OcularSky

What it does

OcularSky is a portable, affordable alternative to fundus cameras, used by trained personnel in various settings like pharmacies, schools, and rural areas. It utilizes smartphones with an AI model and a 3D-printed attachment to diagnose AMD, Glaucoma, and DR.

Your inspiration

WHO reports that about half of vision impairment cases could have been prevented with timely intervention. Leading causes include AMD, Glaucoma, and Diabetic Retinopathy. Regular eye check-ups are expensive and often not covered by insurance. Many regions also lack access to healthcare services, making early diagnosis difficult. Fundus cameras, costing at least $2,000, limit availability to certain hospitals and clinics. This inspired the creation of OcularSky, an affordable, accurate alternative using smartphones, AI models, and a 3D-printed attachment, enabling regular eye check-ups in various settings, especially rural areas.

How it works

OcularSky operates as an application on any smartphone, connected to the cloud where pre-trained AI models diagnose the captured images. To use it, the healthcare practitioner places the smartphone on a 3D-printed gadget equipped with a stabilizing tube to focus on the ocular region. The smartphone's camera and flashlight are aligned with the tube. Once properly positioned, the device is placed about 5 centimeters away from the patient’s eye. The healthcare practitioner captures the image, and the application automatically crops it and uploads it to the cloud for instant diagnosis. For a clearer image, it is recommended that the healthcare practitioner uses pupil dilation eye drops as usual.

Design process

Given the sensitivity of health-related topics, our first step was to scientifically prove the accuracy of our approach. We reviewed all relevant research papers from the past decade and identified gaps in both research and commercial applications. Our next goal was to train our AI model to accurately identify the three targeted ocular diseases. To achieve this, we needed a large dataset of images. We utilized multiple open-source datasets captured using fundus cameras and collaborated with an ophthalmologist from Amina Hospital, Ajman, to capture a dataset using a smartphone for training and testing. Upon achieving desirable results, the AI model was deployed to the cloud for fast processing, and we designed a smartphone application to interface with the AI model. During the design process, we aimed to minimize errors from both the AI model and the user capturing the image. To achieve this, we designed a 3D-printed gadget that acts as a stand, allowing the smartphone to sit securely. The gadget features an adjustable design and a tube to focus the flashlight directly on the ocular region for clear results. Finally, once the prototype was complete, we verified its effectiveness with the ophthalmologist by capturing live images to test its accuracy.

How it is different

OcularSky offers a novel alternative to traditional fundus cameras at a total cost of less than USD 200. By leveraging available smartphones and AI, it eliminates the need for expensive medical equipment, making it accessible to a larger group of people. Unlike existing solutions that typically diagnose only one ocular disease, OcularSky enables early diagnosis of three ocular diseases. OcularSky is user-friendly and can be used by any healthcare practitioner or trained personnel, facilitating regular eye check-ups in various settings such as pharmacies, schools, universities, elderly care homes, and especially in rural areas around the world.

Future plans

Like all AI models, continuous accuracy improvements are necessary. We plan to collaborate with multiple hospitals and clinics, both nationally and internationally, to collect a larger dataset for refining and training the AI model. Additionally, we aim to expand the range of diagnosed diseases for a more comprehensive eye health assessment. Revisions to the 3D-printed gadget prototype and smartphone application will be made based on user feedback to enhance the overall experience.

Awards

- Shortlisted IEEE Senior Design Project at the IEEE UAE Student Day. - Shortlisted research at 2024 Advances in Science and Engineering Technology (ASET) International Conferences.