Pith. sign in

REVIEW 3 major objections 5 minor 26 references

Medical Large Vision Language Models with Multi-Image Visual Ability

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read The paper claims that an 83,200-question instruction dataset, Med-MIM, can teach medical vision-language models to reason across multiple images.

desk verdict Med-MIM is a genuinely new dataset for medical multi-image QA, but the paper's headline claims outrun its evaluation: the held-in benchmark is drawn from the training data and the held-out benchmarks share the same synthesis pipeline. read the letter →

arxiv 2505.19031 v1 pith:Q2RUWMI3 submitted 2025-05-25 cs.CV cs.AI

classification cs.CVcs.AI
keywords medicallargevision-languagemodelsmulti-imageunderstandinginstructiontuningvisualquestionansweringtemporalco-referencebenchmarkGPT-4ogeneration
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

Medical vision-language models answer questions about single images well, but clinical work often needs two or three images at once: a patient's previous and current chest X-ray, two mammographic views, several MRI sequences. This paper claims that a new instruction dataset called Med-MIM, with 83,200 question-answer pairs spanning four multi-image abilities, can teach medical LVLMs to handle such cases. The authors fine-tune two open-source models, Med-Mantis and MIM-LLaVA-Med, on this dataset and report that both beat six strong baselines on their held-in benchmark and improve on their base versions on two held-out benchmarks. If the claim holds, Med-MIM is a reusable resource for upgrading medical AI from single-image perception to multi-image clinical reasoning.

What carries the argument

The Med-MIM instruction dataset is the engine: 83.2K QA pairs in five imaging domains, organized into inherent and composed subsets, with each sample tagged to one of four visual abilities. Inherent samples are generated by GPT-4o from medical reports attached to naturally multi-image collections; composed samples are made by taking single-image LLaVA-Med QA pairs and adding location prefixes like 'in the first image.' The four ability labels drive both training and the held-in benchmark, while held-out benchmarks MIM-RAD and MIM-ODIR are built with the same generation recipe from VQA-RAD and ODIR. Interleaved image-text formatting and standard negative log-likelihood fine-tuning let the two base models ingest up to three images at a time.

What would settle it

Assemble a new multi-image medical QA test from a source never touched by the Med-MIM generation pipeline, such as fresh human-written questions about images from a different institution; if Med-Mantis and MIM-LLaVA-Med then show no advantage over their base models, the dataset's effect is distribution-specific rather than a general multi-image ability.

Watch

Extended reading notes

Core claim

The central claim is that instruction tuning with the Med-MIM dataset produces the gain: fine-tuning on 83.2K multi-image QA pairs converts a general multi-image model (Mantis) and a single-image medical model (LLaVA-Med) into Med-Mantis and MIM-LLaVA-Med, both of which outperform the original models and all open-source baselines on the Med-MIM benchmark. On the held-in subset, Med-Mantis reaches 74.86% close-ended accuracy on temporal questions and 80.40% on co-reference, while MIM-LLaVA-Med improves over LLaVA-Med by 35.44 points on temporal close-ended accuracy. On the held-out MIM-RAD and MIM-ODIR benchmarks, both fine-tuned models show consistent gains over their base versions, and MIM-LLaVA-Med attains the best open-source scores on MIM-ODIR. The paper interprets these results as evidence that the Med-MIM instruction dataset effectively enhances multi-image understanding in the medical domain.

Load-bearing premise

The claim depends on the benchmarks being genuinely different from the training data, so the reported gains measure transferable multi-image skill rather than memorization of near-duplicate examples.

Editorial extensions

If this is right

  • The Med-MIM instruction dataset can be reused as a training resource by any open-source medical LVLM, so future models can acquire multi-image capability without collecting new clinical image pairs.
  • The four-ability split gives researchers a standard way to diagnose where a medical LVLM fails: temporal change, multi-view reasoning, image comparison, or locating content across images.
  • The Med-MIM benchmark provides a quantitative target for medical multi-image QA, with close-ended accuracy and open-ended lexical scores, so model progress on this skill can be tracked over time.
  • The reported gains imply that supervised fine-tuning on multi-image instruction data is a viable path for open-source medical models to approach or exceed closed-source performance on held-in clinical multi-image questions.

Reading between the lines

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

  • A natural next experiment is to fine-tune a larger or stronger general-purpose multi-image backbone on Med-MIM to see whether the medical gains compound with model scale.
  • Because the composed subset is built by adding location prefixes to single-image QA pairs, it plausibly teaches co-reference mostly through surface phrasing; whether the skill survives rephrased location questions is an open testable question.
  • The same inherent/composed recipe could be applied to other paired medical sources, such as longitudinal CT or pathology slide series, to see whether the multi-image abilities generalize beyond chest X-ray and mammography.
  • The ablations leave the inherent-versus-composed mix unexplored factorially; a controlled experiment varying only that mix would show which data source carries each ability.
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.

Referee Report

3 major / 5 minor

Summary. The paper introduces Med-MIM, a medical multi-image instruction dataset of 83.2K QA pairs spanning four visual abilities (temporal understanding, reasoning, comparison, co-reference), and a corresponding Med-MIM benchmark with held-in and held-out subsets. The authors fine-tune LLaVA-Med and Mantis on Med-MIM to produce MIM-LLaVA-Med and Med-Mantis, and report that both achieve superior performance on held-in and held-out benchmark subsets, concluding that fine-tuning with Med-MIM significantly enhances multi-image visual abilities in the medical domain.

Significance. If the claims are substantiated, the Med-MIM dataset and benchmark would be a useful resource for training and evaluating medical LVLMs on multi-image reasoning, providing a taxonomy of abilities and two tuned model checkpoints. The paper is honest in reporting open-ended results and includes ablations on four visual abilities and dataset size. The held-out benchmarks (MIM-RAD, MIM-ODIR) are a constructive attempt to test generalization, and the positive deltas over base models in those benchmarks are suggestive. However, the evaluation design has a load-bearing weakness: the held-in benchmark is derived from the same instruction data used for training, and the held-out benchmarks are produced by the same GPT-4o synthesis pipeline as the training data, so the current evidence does not clearly separate learned generalization from memorization or style transfer.

major comments (3)
  1. [Section 2, 'Multi-image Visual Abilities Evaluation via Med-MIM Benchmark'] The held-in Med-MIM benchmark is described as 'derived from the Med-MIM instruction dataset,' yet the paper does not state that the benchmark examples are disjoint from the 83.2K instruction samples used for fine-tuning. No patient-level split, image-level de-duplication, or question-level filtering is reported. Consequently, the large held-in gains in Table 1 (e.g., MIM-LLaVA-Med temporal close +35.44, Med-Mantis co-reference close +44.69) could reflect memorization of exact or near-duplicate QA pairs rather than improved multi-image visual ability. The central conclusion that fine-tuning 'significantly enhances multi-image visual abilities' rests on these numbers, so a disjoint, de-duplicated evaluation split is necessary.
  2. [Section 2, 'Held-out part' and Section 3, Table 1] The held-out benchmarks MIM-RAD and MIM-ODIR are constructed using the same generation procedures as the Med-MIM instruction dataset: MIM-RAD uses the composed-multi-image procedure with location suffixes, and MIM-ODIR applies the same GPT-4o prompt/refinement pipeline as the inherent Med-MIM data. This makes the held-out evaluation distribution closely resemble the training distribution in format and synthetic answer style. The reported improvements over base models on MIM-RAD and MIM-ODIR may thus partly reflect adaptation to the GPT-4o answer format rather than transferable medical multi-image reasoning. The paper should either use existing independently constructed medical multi-image benchmarks or explicitly demonstrate that performance holds on benchmarks not generated by the Med-MIM pipeline.
  3. [Section 3, Table 1 and ablation figures] All reported numbers are from single runs without error bars or significance tests. The open-ended subsets are very small (e.g., 30 examples each for temporal and reasoning on the held-in benchmark), so observed deltas of a few points can easily be within random variation. For example, the MIM-LLaVA-Med vs. LLaVA-Med held-out MIM-RAD open improvement is only +0.22 percentage points, and several other open deltas are in the 2–7 point range. The paper should report variance across multiple seeds (or bootstrap confidence intervals) and should note the small sample size caveat in the discussion.
minor comments (5)
  1. [Abstract and Introduction] The paper states that Med-MIM is 'the first focused effort on medical multi-image analysis'; this is a strong claim that would benefit from a more careful comparison with prior multi-image medical VLM works beyond the references cited, such as recent multi-image medical LVLM papers that may have appeared by the time of submission.
  2. [Section 2, 'Instruction Tuning'] Equation (1) is a standard next-token NLL loss; the notation p(r_k|I, q_{1:S}, r_{1:k-1}) is clear, but the sentence immediately after the equation repeats the definition. Consider tightening this paragraph.
  3. [Section 3, 'Ablation on Four Visual Abilities'] Fig. 4(c) reports results for '0%, 25%, 50%, 75%, and 100%' of the instruction dataset. The 0% point is presumably the base model without fine-tuning; this should be stated explicitly in the caption or text to avoid ambiguity.
  4. [Section 2, 'Co-reference'] The co-reference category is applied to both the inherent and composed subsets, but the inherent co-reference examples ask about image location (e.g., 'which image... represents the view that captures the breast from above?'), which is closer to spatial/view understanding than co-reference as defined in the natural-image MANTIS taxonomy. The paper should clarify the relationship or justify the category naming.
  5. [General] The manuscript does not include a limitations section. Given the evaluation overlap concerns above, a short paragraph stating the potential distribution shift between synthetic GPT-4o-generated QA pairs and real clinical multi-image questions, and the lack of error bars, would improve scientific transparency.

Circularity Check

1 steps flagged · score 6.0 of 10

Held-in benchmark is derived from the Med-MIM instruction dataset used for fine-tuning, with no described split; its gains are in-sample fits, not independent predictions.

  1. fitted input called prediction [Sec. 2, 'Multi-image Visual Abilities Evaluation via Med-MIM Benchmark' and Sec. 2, 'Instruction Tuning with Med-MIM Instruction Dataset']
    "To comprehensively evaluate four multi-image visual abilities, we construct the held-in Med-MIM Benchmark derived from the Med-MIM instruction dataset, which includes 2,968 closed-type examples (903, 454, 208, and 1,403 for temporal, reasoning, comparison, co-reference abilities, respectively) and 256 open-type examples (30, 30, 136, and 60, respectively)."

    Both models are fine-tuned on the Med-MIM instruction dataset for 3 epochs, and the held-in benchmark is 'derived from' that same dataset. The paper provides no train/test split, disjointness guarantee, or de-duplication, so the large held-in gains (e.g., MIM-LLaVA-Med temporal close +35.44, Med-Mantis co-reference close +44.69) can be memorization of the training QA pool. Using these in-sample scores as evidence that the dataset 'significantly enhances multi-image visual abilities' is calling a fitted input a prediction: the evaluation distribution is, by construction, the training distribution.

full rationale

The paper's central generalization claim is only partly supported. The held-in portion of the Med-MIM benchmark is explicitly derived from the same Med-MIM instruction dataset used for fine-tuning, and no split is reported, so those results are not independent evidence. The held-out MIM-RAD and MIM-ODIR benchmarks use external image sources, which provides some independent signal; however, they are synthesized with the same GPT-4o procedure as the training data, so format/style overlap remains a confound. There is no load-bearing self-citation chain or imported uniqueness theorem. Because the largest and most emphasized held-in gains reduce to in-sample performance, but held-out external-data results provide some non-circular content, the appropriate score is 6.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the validity of GPT-4o-generated QA pairs as clinical ground truth, on the transfer of the four-ability taxonomy to medicine, and on benchmark sets that are not decontaminated from the training data. Construction thresholds (9K temporal, 4.4K reasoning, sequence length 3, epochs 3) are hand-set without sensitivity analysis.

free parameters (4)
  • maximum multi-image sequence length = 3
    Capped at three images to balance dataset richness against fine-tuning compute constraints (Section 2).
  • temporal subset size = 9K
    Filtered from LUMIERE and EMBED without stated inclusion criteria (Section 2).
  • reasoning subset size = 4.4K
    Restricted to two-view mammography diagnosis in EMBED (Section 2).
  • fine-tuning epochs = 3
    Chosen for fine-tuning on 4 A40 GPUs over 72 hours; no early-stopping or hyperparameter search reported (Section 2).
assumptions (5)
  • domain assumption GPT-4o generates medically accurate QA pairs from radiology reports without seeing the images (language-only prompting).
    Section 2: 'we prompt language-only GPT-4o to generate multi-image QA pairs by leveraging the medical reports'. If reports and images disagree or reports contain errors, the QA labels are wrong; human review is mentioned but not quantified.
  • ad hoc to paper The Mantis four-ability taxonomy (temporal, reasoning, comparison, co-reference) is appropriate for medical multi-image tasks.
    The paper adopts the taxonomy from natural-image work [8] but does not validate that these four abilities cover or correctly partition clinical multi-image reasoning (Section 2).
  • ad hoc to paper The held-in benchmark, derived from the Med-MIM instruction dataset, is a valid out-of-sample evaluation set.
    Section 2: held-in benchmark is 'derived from the Med-MIM instruction dataset'; no train/test split or deduplication is described, so its validity rests on unstated separation.
  • ad hoc to paper MS-CXR-T tasks are treated as comparison rather than temporal understanding even though the source data is temporal.
    Section 2: 'although MS-CXR-T contains temporal relations, the actual task is to determine progression status... which requires the model to find differences between two visits.' This reclassification is a modeling choice that affects subset composition.
  • domain assumption Human quality control of GPT-4o outputs is sufficient.
    Section 2 mentions 'human reviewers as a quality control measure' but gives no count, agreement, or audit procedure.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Medical Large Vision Language Models with Multi-Image Visual Ability." pith.science (2026). https://pith.science/paper/Q2RUWMI3

@misc{pith2026250519031,
  author       = {Pith},
  title        = {Pith review of: Medical Large Vision Language Models with Multi-Image Visual Ability},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q2RUWMI3}},
  note         = {Machine review of arXiv:2505.19031}
}
read the original abstract

Medical large vision-language models (LVLMs) have demonstrated promising performance across various single-image question answering (QA) benchmarks, yet their capability in processing multi-image clinical scenarios remains underexplored. Unlike single image based tasks, medical tasks involving multiple images often demand sophisticated visual understanding capabilities, such as temporal reasoning and cross-modal analysis, which are poorly supported by current medical LVLMs. To bridge this critical gap, we present the Med-MIM instruction dataset, comprising 83.2K medical multi-image QA pairs that span four types of multi-image visual abilities (temporal understanding, reasoning, comparison, co-reference). Using this dataset, we fine-tune Mantis and LLaVA-Med, resulting in two specialized medical VLMs: MIM-LLaVA-Med and Med-Mantis, both optimized for multi-image analysis. Additionally, we develop the Med-MIM benchmark to comprehensively evaluate the medical multi-image understanding capabilities of LVLMs. We assess eight popular LVLMs, including our two models, on the Med-MIM benchmark. Experimental results show that both Med-Mantis and MIM-LLaVA-Med achieve superior performance on the held-in and held-out subsets of the Med-MIM benchmark, demonstrating that the Med-MIM instruction dataset effectively enhances LVLMs' multi-image understanding capabilities in the medical domain.

Figures

Figures reproduced from arXiv: 2505.19031 by the authors.

Figure 1
Figure 1. Illustration of the pipeline to generate Med-MIM dataset and benchmark. ized medical VLMs: MIM-LLaVA-Med and Med-Mantis, tailored for multi-image analysis. We compare these models against several state-of-the-art VLMs, and experimental results demonstrate that our models exhibit superior multi-image visual capabilities on both the held-in and held-out Med-MIM benchmarks. 2 Method We create the Med-MIM Instruction da… view at source ↗
Figure 2
Figure 2. Comprehensive analysis of the Med-MIM dataset: (a) Image modality com￾position; (b) Distribution of the question types; (c) Inherent multi-image dataset distribution categorized by four core visual abilities. of medical images, we construct a composed multi-image dataset by manually grouping multiple images from the LLaVA-Med VQA [12] dataset. We cap the length of multi-image sequences at a maximum of three, allowin… view at source ↗
Figure 3
Figure 3. Example of generated multi-image QA pair in the MS-CXR-T dataset. as well. To encode proper location information, the original single-image QA pairs are updated by adding prefixes such as “In the first image” or “In the second image.” (as illustrated in Fig.1). Instruction Dataset Generation. After collecting the multi-image sets, we in￾tegrate the textual data for each image and generate the corresponding question￾… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Ablation results on (a) four visual ability subsets on the Med-Mantis, (b) the performance of Med-Mantis and MIM-LLaVA-Med with and without the composed Med-MIM instruction dataset, and (c) the impact of different sizes of the Med-MIM instruction dataset on the Med-Man…
Figure 5
Figure 5. Figure 5: Case studies for LVLMs in medical multi-image understanding. results indicate that Med-Mantis and MIM-LLaVA-Med can produce more ac￾curate response compared with original Mantis and LLaVA-Med models. 4 Conclusion In this work, we present the Med-MIM instruction dataset…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

26 extracted references · 14 canonical work pages

  1. [1]

    Cancers 15(2), 545 (2023)

    Alleman, K., Knecht, E., Huang, J., Zhang, L., Lam, S., DeCuypere, M.: Multi- modal deep learning-based prognostication in glioma patients: a systematic review. Cancers 15(2), 545 (2023)

  2. [2]

    arXiv preprint arXiv:2404.00578 (2024)

    Bai, F., Du, Y., Huang, T., Meng, M.Q.H., Zhao, B.: M3d: Advancing 3d medical image analysis with multi-modal large language models. arXiv preprint arXiv:2404.00578 (2024)

  3. [3]

    Bannur, S., Hyland, S., Liu, Q., Pérez-García, F., Ilse, M., de Castro, D.C., Boeck- ing, B., Sharma, H., Bouzid, K., Schwaighofer, A., et al.: Ms-cxr-t: Learning to exploit temporal structure for biomedical vision-language processing (2023)

  4. [4]

    Biological psychiatry: cognitive neuroscience and neuroimaging1(3), 230–244 (2016)

    Calhoun, V.D., Sui, J.: Multimodal fusion of brain imaging data: a key to find- ing the missing link (s) in complex mental illness. Biological psychiatry: cognitive neuroscience and neuroimaging1(3), 230–244 (2016)

  5. [5]

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

    Chen, Z., Wu, J., Wang, W., Su, W., Chen, G., Xing, S., Zhong, M., Zhang, Q., Zhu, X., Lu, L., et al.: Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 24185–24198 (2024)

  6. [6]

    arXiv preprint arXiv:2410.21276 (2024)

    Hurst, A., Lerer, A., Goucher, A.P., Perelman, A., Ramesh, A., Clark, A., Os- trow, A., Welihinda, A., Hayes, A., Radford, A., et al.: Gpt-4o system card. arXiv preprint arXiv:2410.21276 (2024)

  7. [7]

    Radiology: Artificial Intelligence5(1), e220047 (2023)

    Jeong, J.J., Vey, B.L., Bhimireddy, A., Kim, T., Santos, T., Correa, R., Dutt, R., Mosunjac, M., Oprea-Ilies, G., Smith, G., et al.: The emory breast imaging dataset (embed): A racially diverse, granular dataset of 3.4 million screening and diagnostic mammographic images. Radiology: Artificial Intelligence5(1), e220047 (2023)

  8. [8]

    Transactions on Machine Learning Research 2024 (2024), https://openreview.net/forum?id=skLtdUVaJa

    Jiang, D., He, X., Zeng, H., Wei, C., Ku, M.W., Liu, Q., Chen, W.: Mantis: Inter- leaved multi-image instruction tuning. Transactions on Machine Learning Research 2024 (2024), https://openreview.net/forum?id=skLtdUVaJa

Show all 26 references
  1. [9]

    Archives of neurology66(10), 1254–1259 (2009)

    Johnson, D.K., Storandt, M., Morris, J.C., Galvin, J.E.: Longitudinal study of the transition from healthy aging to alzheimer disease. Archives of neurology66(10), 1254–1259 (2009)

  2. [10]

    Nature Machine Intelligence3(4), 288–298 (2021)

    Korot, E., Guan, Z., Ferraz, D., Wagner, S.K., Zhang, G., Liu, X., Faes, L., Pon- tikos, N., Finlayson, S.G., Khalid, H., et al.: Code-free deep learning for multi- modality medical image classification. Nature Machine Intelligence3(4), 288–298 (2021)

  3. [11]

    Scientific data 5(1), 1–10 (2018)

    Lau, J.J., Gayen, S., Ben Abacha, A., Demner-Fushman, D.: A dataset of clinically generated visual questions and answers about radiology images. Scientific data 5(1), 1–10 (2018)

  4. [12]

    Advances in Neural Information Processing Systems36 (2024)

    Li, C., Wong, C., Zhang, S., Usuyama, N., Liu, H., Yang, J., Naumann, T., Poon, H., Gao, J.: Llava-med: Training a large language-and-vision assistant for biomedicine in one day. Advances in Neural Information Processing Systems36 (2024)

  5. [13]

    In: Benchmarking, Measur- ing, and Optimizing: Third BenchCouncil International Symposium, Bench 2020, 10 X

    Li, N., Li, T., Hu, C., Wang, K., Kang, H.: A benchmark of ocular disease intelli- gent recognition: One shot for multi-disease detection. In: Benchmarking, Measur- ing, and Optimizing: Third BenchCouncil International Symposium, Bench 2020, 10 X. Yang et al. Virtual Event, No...

  6. [14]

    Advances in neural information processing systems36 (2024)

    Liu, H., Li, C., Wu, Q., Lee, Y.J.: Visual instruction tuning. Advances in neural information processing systems36 (2024)

  7. [15]

    URL https://arxiv

    Liu, Z., Chu, T., Zang, Y., Wei, X., Dong, X., Zhang, P., Liang, Z., Xiong, Y., Qiao, Y., Lin, D., et al.: Mmdu: A multi-turn multi-image dialog understanding benchmark and instruction-tuning dataset for lvlms, 2024c. URL https://arxiv. org/abs/2406.11833

  8. [16]

    arXiv preprint arXiv:2403.05525 (2024)

    Lu, H., Liu, W., Zhang, B., Wang, B., Dong, K., Liu, B., Sun, J., Ren, T., Li, Z., Yang, H., et al.: Deepseek-vl: towards real-world vision-language understanding. arXiv preprint arXiv:2403.05525 (2024)

  9. [17]

    arXiv preprint arXiv:2408.02718 (2024)

    Meng, F., Wang, J., Li, C., Lu, Q., Tian, H., Liao, J., Zhu, X., Dai, J., Qiao, Y., Luo, P., et al.: Mmiu: Multimodal multi-image understanding for evaluating large vision-language models. arXiv preprint arXiv:2408.02718 (2024)

  10. [18]

    In: Machine Learning for Health (ML4H)

    Moor, M., Huang, Q., Wu, S., Yasunaga, M., Dalmia, Y., Leskovec, J., Zakka, C., Reis, E.P., Rajpurkar, P.: Med-flamingo: a multimodal medical few-shot learner. In: Machine Learning for Health (ML4H). pp. 353–367. PMLR (2023)

  11. [19]

    Nature communications 13(1), 4566 (2022)

    Poulakis, K., Pereira, J.B., Muehlboeck, J.S., Wahlund, L.O., Smedby, Ö., Volpe, G., Masters, C.L., Ames, D., Niimi, Y., Iwatsubo, T., et al.: Multi-cohort and longitudinal bayesian clustering study of stage and subtype in alzheimer’s disease. Nature communications 13(1), 4566 (2022)

  12. [20]

    BMC Pulmonary Medicine 20, 1–9 (2020)

    Rousan,L.A.,Elobeid,E.,Karrar,M.,Khader,Y.:Chestx-rayfindingsandtempo- ral lung changes in patients with covid-19 pneumonia. BMC Pulmonary Medicine 20, 1–9 (2020)

  13. [21]

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

    Seyfioglu, M.S., Ikezogwo, W.O., Ghezloo, F., Krishna, R., Shapiro, L.: Quilt- llava: Visual instruction tuning by extracting localized narratives from open-source histopathology videos. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp....

  14. [22]

    In: The Thirty-eighth An- nual Conference on Neural Information Processing Systems

    Srivastav,V.,Navab,N.,Padoy,N.,etal.:Procedure-awaresurgicalvideo-language pretraining with hierarchical knowledge augmentation. In: The Thirty-eighth An- nual Conference on Neural Information Processing Systems

  15. [23]

    Scientific data9(1), 768 (2022)

    Suter, Y., Knecht, U., Valenzuela, W., Notter, M., Hewer, E., Schucht, P., Wiest, R., Reyes, M.: The lumiere dataset: Longitudinal glioblastoma mri with expert rano evaluation. Scientific data9(1), 768 (2022)

  16. [24]

    arXiv preprint arXiv:2308.02463 (2023)

    Wu, C., Zhang, X., Zhang, Y., Wang, Y., Xie, W.: Towards generalist foundation model for radiology. arXiv preprint arXiv:2308.02463 (2023)

  17. [25]

    In: Proceedings of the 41st International Conference on Machine Learning

    Yang, J., Su, B., Zhao, W.X., Wen, J.R.: Unlocking the power of spatial and temporal information in medical multimodal pre-training. In: Proceedings of the 41st International Conference on Machine Learning. ICML’24, JMLR.org (2024)

  18. [26]

    arXiv preprint arXiv:2305.10415 (2023)

    Zhang, X., Wu, C., Zhao, Z., Lin, W., Zhang, Y., Wang, Y., Xie, W.: Pmc-vqa: Visual instruction tuning for medical visual question answering. arXiv preprint arXiv:2305.10415 (2023)

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.