Artificial Intelligence & Machine Learning in Ophthalmology
Machine Learning (ML) and Artificial Intelligence solutions have shown powerful diagnostic performances, especially to interpret and analyze medical images and identify pathological lesions. In ophthalmology, deep learning and machine learning are being used to identify diseases, study images, make accurate intraocular lens calculations, and improve surgical outcomes.
Early detection of eye problems is critical as it helps identify conditions such as diabetic retinopathy, glaucoma, and age-related macular degeneration (AMD). With its deep learning and image analysis algorithms, AI can detect lesions’ presence and the extent and grade the disease based on accepted clinical standards. One of the most commendable recent findings was that a deep learning algorithm could reliably predict fundus images for cardiovascular risk factors and demographics through an AI prediction model.
Several startups harness AI to provide ‘visual solutions’ through deep learning and machine learning techniques. Leben Technologies Care Pte. Ltd., a startup in Deep Learning AI healthcare, creates ways to resolve the global question of preventable blindness. With a 4-step Deep Convolution Neural Network, Netra.AI uses cutting-edge algorithms developed in partnership with some of the world’s leading experts in Retina. Netra.AI allows for automated retinal image processing and point of treatment, thereby reducing the frequency of blindness by immediately recognizing at-risk patients and providing medical practitioners with insights to allow better diagnostic results.
OrCam leverages the power of artificial vision by integrating groundbreaking AI tech into a discrete wearable platform. This device claims to empower the lives of visually impaired people as it offers a simple, easy-to-use interface to help visually impaired people in their day-to-day activities. RetinAI enables the clinician in ophthalmology, such as analysis of eye-related diagnosis, clinical workflows, and patient monitoring.
Collaborative projects employing AI to assess the neovascular age-related macular degeneration (nAMD) using optical coherence tomography (OCT) imaging is being undertaken by TRetinAI Medical AG and Novartis Pharma AG. Many are using RetinAI Discovery, a data management software, to analyze imaging data, compute imaging biomarkers, and follow up changes over time for many commonly occurring eye diseases.
In diabetic retinopathy detection, EyeArt AI Eye Screening System is available as a validated AI technology. Over two million clinical images from more than half a million patient visits worldwide were used to train the machine. It works by analyzing the retinal images of patients obtained using an integrated fundus camera and identifying the disease in less than a minute.
C. Light Technologies, in collaboration with the University of California (UC), Berkeley’s SkyDeck accelerator aims to open new windows to neurological health by measuring the minute motions of the human eye. Retinal movements are tracked on a cellular level using machine learning techniques with the help of a new tracking scanning laser ophthalmoscope device which is also a modern way to test people for disorders such as amyotrophic lateral sclerosis, Alzheimer’s or Parkinson’s disease. Pr3vent is an AI-based startup that monitors newborn eye conditions by screening their retinal images to highlight possible eye problems that need follow-up and care. The company is in the FDA pre-submission stage and hopes to get clearance by the end of 2020. Similarly, Medi Whale Inc. is a startup that owns an auto-diagnostic system in which artificial intelligence software and fundus camera are integrated for ophthalmologic purposes.
AEYE Health uses its camera-based AI software and neural networks to compare the images from a patient’s eye scan with AEYE Health’s online database of healthy and diseased eye images. It makes screening of the eye hassle-free as the company’s software can process and deliver the results of a scan of the fundus (the lower part of the eyeball opposite the pupil) in just a couple of minutes from any camera. Even a standard smartphone will do!
Eyewey owns a particular Nexart-developed Visual AI system that recognizes a much more comprehensive range of things through deep learning. It also understands texts and recognizes speech. Image Recognition and Deep-Learning to generate Visual AI make the app user-friendly.
Dhi Ajna platform is a cloud-based tool that employs Artificial Intelligence, Machine Learning & Deep Learning to help diagnose diabetic retinopathy, pediatric cataract, glaucoma, and macular degeneration. It specializes in imaging of the fundus and provides disease predictions.
The Future is bright.
AI is commonly used for decision making than an actual diagnosis. Due to the shortage of professionals in ophthalmology and long waiting queues, there is a need for autonomous diagnostic screening by AI and decision making.
Although AI models based on deep learning give robust results in the research environment, care should be taken before implementing it in a real-world clinical setting. Although significant advances have been there in image-based classification using deep learning, some drawbacks still exist, such as inadequate clarity, weak integration with previous hierarchical experience, and inflexibility. Nonetheless, developments in AI and ML in ophthalmology has delivered an excellent performance that surpasses human abilities now has reached a tipping point.
AI will not replace human ophthalmologists but only be a digital helping hand to gather information from large quantities of multivariate data and enable quick and accurate decision making. Clinicians should leverage the full potential of AI in a clinical and societal context to bring about transformation in the field of ophthalmology.