With the help of a service-oriented team, the Sankara Eye Foundation India is a nonprofit organization operating throughout all of India that seeks to eradicate blindness that is curable and preventable. Diabetic retinopathy (DR) is a leading cause of preventable blindness in adults in India. The Sankara Eye Foundation and Leben Care, a Singaporean company, worked together to develop the cloud-based AI software platform Netra in order to solve this problem. Based on Intel-powered technology, Netra.AI uses deep learning to quickly and accurately diagnose retinal problems in patients, matching the accuracy of medical professionals.
Difficulty
India is home to one of the highest numbers of diabetics worldwide; by 2030, there could be an alarming 98 million cases1. India has the biggest population of visually impaired and blind people in the world, with over 55 million visually impaired and 8 million blind. For people who are of working age, DR is the primary cause of blindness and vision loss. Consequently, halting the damage requires early detection and therapy. But given that 70% of Indians live in rural areas, it is extremely concerning that there aren’t enough qualified ophthalmologists to diagnose DR, particularly in isolated areas. Additionally, there is still a significant disparity in the nation’s patient base and the availability of the infrastructure and necessary medical services.
Resolution
In order to overcome this obstacle, the Sankara Eye Foundation and Leben Care joined together to put into practice a cloud-based artificial intelligence (AI) solution based on the Intel® Xeon® Scalable CPU architecture and driven by Intel® Deep Learning Boost (Intel® DL Boost). These CPUs can achieve even higher levels of embedded AI performance thanks to Intel DL Boost. Leben Care was able to implement its intelligent solution—Netra.AI, a comprehensive retina risk assessment software-as-a-service platform available on cloud—by combining the power of the Xeon Scalable platform, DL Boost, and Intel® Advanced Vector Extension 512 (Intel® AVX-512) with Amazon EC2 C5 instances. It is an excellent technique for screening retinal illnesses in a large population with limited infrastructure and resources for tertiary healthcare since it can distinguish between an unhealthy and healthy retina.
Outcome
Intel® architecture-based Sankara Eye Foundation was able to precisely identify DR thanks to Netra.AI, allowing treatment to prevent additional vision loss and reduce the rate of blindness. In just two minutes after uploading the photos, Netra.AI generates a detailed report that allows optometrists or imaging technicians to give patients who need a hospital referral immediate advice. In addition, the solution detected any DR with outstanding sensitivity and accuracy (98.5% and 99.7%, respectively).
Using Innovation to Reduce Drastic Vision Loss
In order to lessen the incidence of DR-related blindness, regular diabetic screening is becoming increasingly important in India. Unfortunately, the lack of qualified retina specialists in India makes it difficult to screen asymptomatic people effectively, which causes patients to present with advanced diabetic eye disease later in life. Fundus photographs can be manually graded by qualified graders or retina specialists as part of fundus photograph-based DR screening, which can be carried out in place of physical screening. AI-based DR detection has created a new avenue for DR screening and is developing quickly. The reduction of blindness in patient outcomes is greatly impacted by the use of this technique in the diagnosis of DR patients who are referable.
To control and cure diabetic retinopathy (DR), all diabetic patients nowadays should be referred to a retina expert. Netra was used by Sankara Eye Foundation India.Artificial intelligence (AI) built on Intel architecture was able to accurately identify DR that poses a risk to vision and significantly lessen the screening workload for vitreoretinal (VR) surgeons.
Netra.AI is a cloud-based, all-inclusive software-as-a-service platform for retinal risk assessment, created by Leben Care in association with Sankara Eye Foundation India. Fundus cameras are used to take pictures of patients’ eyes, which are then uploaded to a machine learning-based platform that gets anonymised patient data via an API or web site. Netra.AI can identify retinal pathologies such as diabetic eye disease (DR), glaucoma, macular degeneration, and others using these photos. These conditions call for emergency medical intervention.
Netra.AI offers the ability to be utilized as both online and offline modules, such as a stand-alone box. It has been trained to be device-agnostic to specialized low-powered microscopes with attached cameras. The method combines state-of-the-art algorithms with a four-step deep convolutional neural network (DCNN), created in partnership with top retina specialists. This neural network assists in the following tasks: distinguishing DR stage, identifying general quality distortion for automated picture quality assessment, distinguishing retinal photographs from non-retinal images, and tagging lesions based on pixel density in the fundus images.
The solution is an excellent tool for screening for retinal illnesses in a wide population, particularly in those with inadequate infrastructure and resources for tertiary treatment, as it can distinguish between an abnormal and normal retina. The model has been taught to recognize the various phases of diabetic retinopathy and to recommend early referral or ongoing surveillance for the patient. Moreover, it can detect glaucoma, which might be a useful tool for early detection and treatment of this degenerative condition that eventually results in blindness.
