AI in Eye Disease Detection & Intervention

What is Project A-Eye

a. Project Objective

Worsening eyesight has devastating personal and socioeconomic consequences for patients, families and communities, despite being potentially preventable and treatable.

Almost all vision impairment and blindness from diabetic eye disease can be prevented through diabetes management, early detection of eye problems through regular eye exams, and timely treatment.

The International Council of Ophthalmology survey shows that despite there being over 200 000 ophthalmologists worldwide, there is a current and anticipated future shortfall on the number of ophthalmologists in both developing and high-income countries.[1]

Project A-Eye is a product development project, aiming to harness the capabilities of artificial intelligence & machine learning in the prevention, detection and treatment of eye diseases in clinical practice to improve clinical efficiency and patient throughput.

The project aims to develop an algorithm that automatically analyses data generated by optical  cocherence tomography (OCT) scanners, which are one of the most important ophthalmic diagnostic imaging devices.

The algorithm will support and/or complement the work of medical professionals. The product development is justified by the growing and almost insatiable demand for human resources in ophthalmic imaging diagnostics.

The finished product will be a cloud service offered to eye clinics (B2B) to screen, monitor and track patient disease progression. The technology is inexpensive, fits into the clinical workflow, increases patient throughput and enhances clinical efficiency of highly trained expensive staff by 80%.

b. Methodology

Six ophthalmology specialist (“observers”) with extensive experience in examining OCT images were engaged to “teach” our artificial intelligence. The AI algorithm was taught to:

  • first filter out scans that are not usable for analysis
  • separate healthy eyes from disease impacted eyes
  • establish severity of symptoms and urgency of treatment

c. Results

The project has processed over 100,000 Radial Scan images made by an Avanti OCT device, manufactured by OptoVue, Inc. (Fremont, CA, USA). These scans are the latest technology in retinal, optic nerve and anterior segment OCT imaging to deliver unprecedented views and in-depth analysis of ocular structures. The scans were made in a real-life environment.

For machine learning and analysis convolution neural networks, ResNet and DenseNet models were used.

The project accomplished a critical milestone in 2022 when the precision of the algorithm reached that of human observers.

d. Validation of Results:

Eye Clinic angel funded research (200,000 + EUR invested)

Clinical validation successful

Complete working Proof of Concept has been built

Strong professional team has been set up with Ophthalmology, AI, IT & Legal specialists

3 Ophthalmology Departments of different medical  universities support the project with continuous data source

There exists no known competitor has similar working product or service

e. The Investment Ask & Offer

Funding is required for the following purposes:

Expanding image base to include non-European populations

Expanding knowledge base to identify and track rare diseases

Software development for server and client side

Business development, awareness campaigns

Certification, clinical roll-out

Offer

A minority stake in the operating company that holds the IP and provides the service to eye clinics

Founders

  • dr. Gyorgy Szabo – Angel investor
  • dr. Arpad Dunai – Ophthalmology
  • Gabor Armuth – IT
  • Gyorgy Retek – Engineering
  • dr. Peter Bajcsi – Legal

Download

To download relevant files please check our Downloads section.

Contact

Project A-Eye is implemented in Budapest, Hungary.

For further information please contact:

dr. Peter Bajcsi
peter@bajcsi.hu
+36203236033


[1] https://icoph.org/advocacy/data-on-ophthalmologists-worldwide/
Downloaded 15 May 2023.