MICCAI 2026 Open Data Micro-Grants

MICCAI 2026 Open Data Micro-Grants is sponsored by MICCAI Society. These grants support the creation of original, publicly available open datasets in medical imaging and clinical AI, with a focus on health equity, diversity, and under-represented populations.

Unlock Medical Machine Learning with Open Data We are pleased to announce the third edition of Open Data at MICCAI 2026. As medical machine learning continues to advance, access to diverse, representative, and inclusive datasets remains a key bottleneck for addressing global healthcare challenges.

The Open Data initiative aims to foster collaboration and innovation by encouraging the sharing of high-quality medical imaging datasets within the MICCAI community. Building on the success of the first two editions, Open Data at MICCAI 2026 continues to place a strong emphasis on underrepresented populations and diseases.

Despite substantial progress in medical imaging research and the growing availability of public datasets, significant gaps persist, particularly for regions such as the Middle East. By showcasing datasets that reflect underrepresented populations and clinical conditions, this initiative seeks to reduce disparities in healthcare research and promote more equitable and inclusive machine learning development.

As in previous years, we are creating the space and community that supports and provides more publicly available, high-quality medical imaging datasets, with a particular focus on underrepresented populations and diseases. Details regarding the submission process and storage guidelines will be shared at a later stage.

The initiative remains closely aligned with the principles of FAIR data and Data-centric AI, emphasizing the critical role of well-curated data in building reliable and generalizable models. After all, there are no good models without good data. New for 2026, thanks to the MICCAI Health Equity Grant, we are pleased to introduce micro-grants for dataset providers.

These micro-grants aim to support researchers and institutions in preparing, curating, and sharing high-impact datasets, particularly those representing underrepresented populations and regions. Further details regarding eligibility, application procedures, and grant amounts are provided below. MICCAI 2026 will take place from September 27th - October 1st, 2026, in Strasbourg.

The Open Data event will run in parallel with the main conference. If you are attending MICCAI, you are welcome to visit the Open Data sessions, and no separate registration is required. Datasets and related works will be selected through a peer-reviewed paper submission process.

The MELBA journal will serve as the official journal for submissions from the MICCAI Open Data track. We look forward to welcoming you to the third Open Data session at MICCAI 2026 and to continuing our collective efforts toward more inclusive and equitable medical AI research.

To maximize the impact of shared medical imaging datasets, we have established a set of clear guidelines for data uploading to ensure accessibility and usability across the research community. By following the instructions below, you can contribute to a well-organized, reusable, and collaborative open-data resource. Important All data—including imaging and clinical variables—must be anonymized.

Any information that could link the data to a patient's identity or medical records must be removed. 1. Hosting and Accessibility Data must be hosted in a permanent repository.

While we do not impose a single option, we ask the providers to host their data in a permanent repository (E.g. Synapse, TCIA, Zenodo, Harvard Dataverse etc…). For a more extensive list see MELBA instructions . Data cannot be hosted on services such as GitHub, Google Drive, OneDrive etc. The dataset must be publicly accessible.

This can be through direct download or via a request-access form. Restricted access datasets must provide a clear process for requesting access. If you do not have a dedicated storage space , we recommend using Synapse .

Free uploads are limited to datasets < 100 GB. For larger datasets or other hosting needs, please contact the organizing team early to explore options. When using Synapse, please follow the data uploading instructions .

2. Data Organization and File Structure Organize your data logically and consistently: Use a hierarchical folder structure (e.g., /SubjectID/SessionID/Modality/) Maintain fixed and descriptive filenames (e.g., image. nii.

gz , brain_mask. nii. gz , metadata.

json ). Use one folder per subject. For each subject use one folder per session.

For each session use one folder per scan/modality. Use modality-specific data standards wherever possible: DICOM is strongly recommended where possible. Other formats such as NIfTI, PNG … etc are acceptable, but original DICOMs/DICOM headers or metadata files including acquisition parameters should be included where possible.

3. Patient and Demographic Information To support fairness and bias analysis, please include anonymized patient demographics where available (eg. Age, Sex, Race/Ethnicity).

Clearly indicate in your documentation which fields are provided and their format. 4. Task Definition and Labels Clearly define the intended task(s) for your dataset (e.g., segmentation, classification, detection).

Indicate in your documentation: which files contain labels, the label format (e.g., masks, CSV annotations), any relevant class definitions or codes used. 5. Documentation Requirements Each dataset must include a README.

md file at the root level describing the dataset documentation: The dataset structure and naming conventions. Included modalities and tasks. Patient information fields.

Licensing terms and data use conditions. A data dictionary or schema for tabular metadata. Scripts for downloading, preprocessing, or parsing the data.

6. Naming and Consistency Use consistent naming conventions across the dataset. Avoid ambiguous file names like 12345x6.

nii , final. png . Prefer informative structure and naming like sub-001/ses-MR-01/T1/image.

dcm or lung_mask. nii. gz .

Ideally, at the end of the upload your dataset would have a structure similar to the picture below: New medical imaging datasets that encompass diverse demographics, ethnicities, and medical conditions. This year is especially focused on the Middle East, but we also welcome datasets from other (underrepresented) populations or diseases. Updated or re-designed datasets based on previously publicly available data.

Additional topics include: Dataset collection and annotation techniques. Data augmentation strategies for improving dataset diversity. Ethical considerations in data sharing and privacy preservation.

Applications of open data in medical image analysis, diagnosis, and treatment planning. Challenges and opportunities in accessing and utilizing underrepresented datasets.

Paper submission deadline Notification of acceptance Camera-ready paper submission August 2, 2026, 23:59 CET We welcome submissions of papers presenting novel datasets, particularly those from the Middle East and other underrepresented populations and diseases - including methodologies for the data collection and curation, and innovative approaches for utilizing them in medical imaging research.

Encourage and empower through an open repository the sharing and dissemination of open-access datasets to facilitate collaboration and reproducibility. Promote awareness and understanding of the importance of inclusivity and representative data in developing robust and equitable healthcare solutions.

Facilitate networking opportunities among researchers, data custodians, and stakeholders interested in leveraging open data for medical machine learning. Guidelines on data submission, including upload procedures and storage details, will be provided at a later stage. Adhere to the FAIR data guidelines.

Any associated code should be open source. We invite authors to carefully review our updated author guidelines document: PLEASE CLICK HERE TO VIEW OPEN DATA MICCAI 2026 AUTHOR GUIDELINES New this year : to improve our review process, we now require the authors to provide the link to their repository upon submission.

If the dataset is going to be shared with restricted access, the authors are invited to provide a private link or create a reviewer account for the Open Data 2026 reviewers to be able to access and review the repository. For more information on licensing and ethics, please read the information in the guide below: DATA, LICENSING & ETHICS 101 Authors are invited to share their papers as pre-prints, including during the submission phase.

This is a single-blind review process. For general guidelines on "What Makes a Good Review" please refer to the corresponding section in the MICCAI reviewer guidelines . For this track, pay special attention on the Submission guidelines listed above - scope and criteria, summarized below (in order of priority): Data availability and adherence to the FAIR data principles.

Licensing, potential use cases, and ethical considerations/approvals. Methods used for the data and meta-data creation/collection: from the methods and equipment used for the acquisition to final processing, and the AI-ready state of the dataset (cleaning, curation, possible suggested splits, etc.). Clarity of the dataset specifics.

Dataset usage showcase(s) with evaluation results. We are pleased to announce Open Data Micro-Grants as part of the Health Equity Award. These grants support the creation of original, publicly available open datasets in medical imaging and clinical AI, with a focus on health equity, diversity, and under-represented populations .

Data preparation and curation Annotations and quality control Documentation and public data release Travel costs are not eligible.

4 x Small grants (USD 650) 4 x Large grants (USD 1,200) Researchers from academia, hospitals, or industry No geographic or career-stage restrictions Public release with an open license Original datasets (extensions of public datasets allowed) Ethical approval already obtained Funding released after paper submission (acceptance not required) Proposal submission: February 22nd, 2026, 23. 59 CET Decision: March 01, 2026, 23.

59 CET PLEASE CLICK HERE TO VIEW OPEN DATA MICRO-GRANTS 2026 RESULTS The Microsoft CMT service was used for managing the peer-reviewing process for this conference. This service was provided for free by Microsoft and they bore all expenses, including costs for Azure cloud services as well as for software development and support. Martijn P.

A. Starmans, PhD Assistant Professor AI for Integrated Diagnostics (AIID) Dept. of Radiology & Nuclear Medicine, Dept.

of Pathology Erasmus University Medical Center, Rotterdam, the Netherlands Apostolia Tsirikoglou, PhD Research Specialist, AI for Breast Imaging Dept. of Oncology-Pathology, Karolinska Institutet, Sweden Lidia Garrucho Moras, PhD Postdoc in AI for Medical Imaging Artificial Intelligence in Medicine Lab University of Barcelona, Barcelona, Spain PhD Candidate, Medical Image Analysis and Federated Learning Dept.

of Radiology & Nuclear Medicine Erasmus University Medical Center, Rotterdam, the Netherlands Laura Arbelaez Ossa, MD, PhD AI Ethicist, Implementation and regulatory affairs for Digital Health Artificial Intelligence in Medicine Lab University of Barcelona, Barcelona, Spain