Rahele - The Future of Eye Care with AI

Transforming Eye Care with Artificial Intelligence

By: Dr Raheleh Kafieh

Transforming eye care with AI: Enhancing vision, empowering clinicians, and revolutionizing early diagnosis for a brighter, healthier future.

Where the journey begins

index

Exploring th Brain through the Eye

Introduction

The AI tool for Diagnosis

Eye Imaging Technologies

Meet the Team

Research Overview

AI in Action

Contact

Publications

Vision loss affects millions worldwide, yet many sight-threatening diseases can be detected early with the right tools. Traditional screening methods are time-consuming and dependent on specialist availability, leading to delays in diagnosis and treatment. Artificial Intelligence (AI) is changing this landscape—enhancing accuracy, automating referrals, and making eye care more accessible. Our research harnesses AI-driven ocular imaging to revolutionise early diagnosis, optimise patient pathways, and improve healthcare outcomes.

INTRODUCTION

What is AI?

💡 Click to explore the challenges in current eye care before AI was introduced.

Have you ever wondered how your eyes transform light into the vibrant world you see around you?

🤔✨ Dive into the fascinating mechanics of vision and uncover the secrets of the human eye!

📌 Key Message: Eye diseases like glaucoma, diabetic retinopathy, and age-related macular degeneration affect millions worldwide. AI can revolutionize early detection, automate referrals, and improve accessibility.

Our research is bridging the gap between technology and clinical practice.

Optical Coherence Tomography (OCT):

Captures high-resolution cross-sectional images of the retina.provide context

Exploring Eye Imaging Technologies

Modern eye imaging technologies help us see beyond the surface, allowing for early disease detection and precise diagnostics.

Fluorescein Angiography (FA):

Fundus Photography:

Provides a detailed image of the back of the eye, including the retina and optic nerve.

Uses dye to visualize blood flow in the retina and detect vascular issues.

Visual Field Testing

Maps a person’s peripheral vision to detect blind spots or nerve damage.

Research Overview

Our research focuses on developing AI-driven tools for accessible, high-quality eye care. Three interconnected projects—VISTA, EYESAVE, and OCTAGE—align with this vision, leveraging AI for home-based screening, early disease detection, and age-related ocular health assessment.

OCTAGE: Monitoring the ageing brain via eyes.

EYESAVE: AI-enabled Glaucoma Triage

VISTA: Vision Intelligence for Smart Telehealth Assessment

VISTA: Vision Intelligence for Smart Telehealth Assessment

We collaborate with Siloton, a Bristol-based startup developing a handheld OCT device for monitoring AMD patients at home. Our AI contributions include:

• image quality control,
• enhancing resolution to hospital standards, and
• real-time analysis to detect disease progression, supporting patient autonomy and reducing hospital visits.

https://www.siloton.com/

EYESAVE: AI-enabled Glaucoma Triage

Glaucoma is a leading cause of permanent blindness, affecting 4% of people over 50 in the UK. It often goes unnoticed until it severely impacts vision, and diagnosing it requires several tests, which can be slow and prone to errors. Other health conditions, like heart disease, also play a role in managing glaucoma.Optical Coherence Tomography (OCT) helps doctors detect eye damage, but reading these images needs specialized training, which is scarce. Our project aims to create an AI tool that combines eye scans and health data to assess glaucoma risk, helping doctors make faster, more accurate treatment decisions.

OCTAGE: Monitoring the ageing brain via eyes

Age-related brain changes impact our lives, and the retina is the only part of the brain we can see directly. OCT and OCTA imaging can capture detailed retina images and are used for conditions like diabetes and glaucoma, but they also have potential for studying ageing and diseases like Alzheimer’s.This project aims to develop AI methods to analyze retinal changes with OCT data, creating a detailed image database from NHS and national resources. The goal is to design AI tools that can detect healthy ageing versus conditions like Alzheimer’s and track changes in the retina to identify early signs of age-related diseases.

What are we solving?

A Unified Vision

VISTA

Together, these projects create a comprehensive AI-driven ecosystem, bridging telehealth, community optometry, and hospital diagnostics to transform eye care.

EYESAVE

AI in Action: How It Works

🔬 Step-by-Step Process

Data Collection: OCT scans, fundus images, visual field data. AI Training & Validation: Deep learning models trained on diverse datasets. Clinical Integration: AI-assisted decision-making for early diagnosis and referrals. Patient Impact: Faster, more accurate diagnoses with reduced false referrals.

📌 Click here: A flowchart showing AI’s journey from data input to clinical decision-making.

Exploring the Brain Through the Eye

The connection between the eye and the brain provides valuable insights into diagnosing neurological diseases, as imaging the eye can reveal early signs of conditions such as multiple sclerosis, Alzheimer's, and Parkinson's through changes in the optic nerve and retinal structures.

Multiple sclerosis (MS) is a condition that affects the brain and spinal cord, causing problems with movement, balance, and vision. It happens when the immune system mistakenly attacks the protective covering of nerves, leading to communication issues between the brain and the rest of the body. Symptoms can vary widely, from fatigue and muscle weakness to difficulty walking and numbness. MS is a lifelong condition, but treatments can help manage symptoms and slow its progression.

Sample study: Multiple sclerosis

An interesting fact about multiple sclerosis (MS) is that the eye can provide early clues about the disease. Optical coherence tomography (OCT), a non-invasive eye scan, can detect thinning of the retina, which may indicate nerve damage even before other MS symptoms appear. This makes the eye a valuable window into brain health and a potential tool for early diagnosis.

INTRDUCTION HERE

MS affects

facts

over 2.3 M people worldwide

The average age of diagnosis is between 20 and 40 years old, though it can occur at any age. Around 85% of people with MS are initially diagnosed with relapsing-remitting MS (RRMS), which means symptoms come and go. The risk of developing MS in the general population is approximately 1 in 330, but it increases to 1 in 40 if a first-degree relative has it.

+2.5-5%

First degree relatives

Ratio of women with MS to men is three or four to one.

+15%

have family or relatives with MS

Total

+25%

Identical twins

From Brain to the Eye in MS

Retina is outgrowth of the brain and the only part of the Central Nervous System that can be seen directly from the outside

Optical Coherence Tomography

From Brain to the Eye in MS

Drag the correct person (healthy or MS) to match the corresponding OCT image.

Challenging, isn't it?

From Brain to the Eye in MS

Easy job?

Not really!

Any solution?

From Brain to the Eye in MS

Good for Segmentation

Work with large volume of data

AI

Can do quality control

Good for Classification

The AI tool for diagnosis from OCT images

Predicted benefits of the AI diagnosis tool

Benefits

For patients

For healthcare system

meet the team

Innovators Behind the Vision

Dr Raheleh Kafieh

Prof Anya Hurlbert

Prof Jaume Bacardit

Visual Neuroscience Newcastle University

AI and Machine learning in health Durham University

AI and Machine learning Newcastle University

Dr Dexter Canoy

Prof Jenny Read

Prof David Steel

Epidemiology and Public Health Newcastle University

Computational vision Newcastle University

Retinal surgery consultant

meet the team

ALUMNI

Innovators Behind the Vision

ALUMNI

Dr Fedra Hajizadeh

Dr Roya Arian

Mansooreh Montazerin

Dr Shima Khodabandeh

Prof Fereshteh Ashtari

PDRA Durham University

Consultant ophthalmologist

Consultant neurologist

PGR The University of Southern California

PhD in Bioengineering Isfahan Univ of Med Sciences

Sajed Rakhshani

Dr Zahra Amini

Dr Amir Riazi

Christian Taylor

Regan Cain Bolton

Senior PDRA Aston University

PDRA Newcastle University

PGR Isfahan Univ of Med Sciences

PhD in Bioengineering Isfahan Uni of Med Sciences

MEng Durham University

Knectt Lendoye L'Eyebe

Asieh Solatinpour

Charlotte Corry

Benjamin Bajpai

Ali Aghababaei

PGR Newcastle University

AI researcher

AI expert and Medical student Isfahan Univ of Med Sciences

BEng Durham University

BEng Durham University

Nazanin Faghih Mirzaei

Charlie Farrar

Angus Milton

Reyhaneh Esmailizadeh

Nicholas Njopa-Kaba

PGR West University of London

MEng in Bioengnieering Durham University

MEng Durham University

AI researcher

Management Consultant

Aaron Gifford

Sivashen Naidoo

Morgan Lant

Shwasa Iyer

BEng Durham University

BEng Durham University

MEng in BioengnieeringDurham University

Graduate Research Assistant University of Exeter

ALUMNI

Aref Habibi

Amirali Arbab

BEng Isfahan University of Technology

BEng Isfahan University of Technology

contact

Durham University

Dr Rahele Kafieh

+44 (0) 1913344984

Raheleh.kafieh@durham.ac.uk

Department of Engineering | Room E109 | Christopherson Building Durham University | Lower Mountjoy | South Road | Durham DH1 3LE

selected publications on this topic

Articles in indexed journals

VISTA: Vision Intelligence for Smart Telehealth Assessment

We collaborate with Siloton, a Bristol-based startup developing a handheld OCT device for monitoring AMD patients at home. Our AI contributions include:

• image quality control,
• enhancing resolution to hospital standards, and
• real-time analysis to detect disease progression, supporting patient autonomy and reducing hospital visits.

https://www.siloton.com/

The data is fed into the algorithm during the training stage.

Eventually, we have a trained model for classification.

AI tool that can classify unknown OCT images as either LIKELY MS's or healthy

For classification, agian the clinicians make the known data.

AI Training

Clinical Integration

Patient Impact: Faster, more accurate diagnoses with reduced false referrals.

Preprocessing & Augmentation: Data cleaning, normalization, synthetic data generation

AI-Assisted Decision-Making: Automated triage, risk assessment, referral suggestions

Model Development: Deep learning models for image analysis, machine learning for tabular data

Physician Adoption & Feedback: Human-AI collaboration, refining models based on clinical input

Tabular Data: Visual field data, patient demographics, clinical history

External Validation: Testing on independent datasets, real-world clinical evaluation

Image Data: OCT scans, fundus images

Internal Validation: Performance assessment on training/validation datasets

AI validation

Data Collection

A flowchart showing AI’s journey from data input to clinical decision-making.

Low dimensional and most relevant representation of the 3D OCT data for each eye

EYESAVE: AI-enabled Glaucoma Triage

Glaucoma is a leading cause of permanent blindness, affecting 4% of people over 50 in the UK. It often goes unnoticed until it severely impacts vision, and diagnosing it requires several tests, which can be slow and prone to errors. Other health conditions, like heart disease, also play a role in managing glaucoma.Optical Coherence Tomography (OCT) helps doctors detect eye damage, but reading these images needs specialized training, which is scarce. Our project aims to create an AI tool that combines eye scans and health data to assess glaucoma risk, helping doctors make faster, more accurate treatment decisions.

OCTAGE: Monitoring the ageing brain via eyes

Age-related brain changes impact our lives, and the retina is the only part of the brain we can see directly. OCT and OCTA imaging can capture detailed retina images and are used for conditions like diabetes and glaucoma, but they also have potential for studying ageing and diseases like Alzheimer’s.This project aims to develop AI methods to analyze retinal changes with OCT data, creating a detailed image database from NHS and national resources. The goal is to design AI tools that can detect healthy ageing versus conditions like Alzheimer’s and track changes in the retina to identify early signs of age-related diseases.

Ophthalmologists managed high patient volumes, resulting in increased workloads and inconsistent diagnostic interpretations. This variability in assessments often affected treatment decisions.

Workload and Diagnostic Variability

Expensive diagnostic tools limited access to timely care, especially in underserved areas. Mass screening programs were resource-intensive, and follow-up monitoring was hindered by logistical barriers, making it difficult to reach high-risk populations.

Cost and Screening Challenges

Eye diseases like glaucoma and diabetic retinopathy often progressed silently, with delayed diagnosis due to reliance on manual assessments. Additionally, specialist care was scarce in rural areas, leading to long wait times and worsening conditions.

Delayed Diagnosis and Limited Access

Watch this video to get familiar with the anatomy, parts, and function of the human eye! 👁️

We developed a specific a WEB APP TO COLLECT THE EXACT LOCATION of the cellular layers, called segmentation, by help of the trainee ophthalmologists.

The known data is then processed and a trained network results. This process is called Training stage.

Our algorithm WATCHES AND LEARNS how the human performs the segmentation.

The clinicians perform the segmentation of the cellular layers in a limited number of images and this the KNOWN DATA.

Now, for every new OCT data, 2D and 3D segmentation can be found automatically.

VISTA enables remote eye monitoring, reducing reliance on clinical visits. EYESAVE enhances early detection, improving diagnostic accuracy and minimizing false positives. OCTAGE identifies biomarkers of aging-related eye diseases using advanced imaging.