How AI can help diagnose neurological disease
A retinal scan and AI can detect tiny eye movements that can help diagnose and treat difficult neural disorders much earlier than before.
More about artificial intelligence
What if you could diagnose multiple sclerosis (MS), Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis (ALS), and concussions with a 10-second scan of the retina? Now, a combination of cellular-level imaging and artificial intelligence (AI) presents that possibility. It tracks movements of the retina instead of pupil movement. Because the retinal tracking technology measures eye motion on a cellular scale, it can detect movements as small as 1/100 the size of a human hair, which is about 120 times more sensitive than other eye-tracking systems.
The new technology captures the best of big data imaging and the artificial intelligence that operates on it. Imaging is more granular, so both AI and machine learning (ML) are able to operate on more data points. This has the capability of revealing not only eye health, but underlying medical conditions so the conditions can be treated.
“We know that disease conditions such as multiple sclerosis are characterized by specific eye movements,” said Dr. Christy Sheehy, CEO of C. Light Technologies, a neurotech and AI company. “With retinal scanning technology, we can detect small changes in eye movements sooner, with video visibility down to the one-micron level. In other pupil scanning technologies, the smallest size for eye movement observation is in the 2- to 4-millimeter range.”
More microscopic observation and detection of eye movements and what they mean pave the way for better medical outcomes because disease conditions are detected earlier. Early detection is facilitated with the help of artificial intelligence that uses a framework of neural networks and advanced statistical modeling.
“Each video that we take of the eye is approximately 10 seconds in length,” Sheehy said. “It contains over 5,000 data points, and records six of these videos per patient.”
With the assistance of machine learning that can detect and assess the data patterns of eye movement, present retinal scans can be compared to past results. “This enables us to chart the progression or lack of progression of a disease,” Sheehy said.
In the case of MS, which is the first disease the technology is being applied to, double vision and instances of involuntary eye movement are symptomatic.
“We can learn a lot from this information,” Sheehy said. “On the one hand, a change in eye behavior could denote a progression of the disease—but it could also mean that a particular drug being used to treat the disease isn’t effective. In this latter case, it could signal a care provider to reevaluate and adjust the drug therapy being used so the patient can achieve better results.”
What can others learn from this AI technology?
Technologies like advanced retinal scans and AI/ML evaluation offer lessons for others who are enabling and implementing technologies that operate on big data.
- It’s not necessarily enough to settle for what imaging and other big data capture technologies offer when it’s possible you might uncover more meaningful data visibility if you sharpened the lens of the equipment that was harvesting the raw data points.
- Artificial intelligence, which uses an intellectual framework for assessing and evaluating a specific problem, and machine learning, which can add to AI performance by detecting and analyzing patterns of data, are a dynamic combination.
- Some of the biggest big data, AI and ML breakthroughs occur when experts in a field of study or a business put their empirical knowledge to work, focus on specific problems that have been hard to solve, and get them solved.