Diabetes is the leading cause of blindness in the working-age population. In 2010, over 126 million people had diabetic retinopathy (DR), and it is projected to increase to over 191 million by 2030.
DR occurs when delicate blood vessels in the retina leak blood and other fluids. In advanced stages, abnormal blood vessels grow into the retina causing permanent scarring, vision impairment or blindness. However, early detection and treatment can reduce the risk of blindness by 95 percent.
“Unfortunately diabetic retinopathy usually does not have any symptoms in the early stages,” says Janet Leasher, O.D., M.P.H., co-author of an article published in the peer-reviewed journal Diabetes Care. This fact complicates early detection and diagnosis – and sight-saving treatment.
Health organizations worldwide have instituted screening programs in an effort to decrease DR related morbidity and healthcare costs. To optimize screening accuracy and efficiency, considerable research in automatic retinal image analysis has emerged, and various methods are commercially available. Screening involves taking a photograph of the retina and must be done every year throughout a diabetic patient’s life. While able to detect later stages of DR, most systems cannot reliably recognize the earliest retinal changes. Plus, they require trained personnel to take the photos and manually interpret the results – a significant barrier, especially in impoverished areas of the globe where the prevalence of DR is the highest and the number of ophthalmic professionals is critically low.
Taking early DR detection to the next level, new medtech startup DreamUp Vision has developed an innovative retinal analysis engine based on state-of-the art machine learning known as deep learning. Using big data analysis, the technology takes photos of the retina and can detect tiny changes to identify early stages of retinopathy. Their goal: help healthcare professionals benefit from the latest advances in machine learning to detect eye diseases in their earliest stages.
The technology can be implemented both as an online web-application or mobile application. The retinal photo can be taken by any type of fundus camera and can be analyzed online from any location, including the retinopathy telemedicine centers existing in some countries.
“These days we live in the era of Artificial Intelligence revolution via developments done in the field of deep learning algorithms…This allows (us) to outperform the more traditional machine learning methods in performance, and to address issues that were not available to the machine before,” says Edouard Colas, ophthalmologist and Chief Medical Officer at DreamUpVision. He continues, “At DreamUpVision, we have a mission: to screen diabetic people for diabetic retinopathy in its early stages, and prevent blindness.”