari, S., Lawrance, L., Quillent, A., Assi, T., Lassau, N., Chouzenoux, E.: Radiomics-based method for predicting the glioma subtype as defined by tumor grade, idh mutation, and 1p/19q codeletion. Cancers14(7), 1778 (2022)

2022
[16]

Theranostics9(5), 1303 (2019)

Liu, Z., Wang, S., Dong, D., Wei, J., Fang, C., Zhou, X., Sun, K., Li, L., Li, B., Wang,M.,etal.:Theapplicationsofradiomicsinprecisiondiagnosisandtreatment of oncology: opportunities and challenges. Theranostics9(5), 1303 (2019)

2019
[17]

Journal of machine learning research 9(11) (2008) Controllable Skin Synthesis via Lesion-Focused Vector Autoregression Model 11

Van der Maaten, L., Hinton, G.: Visualizing data using t-sne. Journal of machine learning research 9(11) (2008) Controllable Skin Synthesis via Lesion-Focused Vector Autoregression Model 11

2008
[18]

https://doi.org/10.1109/EMBC.2013.6610779

Mendonca, T., Ferreira, P.M., Marques, J.S., Marcal, A.R.S., Rozeira, J.: PH² - a dermoscopic image database for research and benchmarking2013, 5437–5440. https://doi.org/10.1109/EMBC.2013.6610779

2013
[19]

van den Oord, A., Vinyals, O., Kavukcuoglu, K.: Neural discrete representation learning (2018)

2018
[20]

OpenAI blog1(8), 9 (2019)

Radford,A.,Wu,J.,Child,R.,Luan,D.,Amodei,D.,Sutskever,I.,etal.:Language models are unsupervised multitask learners. OpenAI blog1(8), 9 (2019)

2019
[21]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 10684– 10695 (June 2022)

2022
[22]

Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models (2022)

2022
[23]

In: Lee, D., Sugiyama, M., Luxburg, U., Guyon, I., Garnett, R

Salimans, T., Goodfellow, I., Zaremba, W., Cheung, V., Radford, A., Chen, X., Chen, X.: Improved techniques for training gans. In: Lee, D., Sugiyama, M., Luxburg, U., Guyon, I., Garnett, R. (eds.) Advances in Neural Information Pro- cessing Systems. vol. 29. Curran Associates,...

2016
[24]

Clinical Colorectal Cancer 20(1), 52–71 (2021)

Staal, F.C., van der Reijd, D.J., Taghavi, M., Lambregts, D.M., Beets-Tan, R.G., Maas, M.: Radiomics for the prediction of treatment outcome and survival in patients with colorectal cancer: A systematic review. Clinical Colorectal Cancer 20(1), 52–71 (2021). https://doi.org/10...

2021 doi
[25]

In: Globerson, A., Mackey, L., Belgrave,D.,Fan,A.,Paquet,U.,Tomczak,J.,Zhang,C.(eds.)AdvancesinNeural Information Processing Systems

Tian, K., Jiang, Y., Yuan, Z., PENG, B., Wang, L.: Visual autoregressive modeling: Scalable image generation via next-scale prediction. In: Globerson, A., Mackey, L., Belgrave,D.,Fan,A.,Paquet,U.,Tomczak,J.,Zhang,C.(eds.)AdvancesinNeural Information Processing Systems. vol. 37...

2024
[26]

https://doi.org/10.7910/DVN/DBW86T

Tschandl, P.: The HAM10000 dataset, a large collection of multi- source dermatoscopic images of common pigmented skin lesions (2018). https://doi.org/10.7910/DVN/DBW86T

2018 doi
[27]

Nature medicine26(8), 1229–1234 (2020)

Tschandl, P., Rinner, C., Apalla, Z., Argenziano, G., Codella, N., Halpern, A., Janda, M., Lallas, A., Longo, C., Malvehy, J., et al.: Human–computer collabora- tion for skin cancer recognition. Nature medicine26(8), 1229–1234 (2020)

2020
[28]

In: proceedings of Medical Image Computing and Computer Assisted Intervention – MICCAI 2024

Yuan, Z., Fang, Z., Huang, Z., Wu, F., Yao, Y.F., Li, Y.: Adapting Pre-trained Generative Model to Medical Image for Data Augmentation . In: proceedings of Medical Image Computing and Computer Assisted Intervention – MICCAI 2024. vol. LNCS 15005. Springer Nature Switzerland (O...

2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Bai, Y., Geng, X., Mangalam, K., Bar, A., Yuille, A., Darrell, T., Malik, J., Efros, A.A.: Sequential modeling enables scalable learning for large vision models (2023), https://arxiv.org/abs/2312.00785

arXiv 2023

[2] [2]

Color medical image analysis pp

Ballerini, L., Fisher, R.B., Aldridge, B., Rees, J.: A color and texture based hi- erarchical k-nn approach to the classification of non-melanoma skin lesions. Color medical image analysis pp. 63–86 (2013)

work page 2013

[3] [3]

National Science Review 11(12), nwae348 (10 2024)

Chen, M., Mei, S., Fan, J., Wang, M.: Opportunities and challenges of diffusion models for generative ai. National Science Review 11(12), nwae348 (10 2024). https://doi.org/10.1093/nsr/nwae348

doi:10.1093/nsr/nwae348 2024

[4] [4]

https://doi.org/10.48550/arXiv.1710.05006 10 J

Codella, N.C.F., Gutman, D., Celebi, M.E., Helba, B., Marchetti, M.A., Dusza, S.W., Kalloo, A., Liopyris, K., Mishra, N., Kittler, H., Halpern, A.: Skin lesion analysis toward melanoma detection: A challenge at the 2017 international sym- posium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC). https://doi.org/10...

arXiv doi:10.48550/arxiv.1710.05006 2017

[5] [5]

In: 2024 IEEE International Symposium on Biomedical Imaging (ISBI)

Du, S., Wang, X., Lu, Y., Zhou, Y., Zhang, S., Yuille, A., Li, K., Zhou, Z.: Boosting dermatoscopic lesion segmentation via diffusion models with visual and textual prompts. In: 2024 IEEE International Symposium on Biomedical Imaging (ISBI). pp. 1–5 (2024). https://doi.org/10.1109/ISBI56570.2024.10635486

arXiv 2024

[6] [6]

Esser, P., Rombach, R., Ommer, B.: Taming transformers for high-resolution image synthesis (2021), https://arxiv.org/abs/2012.09841

arXiv 2021

[7] [7]

In: 2024 46th Annual International Con- ference of the IEEE Engineering in Medicine and Biology Society (EMBC)

Farooq, M.A., Yao, W., Schukat, M., Little, M.A., Corcoran, P.: Derm-t2im: Har- nessing synthetic skin lesion data via stable diffusion models for enhanced skin disease classification using vit and cnn. In: 2024 46th Annual International Con- ference of the IEEE Engineering in Medicine and Biology Society (EMBC). p. 1–5. IEEE (Jul 2024). https://doi.org/1...

arXiv 2024

[8] [8]

JAMA Dermatology 153(5), 453–457 (05 2017)

Finnane, A., Curiel-Lewandrowski, C., Wimberley, G., Caffery, L., Katragadda, C., Halpern, A., Marghoob, A.A., Malvehy, J., Kittler, H., Hofmann-Wellenhof, R., Abraham, I., Soyer, H.P., behalf of the International Society of Digital Imaging of the Skin (ISDIS) for the International Skin Imaging Collabora- tion (ISIC), O.: Proposed technical guidelines for...

work page arXiv 2017