Artificial intelligence (AI) has emerged as a transformative technology across various fields, and its applications in the medical domain, particularly in ophthalmology, have gained significant attention. The vast amount of high-resolution image data, such as optical coherence tomography (OCT) images, has been a driving force behind AI growth in this field.
AI-based algorithms have by their precision, reproducibility and speed, the potential to reliably quantify biomarkers, predict disease progression and assist treatment decisions in clinical routine as well as academic studies.
ANAT LOEWENSTEIN, MHA, Professor and Director of the Division of Ophthalmology at Tel Aviv University, illustrated the important role of AI in the use of home OCT
READ THE FULL STATEMENT BELOW
“I’m going to talk about the need for home OCT for all the people, not only for dermatologists. It is a tool for monitoring retina disease. I’m going to focus on home OCT and specify the role of AI in the use of home OCT, the concept of monitoring home OCT, and the evidence that we have on its performance. We have had a great breakthrough last decade in the management of macular degeneration disease, which is the main reason for blindness in people 60 years of age or older in the world.
So we have a new treatment, for which we can give injections in the eye that contain anti-vascular factors, stop the disease, and prevent blindness. However, despite effective treatments that have been proven in many tries, still only about a third of the patients maintain a functional vision. The reason is that the results in the real world are different from the ones in pivotal trials because this needs frequent injections, which cause an unacceptable burden on physicians, patient’s caregivers, and the system, and this brings along a huge non-compliance, with a very important impact on the clinical outcomes.
Here you see OCT images of two eyes of the same patient. On the left side, you see how the patient came to the presentation. On the right side, after two months. You can see the annotation in green by AI showing where the fluid is. It’s very similar after two months in both eyes. However, in the right eye on the top, in between the visits there was no fluid, while in the left eye, there was fluid all along, which could mean that we need more or different treatment.
The home OCT shows what happens in between the visits. The images are required in between the visits. At home, they’re automatically quantified by an AI-based algorithm, and the fluid volume is then plotted as a trajectory over time, giving us very granular information on the temporal fluid dynamics. It alone brings personalized medicine.
So why is AI critical here? Near-daily remote monitoring is not feasible with AI. The physicians cannot practically review all the home images acquired by the patient. The AI-powered algorithm automatically detects and quantifies the presence of subretinal and intraretinal fluid, creating a fluid map, and AI converts these 3D volume scans to 1D fluid volume trajectories. So the data collected over time is displayed longitudinally, giving the physician a very easy way to look at it.
How do we implement the home OCT? Very important challenges are patient training, adherent testing at home, and responding to notifications based on testing. This is the flow. It is not something you buy at the supermarket, the retina specialists prescribe the technology, the patient is referred to the monitoring center, which sends the device to the patient and then takes care of logistic and technical support, and the patient is asked to perform daily scans, all the images on the Cloud are analyzed by AI, and the notification is sent to the physician, either if there is a certain threat or either the physician requested it.
We have a lot of evidence accumulated over the years: practicality (self-image and home use), accuracy (automated fluid quantification), and value (impact on patient management). For the ability to use the home OCT, we have many studies. The most important is the one performed in the USA, involving more than 300 patients, which showed that 90% of these elderly patients could use the device. For accuracy, we have many studies. All of them showed a high correlation between the grading of very expert physicians and the technology. If there’s a significant amount of fluid, then there’s 100% accuracy.
For the impact on patient management, we have very little data because only know we have longitudinal studies, but we have a small study on 15 patients in the US, that showed that when the decision were based on the home OCT, then the interval between injections could be extended from 8 to 15,3 weeks, a very good reduction on the burden of the patient and of the physician. Also, it has been shown that if the retina specialists looked at the results of home OCT, sometimes even if the patient was scheduled to be treated according to a certain regimen, 42% of the time it didn’t even treat him, and in case it was given, the treatment was given a week earlier, so in the more timely manner.
The physician can decide when to do the injection. The physician and AI need to go hand-by-hand, in order to bring benefits to our patients. We have multiple studies that showed the usability of home OCT in AMD patients and the role of AI is essential in remote monitoring services like home OCT. Moreover, physicians led by AI can reduce the burden on practices, while keeping physicians in control of patient management. All of this is to bring better incomes. Thank you very much.”
The FINAL DECLARATION of the 23rd Infopoverty World Conference is now available! The Plan of Action including a list of projects and proposals that emerged from the discussion will be available soon. STAY TUNED!
Comments