See the future with eye imaging data.
The Envision Portal is your launchpad for easily preparing, sharing, and finding eye imaging data using user friendly interfaces and automation tools.
What is the Envision Portal?
Eye imaging data are essential to advancing research in eye health and beyond. Yet, the research community faces challenges in finding and accessing eye imaging datasets that are standardized, well-documented, and ready for reuse. To address this challenge, we are developing the Envision Portal, a cloud-based, open-source platform that provides researchers and AI developers with the tools they need to conveniently share, discover, and reuse eye imaging data.
Learn moreEasy to use, intuitive, and focused on supporting your work.
Whether you are sharing data, searching for datasets, or building AI models, the Envision Portal provides the tools you need to work efficiently and responsibly.
Easy Data Sharing
Contribute your datasets through guided submission workflows. The platform supports data standardization, data de-identification, metadata completeness, and multiple access methods making your data valuable and responsibly reusable by the community.
Convenient Data Discovery
Discover diverse eye imaging datasets using powerful search and filtering tools. The platform act as a registry for all eye imaging datasets, whether they are shared through the Envision Portal or not.
Automation for Efficiency
Automated tools for data formatting, metadata completeness, and de-identification help ensure that are AI-ready while reducing effort for contributors. The platform also includes a novel LLM-based search tool where data consumers can identify the right datasets for their use cases through a series of questions.
FAIR first
Every dataset on the Envision Portal is aligned with the FAIR (Findable, Accessible, Interoperable, Reusable) principles. Rich metadata, standardized formats, and clear documentation make data ready for reuse in research, AI development, and beyond.
Development Timeline
Our development timeline is designed to align with the needs of our users and ensure a smooth transition from prototype to production.
Year 1
Architecture, standards, and base platform
- Define the technical architecture and user workflow to align internally on major design decisions
- Indentify data standards to make datasets FAIR and AI-ready (CDS structure, DICOM for imaging, OMOP for clinical tables)
- Create initial wireframes and begin building the base code for the platform
- Set up initial developer approach for open-source development (public GitHub, contribution-friendly structure)
Year 2
Data access and initial launch
- Manually standardize one new dataset and publish it through the platform
- Build core dataset sharing services: storage, DOI minting, dataset landing pages, and data access workflow
- Index about 10 external eye imaging datasets from other repositories into the Envision Portal database
- Launch Envision Portal for users to access the new dataset and discover indexed external datasets
- Establish user and developer documentation and begin ongoing maintenance of docs
- Start community outreach through conferences and webinars
Year 3
Uploader workflows and automation foundations
- Build user-facing workflows to upload, manage, standardize, and share datasets through Envision Portal
- Start automated tooling for data standardization and preparation
- Implement PHI detection and removal tooling, plus validation support
- Add dataset versioning workflows
- Develop an automated pipeline to detect and index eye imaging datasets from other repositories
- Define federated learning approach with the Alzheimer's Disease Data Initiative (ADDI)
- If needed, build a desktop upload and download app to improve user experience
Year 4
Controlled access, advanced discovery, and reuse
- Continue improving data standardization tooling
- Implement advanced data access features, including controlled access request workflows
- Add advanced dataset discovery features, including robust search and filtering
- Deliver a robust API for integration with AI and ML pipelines
- Add in-portal preview and visualization for common formats (CSV, XLSX, DICOM, BMP, and image formats) before download
- Build initial federated learning capabilities
- Support datasets shared by external groups and expand the number of indexed external datasets
Year 5
Scale, security validation, and long-term sustainability
- Complete federated learning capabilities
- Conduct full security validation to prepare for large-scale independent submissions and sharing
- Establish contribution guidelines and support community-driven extensions of the platform
- Pursue outreach for wide adoption and formal recognition as a trusted domain repository
- Execute the sustainability plan to support development and operations beyond current funding
