CoChat
Sign Up
PreprintCited by 57Multimodal Machine Learning ApplicationsArtificial Intelligence in Healthcare and EducationDomain Adaptation and Few-Shot Learning

MedGemma Technical Report

Open access
Sellergren, Andrew, Kazemzadeh, Sahar, Jaroensri, Tiam, Kiraly, Atilla, Traverse, Madeleine, Kohlberger, Timo, Xu, Shawn, Jamil, Fayaz +73 more·2025·arXiv
DOI 10.48550/arxiv.2507.05201arXiv 2507.05201

Preprint — published as 10.1038/s41591-026-04626-w · Paper page

Open this paper in CoChat — free accountOpen PDF

Abstract

Artificial intelligence (AI) has significant potential in healthcare applications, but its training and deployment faces challenges due to healthcare's diverse data, complex tasks, and the need to preserve privacy. Foundation models that perform well on medical tasks and require less task-specific tuning data are critical to accelerate the development of healthcare AI applications. We introduce MedGemma, a collection of medical vision-language foundation models based on Gemma 3 4B and 27B. MedGemma demonstrates advanced medical understanding and reasoning on images and text, significantly exceeding the performance of similar-sized generative models and approaching the performance of task-specific models, while maintaining the general capabilities of the Gemma 3 base models. For out-of-distribution tasks, MedGemma achieves 2.6-10% improvement on medical multimodal question answering, 15.5-18.1% improvement on chest X-ray finding classification, and 10.8% improvement on agentic evaluations compared to the base models. Fine-tuning MedGemma further improves performance in subdomains, reducing errors in electronic health record information retrieval by 50% and reaching comparable performance to existing specialized state-of-the-art methods for pneumothorax classification and histopathology patch classification. We additionally introduce MedSigLIP, a medically-tuned vision encoder derived from SigLIP. MedSigLIP powers the visual understanding capabilities of MedGemma and as an encoder achieves comparable or better performance than specialized medical image encoders. Taken together, the MedGemma collection provides a strong foundation of medical image and text capabilities, with potential to significantly accelerate medical research and development of downstream applications. The MedGemma collection, including tutorials and model weights, can be found at https://goo.gle/medgemma.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

DataCite · 10.48550/arxiv.2507.05201

Cites

1 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Existing medical MLLMs focus on visual question answering and are largely confined to image-level understanding tasks, failing to achieve fine-grained pixel-level comprehension.

    “Prevailing medical MLLMsfocus on visual question answering (VQA) and are largely confined to image-level understanding tasks, thus failing to achieve fine-grained pixel-level comprehension.”

    From IBISAgent: Reinforcing Pixel-Level Visual Reasoning in MLLMs for Universal Biomedical Object Referring and Segmentation (2026) · cites this paper as [33]

    The Introduction (p. not available) says recent multimodal models have demonstrated utility in standard 2D imaging, while models addressing more complex imaging tasks are limited. Section 3.1 (p. not available) includes VQA among existing MedGemma benchmarks, but also describes report generation and image-based multiple-choice tasks. The report does not establish that prevailing medical MLLMs are largely confined to VQA or image-level tasks, or that they fail at pixel-level comprehension. (LLM, conf=0.82)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

The publisher hasn’t deposited this paper’s reference list.

Research works printed but not linked: 4

  1. Yixiong Chen, Shawn Xu, Andrew Sellergren, Yossi Matias, Avinatan Hassidim, Shravya Shetty, Daniel Golden, Alan L Yuille, and Lin Yang. Coca-cxr: Co ntrastive ca ptioners learn strong temporal structures for chest x-ray vision-language understanding. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 78-88. Springer, 2025.
  2. Ary L Goldberger, Luis AN Amaral, Leon Glass, Jeffrey M Hausdorff, Plamen Ch Ivanov, Roger G Mark, Joseph E Mietus, George B Moody, Chung-Kang Peng, and H Eugene Stanley. Physiobank, physiotoolkit, and physionet: components of a new research resource for complex physiologic signals. circulation, 101(23):e215-e220, 2000b.
  3. Ankit Pal, Logesh Kumar Umapathi, and Malaikannan Sankarasubbu. Medmcqa: A large-scale multi- subject multi-choice dataset for medical domain question answering. In Conference on health, inference, and learning, pages 248-260. PMLR, 2022.
  4. Andrew Sellergren, Sahar Kazemzadeh, Tiam Jaroensri, Atilla Kiraly, Madeleine Traverse, Timo Kohlberger, Shawn Xu, Fayaz Jamil, Cían Hughes, Charles Lau, et al. Medgemma technical report. arXiv preprint arXiv:2507.05201, 2025.

Cited by

Sorted by citations

Related

Authors
Sellergren, A., Kazemzadeh, S., Jaroensri, T., Kiraly, A., Traverse, M., Kohlberger, T., Xu, S., Jamil, F., Hughes, C., Lau, C., Chen, J., Mahvar, F., Yatziv, L., Chen, T., Sterling, B., Baby, S. A., Baby, S. M., Lai, J., Schmidgall, S., Yang, L., Chen, K., Bjornsson, P., Reddy, S., Brush, R., Philbrick, K., Asiedu, M., Mezerreg, I., Hu, H., Yang, H., Tiwari, R., Jansen, S., Singh, P., Liu, Y., Azizi, S., Kamath, A., Ferret, J., Pathak, S., Vieillard, N., Merhej, R., Perrin, S., Matejovicova, T., Ramé, A., Riviere, M., Rouillard, L., Mesnard, T., Cideron, G., Grill, J.-B., Ramos, S., Yvinec, E., Casbon, M., Buchatskaya, E., Alayrac, J.-B., Lepikhin, D., Feinberg, V., Borgeaud, S., Andreev, A., Hardin, C., Dadashi, R., Hussenot, L., Joulin, A., Bachem, O., Matias, Y., Chou, K., Hassidim, A., Goel, K., Farabet, C., Barral, J., Warkentin, T., Shlens, J., Fleet, D., Cotruta, V., Sanseviero, O., Martins, G., Kirk, P., Rao, A., Shetty, S., Steiner, D. F., Kirmizibayrak, C., Pilgrim, R., Golden, D., & Yang, L.
ORCID
Venue
arXiv
Publisher
arXiv
Published
2025
ISSN
2331-8422 (ISSN-L)
DOI
10.48550/arxiv.2507.05201
arXiv
2507.05201
OpenAlex
W4414687708
Cited by
49
References
0
Citations by year
2026: 43 · 2025: 6
Type
posted content
Language
en
Open access
Licence
Locations
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • arXiv (Cornell University) (cc-by)
Funders
DeepMind
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
Keywords
MedGemma, Gemma 3, vision encoder, medical image interpretation, out-of-distribution generalization

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review