Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

MedSeg-R: Reasoning Segmentation in Medical Images with Multimodal Large Language Models

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read MedSeg-R turns implicit clinical questions into text answers plus precise segmentation masks in a single end-to-end model.

desk verdict MedSeg-R is a LISA-style model for a new medical reasoning segmentation task, but the MedSeg-QA dataset's [SEG]-to-mask grounding is unverified, making the central claim conditional. read the letter →

arxiv 2506.10465 v1 pith:DGZ6DPJ2 submitted 2025-06-12 cs.CV

classification cs.CV
keywords medicalimagereasoningsegmentationmultimodallargelanguagemodelpixel-levelgroundingvisualquestionansweringMedSeg-QAdataset
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

MedSeg-R introduces and addresses a new task, medical image reasoning segmentation, where a model receives an implicit, open-ended clinical question about an image and must return both a reasoned text response and a segmentation mask of the structure referenced in that response. The paper argues that existing segmentation models require explicit object names, while medical visual question-answering models lack pixel-level grounding, and that a single end-to-end framework can bridge both gaps. To support this, the authors propose an architecture that fuses a multimodal large language model with a SAM-style mask decoder, and they build MedSeg-QA, a dataset of over ten thousand image-mask-conversation triples generated by GPT-4 and refined by physicians. Reported experiments on VQA-RAD, SLAKE, PathVQA, and FLARE 2022 suggest the resulting model achieves state-of-the-art or competitive performance on both biomedical question answering and segmentation, while also demonstrating the new reasoning-segmentation behavior qualitatively.

What carries the argument

The load-bearing mechanism is the [SEG] special token chain: the CLIP-based global visual encoder and Mistral-7B LLM (the global context understanding module) produce a text response containing <p>...</p> [SEG] placeholders, and then a SAM-based grounding encoder with a mask decoder converts those placeholders into pixel masks. The text response is thus the bridge that connects high-level reasoning to pixel-level output, and the training objective combines text cross-entropy loss with binary cross-entropy and DICE losses on the mask.

What would settle it

A direct falsifier is to sample conversations from MedSeg-QA, extract every <p>...</p> [SEG] phrase, and check against the ground-truth mask whether the named structure is actually present and the mask is the correct one; if mismatches are frequent, the trained mask decoder may rely on dataset artifacts rather than true reasoning grounding.

Watch

Extended reading notes

Core claim

The paper's central claim is that, given a medical image and an implicit instruction such as "What might this test reveal about the patient's underlying conditions?", a model can simultaneously generate a coherent textual explanation and a precise segmentation mask of the abnormal region it mentions. MedSeg-R realizes this by letting the LLM produce intermediate tokens and a text response that embeds special tokens of the form '<p> covid-19 </p> [SEG]', which the mask decoder then uses to generate the corresponding mask. The authors further claim that MedSeg-QA, a dataset of over 10,000 physician-refined image-mask-conversation triples, is sufficient to train this reasoning-segmentation capability, and that the same model remains competitive on standard biomedical VQA and segmentation benchmarks.

Load-bearing premise

The central assumption is that every phrase GPT-4 wraps in [SEG] tokens during dataset creation actually corresponds to one of the ground-truth masks in the source segmentation dataset, because the paper does not describe any verification step checking that alignment.

Editorial extensions

If this is right

  • Physicians could ask open-ended diagnostic questions and receive both a written explanation and a localized mask of the region being discussed, without naming the structure explicitly.
  • A single MedSeg-R model can perform standard medical VQA, standard segmentation, and the new reasoning-segmentation task, so one system replaces multiple task-specific components.
  • The MedSeg-QA dataset provides a public resource for training and benchmarking reasoning segmentation across CT, histology, and dermoscopy images.
  • Multi-round conversation support means follow-up questions can refine both the textual diagnosis and the associated mask interactively.
  • Competitive results on FLARE 2022 suggest that adding LLM-based reasoning does not degrade pixel-level segmentation accuracy on standard organs.

Reading between the lines

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

  • If the [SEG]-to-mask alignment in MedSeg-QA is later verified to be accurate, the same three-stage pipeline could be applied to additional modalities such as MRI or ultrasound by changing only the dataset-specific caption prefixes.
  • The model's robustness to paraphrased clinical questions is untested; a stress test with rephrased versions of the same question would reveal whether the reasoning behavior generalizes beyond the training phrasing distribution.
  • A testable extension is to regenerate the dataset with an open-weight LLM instead of GPT-4; if MedSeg-R retains its performance, the annotation pipeline is not tied to a specific commercial model.
  • The reported segmentation numbers on FLARE 2022 are close to nnU-Net, so the practical advantage of the reasoning framework may matter more for implicit-instruction settings than for standard organ segmentation.
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

4 major / 5 minor

Summary. The paper introduces MedSeg-R, an end-to-end multimodal large language model for a new task called medical image reasoning segmentation, where the model must answer an implicit or complex clinical question and simultaneously output a segmentation mask for the structure referenced in its answer. The architecture combines a CLIP ViT-H visual encoder and Mistral-7B LLM with a SAM-based grounding encoder and mask decoder, trained with a weighted text and segmentation loss. The authors also introduce MedSeg-QA, a dataset of over 10,000 image-mask pairs with multi-turn conversations generated by GPT-4 and refined by physician review. Experiments report results on standard biomedical VQA benchmarks (VQA-RAD, SLAKE, PathVQA) and on the FLARE 2022 segmentation benchmark, plus qualitative examples of reasoning segmentation on MedSeg-QA.

Significance. If the central claims hold, the paper addresses a meaningful gap: extending reasoning segmentation from natural images to medical images, where implicit clinical questions require both textual interpretation and pixel-level grounding. The architecture is a reasonable adaptation of established reasoning-segmentation designs such as LISA and GLAMM, and the proposed MedSeg-QA dataset could be a useful community resource if its grounding quality is verified. The paper includes qualitative demonstrations across CT, dermatoscopic, and histological modalities that are suggestive of the intended behavior. However, the current evidence is insufficient to support the paper's stronger state-of-the-art claims: the novel reasoning segmentation task is evaluated only qualitatively, the grounding between [SEG] spans and masks is not verified, and the numerical comparisons lack statistical rigor. The central idea is defensible, but the load-bearing evidence needs substantial additional work.

major comments (4)
  1. [Section 2.3] The dataset construction pipeline never verifies that a [SEG] mention inserted by GPT-4 actually refers to the annotated structure in the source segmentation mask. In the Image Caption Generation stage, GPT-4 receives only the image and a dataset-specific prefix, not the mask or its label. In the Conversations Generation stage, GPT-4 is prompted to insert [SEG] based on refined captions. The physician review checks caption textual accuracy, not the alignment between the <p>...</p> span and the mask. Since source datasets such as Medical Segmentation Decathlon and EBHI-Seg can contain multiple annotated structures per image or structures not mentioned in a generated caption, the model may be trained on mismatched text-mask pairs. This is load-bearing because the central claim that MedSeg-R learns to ground a referenced structure in pixels depends entirely on the correctness of these triples. The authors should add a verification step (e.g., automatic label-to-span matching followed by human review) and report the agreement rate on a sample.
  2. [Section 3.1] The reasoning segmentation task introduced in Section 2.1 is never evaluated quantitatively. Section 3.1 shows only qualitative examples in Figure 3, with no metrics on the MedSeg-QA validation set, no comparison against LISA or other reasoning segmentation baselines, and no evaluation of whether the generated mask corresponds to the noun phrase in the generated text. Without such an evaluation, the claim in the Conclusion that MedSeg-R 'surpasses existing models in segmentation accuracy and reasoning' is unsupported for the very task the paper proposes. The authors should add quantitative reasoning-segmentation metrics (e.g., cIoU or mask accuracy conditioned on the referenced text span) and compare with at least one reasoning-segmentation baseline.
  3. [Table 4] The FLARE 2022 experiment does not evaluate the proposed reasoning segmentation task; it uses an explicit instruction of the form 'Please segment the <class-name> in the medical image' and the fixed response 'Sure, it is [SEG].' This reduces to standard supervised segmentation and does not test implicit instruction understanding. Moreover, the reported DSC for MedSeg-R (0.8718±0.1034) overlaps with nnU-Net (0.8615±0.0790) within one standard deviation, and no significance test is reported, so the claim of superiority over state-of-the-art segmentation methods is not established. The authors should either temper the claim or provide statistical tests (e.g., paired bootstrapping over cases).
  4. [Table 3] The VQA results in Table 3 are reported without error bars, number of runs, or significance tests. Given that differences between methods are often a few points (e.g., MedSeg-R 72.90 vs. VL Encoder–Decoder 71.49 on VQA-RAD Open), single-run results do not support the state-of-the-art claim. The table also contains apparent data issues: the VL Encoder–Decoder row lists the identical Open score 71.49 for both VQA-RAD and PathVQA, and the LLaVA-Med SLAKE Open entry '87.1186.78' appears to be two numbers concatenated without a separator. The authors should correct these entries and provide variance or confidence intervals.
minor comments (5)
  1. [Table 3] The LLaVA-Med SLAKE Open column reads '87.1186.78', which is missing a separator and is ambiguous; it should be '87.11 86.78' or similar.
  2. [Section 2.2] The phrase 'then we first detail the architecture' contains a redundant 'first' and should be reworded for clarity.
  3. [Section 2.3] The paper describes the dataset as containing 'over 10,000 image-mask pairs' but does not state the exact number, the per-modality breakdown, or the number of conversations; this information should be provided in a dataset statistics table alongside Table 2.
  4. [General] The paper does not mention whether the MedSeg-QA dataset and MedSeg-R code will be released; given the dataset's central role, a release statement and a URL would improve reproducibility.
  5. [Section 1] The reference list includes several entries cited only by arXiv number (e.g., [5], [12], [13]) without full venue information where available; the authors should complete the bibliographic details.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the model's central claims are grounded in external benchmarks (VQA-RAD, SLAKE, PathVQA, FLARE 2022), and the MedSeg-QA evaluation is a standard held-out split rather than a derivation that reduces to its own inputs.

full rationale

I walked the paper's derivation chain: the task definition (Sec. 2.1) is a task formulation, not a fitted quantity; the MedSeg-R architecture (Sec. 2.2) is a straightforward combination of a CLIP visual encoder, Mistral-7B LLM, and a SAM-style mask decoder trained with text cross-entropy and mask BCE/DICE losses; the dataset pipeline (Sec. 2.3) uses GPT-4 to generate captions and conversations while masks come from existing segmentation datasets. No equation in the paper defines a prediction in terms of the quantity it claims to predict, and no load-bearing argument reduces to a self-citation. The external evaluations on VQA-RAD, SLAKE, PathVQA, and FLARE 2022 are independent of MedSeg-QA and provide separate support for the model's text and segmentation capabilities. The concern that [SEG]-span-to-mask alignment in MedSeg-QA is unverified is a legitimate data-quality and correctness risk, but it is not circularity: the model is trained on the pairs as given, and the held-out MedSeg-QA split tests generalization from that same distribution, which is standard supervised evaluation rather than a construction that forces the reported outcome. The qualitative reasoning demonstrations on MedSeg-QA are partly self-referential because the test conversations were produced by the same GPT-4 pipeline that generated the training conversations, but this affects the strength of the reasoning claim, not the logical circularity of the derivation. No claims here are equivalent to their inputs by construction.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claims rest on the quality and alignment of the automatically generated MedSeg-QA dataset and on reused public model components. The objective weights and prompt prefixes are chosen by hand but not reported, and there are no new physical or mathematical entities.

free parameters (2)
  • lambda_t and lambda_m = not reported
    Objective weights in L = lambda_t * Lt + lambda_m * Lm; values affect balance between text and mask loss but are not specified.
  • dataset-specific GPT-4 prompt prefixes = not reported in full
    Custom prefixes engineered per source dataset in the caption generation stage; these shape the generated captions and thus the conversation data.
assumptions (3)
  • domain assumption GPT-4-generated captions and conversations, after physician review, are clinically accurate and correctly linked to the segmentation masks of the source datasets.
    Section 2.3 relies on this for dataset quality; if the [SEG] phrases do not faithfully correspond to the mask labels, the mask decoder is trained on mismatched pairs.
  • domain assumption A medical image can be represented by paired image, text instruction, text response with [SEG] tokens, and one or more segmentation masks, and that this training objective transfers to unseen medical modalities.
    The task definition in Section 2.1 assumes such pairs capture clinical reasoning and grounding.
  • domain assumption Standard neural architectures (CLIP ViT-H, Mistral-7B, SAM ViT-H) are used as given and are sufficient for the task.
    The method section uses these pretrained components without modification, so their adequacy is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MedSeg-R: Reasoning Segmentation in Medical Images with Multimodal Large Language Models." pith.science (2026). https://pith.science/paper/DGZ6DPJ2

@misc{pith2026250610465,
  author       = {Pith},
  title        = {Pith review of: MedSeg-R: Reasoning Segmentation in Medical Images with Multimodal Large Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DGZ6DPJ2}},
  note         = {Machine review of arXiv:2506.10465}
}
read the original abstract

Medical image segmentation is crucial for clinical diagnosis, yet existing models are limited by their reliance on explicit human instructions and lack the active reasoning capabilities to understand complex clinical questions. While recent advancements in multimodal large language models (MLLMs) have improved medical question-answering (QA) tasks, most methods struggle to generate precise segmentation masks, limiting their application in automatic medical diagnosis. In this paper, we introduce medical image reasoning segmentation, a novel task that aims to generate segmentation masks based on complex and implicit medical instructions. To address this, we propose MedSeg-R, an end-to-end framework that leverages the reasoning abilities of MLLMs to interpret clinical questions while also capable of producing corresponding precise segmentation masks for medical images. It is built on two core components: 1) a global context understanding module that interprets images and comprehends complex medical instructions to generate multi-modal intermediate tokens, and 2) a pixel-level grounding module that decodes these tokens to produce precise segmentation masks and textual responses. Furthermore, we introduce MedSeg-QA, a large-scale dataset tailored for the medical image reasoning segmentation task. It includes over 10,000 image-mask pairs and multi-turn conversations, automatically annotated using large language models and refined through physician reviews. Experiments show MedSeg-R's superior performance across several benchmarks, achieving high segmentation accuracy and enabling interpretable textual analysis of medical images.

Figures

Figures reproduced from arXiv: 2506.10465 by the authors.

Figure 1
Figure 1. Different tasks in medical image analysis. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. MedSeg-R’s architecture. This figure illustrates our model’s ability of uti￾lizing the reasoning capabilities of the large language model (LLM) to produce detailed text responses y and guide the mask decoder in generating precise segmentation masks m. 2 Method In this section, we first define the medical image reasoning segmentation task in Sec. 2.1, then we first detail the architecture and training objectives of M… view at source ↗
Figure 3
Figure 3. Qualitative evaluation of MedSeg-R on the MedSeg-QA dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. MediRound: Multi-Round Entity-Level Reasoning Segmentation in Medical Images

    cs.CV 2025-11 conditional novelty 5.0 of 10

    MediRound introduces a multi-round, entity-level medical segmentation task, a 177K-dialogue dataset built from SA-Med2D-20M with GPT-5, and a LLaVA-Med/MedSAM baseline whose inference-time judgment-and-correction modu...

Reference graph

Works this paper leans on

32 extracted references · 17 canonical work pages · cited by 1 Pith paper

  1. [1]

    arXiv preprint arXiv:2303.08774 (2023)

    Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F.L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al.: Gpt-4 technical report. arXiv preprint arXiv:2303.08774 (2023)

  2. [2]

    Nature communications13(1), 4128 (2022)

    Antonelli, M., Reinke, A., Bakas, S., Farahani, K., Kopp-Schneider, A., Landman, B.A., Litjens, G., Menze, B., Ronneberger, O., Summers, R.M., et al.: The medical segmentation decathlon. Nature communications13(1), 4128 (2022)

  3. [3]

    Bioengineering (2023)

    Bazi, Y., Rahhal, M.M.A., Bashmal, L., Zuair, M.: Vision–language model for visual question answering in medical imagery. Bioengineering (2023)

  4. [4]

    Cheng, B., Misra, I., Schwing, A.G., Kirillov, A., Girdhar, R.: Masked-attention mask transformer for universal image segmentation (2022)

  5. [5]

    arXiv:2305.06500 (2023)

    Dai, W., Li, J., Li, D., Tiong, A.M.H., Zhao, J., Wang, W., Li, B., Fung, P., Hoi, S.: Instructblip: Towards general-purpose vision-language models with instruction tuning. arXiv:2305.06500 (2023)

  6. [6]

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., Houlsby, N.: An image is worth 16x16 words: Transformers for image recognition at scale (2021)

  7. [7]

    Eslami, S., de Melo, G., Meinel, C.: Does clip benefit visual question answering in the medical domain as much as it does in the general domain? (2021)

  8. [8]

    In: International MICCAI brainlesion workshop

    Hatamizadeh, A., Nath, V., Tang, Y., Yang, D., Roth, H.R., Xu, D.: Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images. In: International MICCAI brainlesion workshop. pp. 272–284. Springer (2021)

Show all 32 references
  1. [9]

    In: Proceedings of the IEEE/CVF winter conference on applications of computer vi- sion

    Hatamizadeh, A., Tang, Y., Nath, V., Yang, D., Myronenko, A., Landman, B., Roth, H.R., Xu, D.: Unetr: Transformers for 3d medical image segmentation. In: Proceedings of the IEEE/CVF winter conference on applications of computer vi- sion. pp. 574–584 (2022)

  2. [10]

    arXiv preprint arXiv:2003.10286 (2020)

    He, X., Zhang, Y., Mou, L., Xing, E., Xie, P.: Pathvqa: 30000+ questions for medical visual question answering. arXiv preprint arXiv:2003.10286 (2020)

  3. [11]

    Isensee,F.,Petersen,J.,Klein,A.,Zimmerer,D.,Jaeger,P.F.,Kohl,S.,Wasserthal, J.,Koehler,G.,Norajitra,T.,Wirkert,S.,Maier-Hein,K.H.:nnu-net:Self-adapting framework for u-net-based medical image segmentation (2018)

  4. [12]

    Jiang, A.Q., Sablayrolles, A., Mensch, A., Bamford, C., Chaplot, D.S., de las Casas, D., Bressand, F., Lengyel, G., Lample, G., Saulnier, L., Lavaud, L.R., Lachaux, M.A., Stock, P., Scao, T.L., Lavril, T., Wang, T., Lacroix, T., Sayed, W.E.: Mistral 7b (2023)

  5. [13]

    arXiv:2304.02643 (2023)

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.Y., Dollár, P., Girshick, R.: Segment anything. arXiv:2304.02643 (2023)

  6. [14]

    Scientific Data11(1), 1283 (2024)

    Kuş, Z., Aydin, M.: Medsegbench: A comprehensive benchmark for medical image segmentation in diverse data modalities. Scientific Data11(1), 1283 (2024)

  7. [15]

    arXiv:2308.00692 (2023)

    Lai, X., Tian, Z., Chen, Y., Li, Y., Yuan, Y., Liu, S., Jia, J.: Lisa: Reasoning segmentation via large language model. arXiv:2308.00692 (2023)

  8. [16]

    Scientific data (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 (2018)

  9. [17]

    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 (2023) MedSeg-R 11

  10. [18]

    In: In- ternational Symposium on Biomedical Imaging (ISBI)

    Liu, B., Zhan, L.M., Xu, L., Ma, L., Yang, Y., Wu, X.M.: Slake: A semantically- labeled knowledge-enhanced dataset for medical visual question answering. In: In- ternational Symposium on Biomedical Imaging (ISBI). IEEE (2021)

  11. [19]

    Nature Communications15, 654 (2024)

    Ma, J., He, Y., Li, F., Han, L., You, C., Wang, B.: Segment anything in medical images. Nature Communications15, 654 (2024)

  12. [20]

    Myronenko,A.:3dmribraintumorsegmentationusingautoencoderregularization. In: Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries: 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 16, 2018, Revise...

  13. [21]

    Peng, Z., Xu, Z., Zeng, Z., Xie, L., Tian, Q., Shen, W.: Parameter efficient fine- tuning via cross block orchestration for segment anything model (2024)

  14. [22]

    Peng, Z., Xu, Z., Zeng, Z., Yang, X., Shen, W.: Sam-parser: Fine-tuning sam effi- ciently by parameter space reconstruction (2023)

  15. [23]

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., Sutskever, I.: Learning transferable visual models from natural language supervision (2021)

  16. [24]

    Biomed- ical Signal Processing and Control p

    Rahimzadeh, M., Attar, A., Sakhaei, S.M.: A fully automated deep learning-based network for detecting covid-19 from a new and large lung ct scan dataset. Biomed- ical Signal Processing and Control p. 102588 (2021)

  17. [25]

    Rasheed, H., Maaz, M., Mullappilly, S.S., Shaker, A., Khan, S., Cholakkal, H., Anwer, R.M., Xing, E., Yang, M.H., Khan, F.S.: Glamm: Pixel grounding large multimodal model (2024)

  18. [26]

    Shi, L., Li, X., Hu, W., Chen, H., Chen, J., Fan, Z., Gao, M., Jing, Y., Lu, G., Ma, D., Ma, Z., Meng, Q., Tang, D., Sun, H., Grzegorzek, M., Qi, S., Teng, Y., Li, C.: Ebhi-seg: A novel enteroscope biopsy histopathological haematoxylin and eosin image dataset for image segment...

  19. [27]

    arXiv preprint arXiv:2303.05977 (2023)

    van Sonsbeek, T., Derakhshani, M.M., Najdenkoska, I., Snoek, C.G., Worring, M.: Open-ended medical visual question answering through prefix tuning of language models. arXiv preprint arXiv:2303.05977 (2023)

  20. [28]

    Nature Medicine30(11), 3129–3141 (Aug 2024)

    Zhang, K., Zhou, R., Adhikarla, E., Yan, Z., Liu, Y., Yu, J., Liu, Z., Chen, X., Davison, B.D., Ren, H., Huang, J., Chen, C., Zhou, Y., Fu, S., Liu, W., Liu, T., Li, X., Chen, Y., He, L., Zou, J., Li, Q., Liu, H., Sun, L.: A generalist vision–language foundation model for dive...

  21. [29]

    arXiv preprint arXiv:2304.13785 (2023)

    Zhang, K., Liu, D.: Customized segment anything model for medical image seg- mentation. arXiv preprint arXiv:2304.13785 (2023)

  22. [30]

    arXiv preprint arXiv:2303.00915 (2023)

    Zhang, S., Xu, Y., Usuyama, N., Xu, H., Bagga, J., Tinn, R., Preston, S., Rao, R., Wei, M., Valluri, N., et al.: Biomedclip: a multimodal biomedical foundation model pretrained from fifteen million scientific image-text pairs. arXiv preprint arXiv:2303.00915 (2023)

  23. [31]

    Nature Methods22(1), 166–176 (Nov 2024)

    Zhao, T., Gu, Y., Yang, J., Usuyama, N., Lee, H.H., Kiblawi, S., Naumann, T., Gao, J., Crabtree, A., Abel, J., Moung-Wen, C., Piening, B., Bifulco, C., Wei, M., Poon, H., Wang, S.: A foundation model for joint segmentation, detection and recognition of biomedical objects acros...

  24. [32]

    arXiv:2304.10592 (2023)

    Zhu, D., Chen, J., Shen, X., Li, X., Elhoseiny, M.: Minigpt-4: Enhancing vision- language understanding with advanced large language models. arXiv:2304.10592 (2023)

Pith tools

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