Visual Diagnostics Become More Accessible

iStock/Feternalcreative

“The eyes are a window to the soul,” the saying goes. From a clinical perspective, the eyes—via the optic nerve—are also a key conduit into the brain. For decades, researchers in psychiatry have been trying to understand this eye-brain connection, with the goal of developing objective, optical diagnostic tests. Experts say that such tests could be particularly useful for disorders that may hinder a person’s insight, such as schizophrenia, autism, and dementia.

One promising area involves tracking someone’s gaze; many studies have scientifically validated the use of eye-tracking to detect psychiatric conditions, but until recently financial and technical hurdles have kept such tools from being widely implemented. Now, as phones and tablets with powerful processors and high-quality cameras become widely available, obtaining a an eye-tracking test may be as simple as downloading an app.

Geraldine Dawson, Ph.D., the director of Duke University’s Center for Autism and Brain Development, has been leading one effort to bring eye-tracking technology to routine pediatric care.

“It has been well established that children with autism pay attention to the world differently; they tend to focus on the nonsocial elements of a scene,” Dawson said. “Tracking a child’s gaze can be a way to screen for autism.”

Technology companies such as Tobii have developed tools that use infrared light to precisely track children’s eye gaze, but these devices are expensive and must be calibrated before every use.

Dawson and Guillermo Sapiro, Ph.D., the James B. Duke Distinguished Professor of Electrical and Computer Engineering at Duke University, set out to create a low-cost eye-tracking app. They created software that tracks the gaze of children as they watch a series of brief videos (each 60 seconds or less). The app uses the cameras available on tablets and phones to track both eye and facial movements of the children as they watch the short videos, which include a conversation between people, people engaged in an activity (for instance, blowing bubbles), and static toys. Children can watch the videos from the comfort of their caregiver’s lap during a routine well-child visit.

A report on the app, published in the August issue of JAMA Pediatrics, described how the eye-tracking app was able to predict which children (aged 16 to 38 months) would be subsequently diagnosed with autism with about 90% accuracy. Dawson noted that the app picked up on previously studied phenomena in children with autism, including their preference for social versus nonsocial stimuli, as well as new differences in gaze patterns. For example, she said, “We found that toddlers with autism had less ability to coordinate their gaze with the flow of a conversation.”

Dawson noted that current clinical checklists used in autism screening are generally reliable but rely on parental reports of children between the ages of 1 and 3 years and rely on identifying behaviors considered typical from a White, European-American perspective. Her clinical work with the eye-tracking app has involved hundreds of diverse families in the Durham, N.C., area and thus far works equally well regardless of the child’s sex or racial/ethnic background. Among other projects, her team is testing how well the app works in infants under 12 months as well as children over 3 years.

Eyes May Offer Clues About Dementia

Visual tests may also aid in detecting people with early stage dementia. “Accurate, early identification of dementia is such a pressing need, yet clinicians have been using the same tools for over 50 years,” noted Chris Kalafatis, M.D., a consultant psychiatrist at the South London and Maudsley NHS Foundation Trust.

As with autism checklists, cognitive tests used to identify people with dementia are prone to bias when given to people of different educational or cultural backgrounds (for example, someone with less formal education may underperform on a word-based test even in the absence of dementia). Cognitive tests also suffer from a practice effect when given repeatedly—people get better with repetition.

A few years back at Cambridge University, neuroscience graduate student Seyed-Mahdi Khaligh-Razavi made an interesting discovery. In visual recognition tests, the brain produced completely different electrical signals when individuals were shown images of animals as opposed to other images. That difference reflects one of the oldest evolutionary jobs of the human brain—the ability to recognize other animals quickly and undertake the appropriate flight or fight response. What’s more, he found that a person’s speed of animal recognition got slower with age or cognitive impairment.

Khaligh-Razavi, along with Cambridge colleague Sina Habibi, Ph.D., and Kalafatis, formed the start-up Cognetivity and developed a five-minute online test known as the Cognetivity Integrated Cognitive Assessment (CognICA). As part of the test, which can be given on any tablet, people are shown a series of black-and-white images for one-tenth of a second. Individuals must then identify whether the image they saw was an animal or not as quickly as possible.

Based on an individual’s response relative to their age, the CognICA algorithm calculates the individual’s overall cognitive health. Habibi noted that because the brain circuit involved in animal recognition passes through multiple brain regions, the algorithm can identify cognitive problems even when only one or two regions of the brain are affected.

Pilot studies found that the CognICA could identify cognitive impairment in 88% of adults with early stage Alzheimer’s disease and that CognICA reliability was not influenced by the participants’ years of education. These studies led to regulatory approval of CognICA in the United Kingdom, and the company is deploying their tool in multiple clinics across the country. Just this month, CognICA was cleared by the Food and Drug Administration, enabling the tool to be marketed as a medical device in the United States.

“We have seen perhaps the most interest from primary care doctors, since this test can close a critical service gap,” Kalafatis said. “It’s hard to regularly follow up with people who screen positive for pre-dementia because the volume is too high and the tools are lacking; the ICA can be used at home, so patients can send in results on their own and physicians can focus on those who are declining more rapidly.”

Dawson said she also sees a future where parents will use eye-tracking apps to monitor their children and provide real-time updates to their physician. “These tools can be used even after a diagnosis is made, for example to track a child’s progress with a prescribed therapy,” she said.

“The COVID pandemic has been an inflection point for digital technologies,” Kalafatis said. “Both patients and providers have taken a greater interest in virtual and online care.” With this new desire to adopt digital technologies, Kalafatis added, “we may finally bring the field of cognitive assessment into the 21st century.”

Dawson’s study was funded by the NIH Autism Centers of Excellence Award, with additional support from the National Institute of Mental Health and other groups. ■