Singapore National Eye Centre

Artificial Intelligence: Improving Diabetes Retinopathy Screening and Treatment.

The Singapore National Eye Centre’s Medical Retina Departmentmanages conditions affecting the retina that are not amenable to surgical treatment. Non-surgical treatment options offered include laser and medications.
A wide range of conditions are currently under the care of the department, the most common of which include diabetic eye disease.
Other conditions managed by the department include retinal complications associated with myopia, heredofamilial retinal dystrophies, myopic retinal degeneration and retinovascular eye disease.
The department is equipped with state-of-the-art outpatient suites in the Diabetes and Metabolism Centre (DMC) and is supported by a comprehensive range of the latest imaging equipment for diagnostic and monitoring purposes.
Treatment capabilities range from full-suite laser facilities (including high-speed multispot Pascal lasers and photodynamic therapy) to a purpose-built intravitreal injection suite with a HEPA filter.
Diabetes Mellitus (DM) is the world’s fastest growing chronic disease. There were 415 million people worldwide who have diabetes, and this number is projected to increase to 642 million in 2040. Diabetic retinopathy (DR), a major side effect of diabetes caused by “leaky” blood vessels in the retina, is a leading cause of vision loss. To prevent DR-related visual impairment, early detection via screening and prompt treatment are necessary.
Retinal photography is the most common and cost-effective DR screening method, where photographs of retinal scans are assessed and graded by trained eye doctors, optometrists, or professional graders. With increasing numbers of diabetics worldwide, the manual assessment of millions of photographs will become unsustainable in the future.
SERI’s research team, in collaboration with the NUS School of Computing, has developed a state-of-the-art artificial intelligence (AI) system to automatically assess retinal photographs and replace human graders, which can significantly reduce public health care costs while maintaining its standards.
This technology incorporates superior AI-based algorithms, which were based on the largest dataset of Asian eyes in the world. The new algorithms were able to reach over 90% sensitivity/specificity for referable DR cases and massively upscale current DR screening capabilities.
The technology is currently undergoing side-by-side clinical validation in order to facilitate its integration into the Singapore Diabetic Retinopathy Programme (SiDRP). It has been licensed to a multi-national firm specialising in imaging in a deal covering major global markets, and has since received substantial enquiries from multiple private entities for licensing.