Retinal and medical imaging
Aain-D: AI for diabetic retinopathy from retinal images
AI that grades diabetic retinopathy from retinal photographs, including a new Radon-transform method released as open-source software.
Project leadDr. Zubair Shah
- 5
- Publications
- 3
- Talks
- 3
- Team members
- 0
- Partners
Summary
We built AI that looks at photographs of the back of the eye and grades how far diabetic eye disease has progressed, so screening can reach more people.
Diabetic retinopathy is a leading cause of preventable vision loss, and early detection depends on regular retinal screening. Aain-D, funded by the HBKU Innovation Centre, set out to build an AI-enabled mobile tool to help diagnose diabetes and stratify diabetic retinopathy from retinal images.
The research introduced RadEx, a nonlinear Radon-inspired transform that turns retinal images into sinograms, and RadFuse, which combines these sinograms with fundus images in deep networks. Tested on the public APTOS-2019 and DDR datasets, the method achieved high grading accuracy and was released as open-source software.
Key findings
- Quadratic kappa for five-class retinopathy grading with RadFuse, with 87% accuracy
- 0.932
- Open-source Python package releasing the nonlinear Radon transform
- RadEx
OverviewWhy it matters · Research question and aims · Impact
Why it matters
- Diabetic retinopathy is a leading cause of preventable blindness, and early treatment works.
- Screening depends on specialists reading many images. AI can help prioritise the people who need care first.
Research question and aims
The main research question- Grade diabetic retinopathy severity from retinal fundus images.
- Develop new image-feature methods based on the Radon transform.
- Evaluate AI models for generating synthetic retinal images.
- Release the methods as reusable open-source software.
Impact
Who benefits and how: patients, clinicians, health services, policy, training (e.g. "Model shared with HMC diabetes clinic")
MethodsDesign, data and methods · Timeline and milestones · Ethics and data protection
Design, data and methods
- Study design
- Study design (e.g. prospective cohort, retrospective analysis, prototype evaluation)
- Setting
- Where the study takes place
- Participants or sample
- Who is included and how many
Analytics and AI methods
How the study works
- Deep learning on retinal fundus images.
- Linear and nonlinear Radon-transform sinograms as an additional input.
- Comparative evaluation of generative models for retinal image synthesis.
Timeline and milestones
Milestones with dates (start, ethics approval, data collection, results, completion)
Ethics and data protection
- Approving body
- e.g. HBKU IRB, HMC MRC
- Approval number
- Approval or reference number
- Data protection
- How data are stored and protected
Team & partnersTeam · Partners and collaborators
Team
Partners and collaborators
Partner organisations and their role (data, clinical site, funding, community)
OutputsOutputs
Outputs
Publications 5
-
Integrating non-linear Radon transformation for diabetic retinopathy grading
Mohsen F, Belhaouari S, Shah Z
Scientific Reports 15: 30706, 2025
Journal articleRead the paper Free version
-
RadEx: an open source Python package for nonlinear Radon transformation
Mohsen F, Islam A, Shah Z, Belhaouari SB
Software Impacts: 100779, 2025
Journal articleRead the paper
-
A comparative evaluation of artificial intelligence-based models in retinal image synthesis
Mohsen F, Aldhubri A, Biswas MR, Shah Z
2nd International Conference on Foundation and Large Language Models (FLLM 2024): 51–57, 2024
Conference paperRead the paper Conference talk
-
Introducing RadEx: adaptive parameterized feature extraction from medical images
Islam A, Mohsen F, Shah Z, Belhaouari SB
Computer Graphics International Conference (CGI 2024): 278–294, Springer, 2024
Conference paperRead the paper Conference talk
-
Mohsen F, Shah U, Islam A, Shah Z, Belhaouari SB
IEEE 26th International Workshop on Multimedia Signal Processing (MMSP 2024): 1–5, 2024
Conference paperRead the paper Conference talk
Show these in the publications list
Talks
Software, datasets and tools
Keywords
Get involved
Use or extend our methods
RadEx is open-source. Contact us to collaborate on retinal AI or to test the methods on new data.
Related projects
Completed · 2022–2026Harnessing the power of artificial intelligence: predicting type 2 diabetes onset
Completed · 2021–2024C.P.-Support: peer support for people living with diabetes
Completed · 2020