OSCR

Learning like a radiologist: a medical vision-language model for radiological image analysis via curriculum learning.

Code ↔ Paper

The paper beside its authors' code: matches between them have not been computed for this paper yet.

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Python · 70 lines · 2.3 KB · no license

  1. import os, sys
  2. import datetime
  3. import dotenv
  4. from pathlib import Path
  5. from icecream import ic
  7. dotenv.load_dotenv(override=True)
  10. def main():
  11. src_dir = os.path.join(os.path.dirname(__file__), 'src')
  12. assert os.path.isdir(src_dir), f"error src/ - {src_dir}"
  13. ic(f"Working Dir: {src_dir}")
  14. os.chdir(src_dir)
  16. # region >>>torchrun
  17. command = (
  18. f"torchrun "
  19. f"--nproc_per_node=4 "
  20. f"--rdzv_endpoint=$HOSTE_NODE_ADDR "
  21. f"--master_port=29507 "
  22. f"-m training.main "
  23. f"--model hf-hub:microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224 "
  24. f"--train-data=TRUE "
  25. f"--data=your_data_csv "
  26. f"--warmup 2000 --lr 1e-4 --beta1 0.9 --beta2 0.98 --wd 0.2 --eps 1e-6 "
  27. f"--precision amp --workers 32 --grad-clip-norm 1.0 "
  28. f"--local-loss "
  29. f"--gather-with-grad "
  30. f"--batch-size 128 "
  31. f"--accum-freq 16 "
  32. f"--epochs 10 "
  33. f"--name {TRAIN_COMMAND__NAME} "
  34. f"--logs {TRAIN_COMMAND__LOGS} "
  35. f"--seed 0 "
  36. f"--report-to tensorboard "
  37. # f"2 > {TRAIN_COMMAND_ERR_TO}"
  38. "--siglip "
  39. )
  40. # endregion <<<
  42. # region >>> Duplicate train call
  43. # Save the command file(or sh/py) to train log folder
  44. Path(os.path.join(TRAIN_COMMAND__LOGS, TRAIN_COMMAND__NAME)).mkdir(exist_ok=True, parents=True)
  45. save_to = os.path.join(TRAIN_COMMAND__LOGS, TRAIN_COMMAND__NAME, "commmand_call_history.txt")
  46. with open(save_to, 'a') as f:
  47. f.write(f"{'=' * 40} {datetime.datetime.now()} {'=' * 40}\n")
  48. f.write(f"Train Session Desc: {Train_Session_Desc}\n\n")
  49. f.write(f"Command: \n")
  50. f.write(command.replace(" --", "\n--"))
  51. f.write("\n\n")
  53. # endregion <<< Duplicate train call
  55. os.system(command)
  56. # subprocess.run(command, shell=True, check=True)
  58. if __name__ == "__main__":
  59. SCI_PROJECT_NFS_ROOT = dotenv.dotenv_values()["SCI_PROJECT_NFS_ROOT"]
  60. ic(SCI_PROJECT_NFS_ROOT)
  62. N_Nodes = 1
  63. NPROC_PER_NODE = 4
  64. TRAIN_COMMAND__LOGS = "./train_log/"
  65. TRAIN_COMMAND__NAME = f"{N_Nodes}_{NPROC_PER_NODE}"
  66. TRAIN_COMMAND_ERR_TO = os.path.join(TRAIN_COMMAND__LOGS, TRAIN_COMMAND__NAME, "sci_err.log")
  68. Train_Session_Desc = ""
  70. main()

run.py at commit 045ad9e, no license · at the source

Overview

Authors: Minhui Tan1,2, Qingxia Wu3, Boyang Zhang2, Genqiang Ren2, Jianlong Nie1,2, Zhong Xue2, Yiqiang Zhan2, Sean Zhou2, Xiaohuan Cao2, Dinggang Shen1,2,4
  1. School of Biomedical Engineering & State Key Laboratory of Advanced Medical Materials and Devices, ShanghaiTech University, Shanghai, China
  2. Shanghai United Imaging Intelligence Co., Ltd., Shanghai, China
  3. Beijing United Imaging Research Institute of Intelligent Imaging, Beijing, China
  4. Shanghai Clinical Research and Trial Center, Shanghai, China
Institutions: ShanghaiTech University (China); United Imaging Intelligence (China) (China); Shanghai Clinical Research Center (China)
Journal: NPJ digital medicine, volume 9, issue 1, article 695
Dates: received 15 July 2025; accepted 25 April 2026; published online 5 June 2026
Type: Research article · Language: English
License: CC BY-NC-ND
Identifiers: DOI 10.1038/s41746-026-02713-3 · PMID 42249121 · PMCID PMC13562648 · OpenAlex W7163703564
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: clinical / translational (subfield)
Methods: Machine learning
Keywords: Computational biology and bioinformatics, Health care, Mathematics and computing, Medical research
Topic: Radiology practices and education (Radiology, Nuclear Medicine and Imaging, Medicine), according to OpenAlex
Funding: National Key Research and Development Program of China (2022YFE0205700); Shanghai Municipal Central Guided Local Science and Technology Development Fund (YDZX20233100001001); National Natural Science Foundation of China (82441023); Beijing Natural Science Foundation (IS24053)
Citations: not cited yet (Europe PMC); 55 references in the paper

Abstract

The abstract is not reproduced here: the paper's license (CC BY-NC-ND) does not allow it. Read it in the paper, at the publisher or on Europe PMC.

Repository

Its files are read in the Code ↔ Paper reader above.

Erickeyan/RadiSim

License: none: the authors keep all their rights
State: the link answers, verified on 27 September 2026
Evidence: files inventoried
Commit: 045ad9e24a7cc426291ae112df0d4107e99afdef, 14 February 2026
Languages: Python (43)
Size: 54 files, 43 scripts
Software Heritage: not archived
Found in: “Data availability”
Holds: README, tests
Not found: license file, CITATION.cff, environment file, continuous integration, documentation
Tools: PyTorch (30 files), NumPy (9 files), Pillow (6 files), Hugging Face Transformers (4 files), pandas (2 files)
Availability: 1 check, the latest on 27 September 2026: the link answers
  • 27 September 2026: the link answers
44 files
At the source: github.com/Erickeyan/RadiSim

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 43 scripts, each with its path and the digest of its content;
  • no match between paragraphs and code yet;
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

Datasets cited

Code and data availability statement

The paper has a code and data availability statement. Its license (CC BY-NC-ND) does not allow reproducing it here; in short, from what the harvester recognized in it:

Read it in the paper: doi.org/10.1038/s41746-026-02713-3.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 10 authors, 4 keywords, 4 funders, 24 references.

Cite

This paper

Tan, M., Wu, Q., Zhang, B., Ren, G., Nie, J., Xue, Z., Zhan, Y., Zhou, S., Cao, X., & Shen, D. (2026). Learning like a radiologist: a medical vision-language model for radiological image analysis via curriculum learning. NPJ digital medicine, 9(1), 695. https://doi.org/10.1038/s41746-026-02713-3

BibTeX

@article{tan2026learning,
author = {Tan, Minhui and Wu, Qingxia and Zhang, Boyang and Ren, Genqiang and Nie, Jianlong and Xue, Zhong and Zhan, Yiqiang and Zhou, Sean and Cao, Xiaohuan and Shen, Dinggang},
title = {{Learning like a radiologist: a medical vision-language model for radiological image analysis via curriculum learning}},
journal = {NPJ digital medicine},
year = {2026},
month = jun,
volume = {9},
number = {1},
pages = {695},
publisher = {Nature Publishing Group},
issn = {2398-6352},
doi = {10.1038/s41746-026-02713-3},
url = {https://doi.org/10.1038/s41746-026-02713-3},
pmid = {42249121},
pmcid = {PMC13562648}
}

RIS

TY - JOUR
AU - Tan, Minhui
AU - Wu, Qingxia
AU - Zhang, Boyang
AU - Ren, Genqiang
AU - Nie, Jianlong
AU - Xue, Zhong
AU - Zhan, Yiqiang
AU - Zhou, Sean
AU - Cao, Xiaohuan
AU - Shen, Dinggang
TI - Learning like a radiologist: a medical vision-language model for radiological image analysis via curriculum learning
T2 - NPJ digital medicine
J2 - NPJ Digit Med
PY - 2026
DA - 2026/06/05
VL - 9
IS - 1
SP - 695
SN - 2398-6352
PB - Nature Publishing Group
DO - 10.1038/s41746-026-02713-3
UR - https://doi.org/10.1038/s41746-026-02713-3
LA - en
ER -

CSL-JSON

{
"id": "10.1038/s41746-026-02713-3",
"type": "article-journal",
"title": "Learning like a radiologist: a medical vision-language model for radiological image analysis via curriculum learning",
"container-title": "NPJ digital medicine",
"author": [
{
"family": "Tan",
"given": "Minhui"
},
{
"family": "Wu",
"given": "Qingxia"
},
{
"family": "Zhang",
"given": "Boyang"
},
{
"family": "Ren",
"given": "Genqiang"
},
{
"family": "Nie",
"given": "Jianlong"
},
{
"family": "Xue",
"given": "Zhong"
},
{
"family": "Zhan",
"given": "Yiqiang"
},
{
"family": "Zhou",
"given": "Sean"
},
{
"family": "Cao",
"given": "Xiaohuan"
},
{
"family": "Shen",
"given": "Dinggang"
}
],
"container-title-short": "NPJ Digit Med",
"volume": "9",
"issue": "1",
"page": "695",
"DOI": "10.1038/s41746-026-02713-3",
"PMID": "42249121",
"PMCID": "PMC13562648",
"ISSN": "2398-6352",
"publisher": "Nature Publishing Group",
"URL": "https://doi.org/10.1038/s41746-026-02713-3",
"language": "en",
"issued": {
"date-parts": [
[
2026,
6,
5
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1016/j.patter.2026.101538 [code]
A multi-modal foundation model for brain disease diagnosis and medical imaging.
Journal: Patterns (New York, N.Y.)
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, clinical / translational, 1 reference
[2] doi:10.1038/s43856-026-01817-x [code]
Visual prompt engineering for multimodal and irregularly sampled medical data.
Journal: Communications medicine
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, clinical / translational, 1 reference
[3] doi:10.1038/s41598-026-50158-1 [code]
NeXtSwin-X: dual-branch cross-attention fusion of ConvNeXt and swin transformer for accurate brain tumor classification from MRI and CT.
Journal: Scientific reports
In common: PyTorch, pandas, NumPy, kaggle.com/datasets/ahmedhamada0
[4] doi:10.1093/braincomms/fcag253 [code]
Disease detection and classification in temporal lobe epilepsy: step-wise versus simultaneous AI decision models in a multisite neuroimaging study.
Journal: Brain communications
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, clinical / translational
[5] doi:10.1038/s41598-026-50791-w [code]
Hybrid Vi+ECNN framework for advanced ADHD diagnostic accuracy in medical imaging.
Journal: Scientific reports
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, clinical / translational
[6] doi:10.1038/s41531-026-01348-1 [code]
Pathology-Anchored Transcranial Sonography: A Cascaded Super-Resolution Deep Learning System for Early-Stage Parkinson's Disease Grading.
Journal: NPJ Parkinson's disease
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, clinical / translational
[7] doi:10.1038/s42003-026-10957-8 [code]
Brain defence by the extracellular matrix protein Cochlin.
Journal: Communications biology
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools
[8] doi:10.1038/s41467-026-76837-1 [code]
Drug screen and machine learning predict neuroprotective agents in a preclinical human model of childhood dementia.
Journal: Nature communications
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools
[9] doi:10.1038/s41467-026-76098-y [code]
A single computational objective can produce specialization of streams in visual cortex.
Journal: Nature communications
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools
[10] doi:10.1162/imag.a.1326 [code]
RAVEN: Robust, generalizable, multi-resolution structural MRI upsampling using autoencoders.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.