News

Tracking eye movements in search of Alzheimer’s

A new tool detects mild cognitive impairment early by using big data analytics and cloud computing.

Alakananda Dasgupta

doi:10.1038/nindia.2019.172 Published online 30 December 2019

MindEye uses eye-tracking technology to detect mild cognitive impairment (MCI), a precursor to Alzheimer’s.

© IIT Gandhinagar

Alzheimer’s disease is a degenerative brain disease and the most common form of dementia. A promising new tool to help in the early diagnosis of Alzheimer’s is being developed by scientists at the Indian Institute of Technology (IIT) Gandhinagar. The non-invasive tool tracks eye movements in response to visual stimuli presented on a computer screen. 

The project, nicknamed 'MindEye', uses eye-tracking technology to detect mild cognitive impairment (MCI), which often begins years before more visible signs and symptoms of Alzheimer’s become evident. The results of the experiment were recently presented at the Faculty of Old Age Psychiatry Annual Conference, organized by the Royal College of Psychiatrists, in Nottingham, United Kingdom.

MindEye is a collaboration between Uttama Lahiri at IIT Gandhinagar, Anirban Dutta at the University at Buffalo, New York, and Abhijit Das at AMRI, Kolkata. It is a low-cost, user-friendly device, integrated with computerized cognitive tests, big data analytics and cloud computing, underpinned by affordable software.

Irregularities in the eye movements of people with Alzheimer’s were first observed in the 1980s by J Thomas Hutton, former director of the Texas Tech Alzheimer’s Center. People with Alzheimer’s show subtle derangements in saccadic eye movements (rapid eye movements that reposition the eye from one fixation point to another, with each fixation lasting 30 to 80 milliseconds) very early in the disease. This derangement, called saccadic latency, increases as the disease progresses.

Lahiri explains, “The subject is presented with multiple visual stimuli in the form of black dots against a white background in various quadrants of a computer screen, which has an in-built high-resolution infrared camera.” In one task, the saccadic latency is measured. In another, the subject is asked to recall the sequence of dots and reproduce the appropriate movements of their gaze. Six such sequential tasks measure subtle changes in the eye movements and, in turn, cognitive decline. The device also measures the angle of deviation, or rotation of the eyeball, and the trajectory made between the fixation points.

There are other tools for diagnosing Alzheimer’s early. Some, like amyloid positron emission tomography scans, are very expensive. Others, like spinal taps, are invasive; and still others, like the Mini-Mental State Examination, and the Montreal Cognitive Assessment, require basic literacy. What sets MindEye apart from these other tools, emphasizes Das, is its easy delivery in a community setting. Although there are forerunners of this type of work, never before has such a large cohort of people been tested – so far, roughly 1,700 people have been recruited in the clinical trial, which is being conducted in the West Bengal state of India.

Dutta says, “India will provide big data for training a better classifier or biomarker than what could be available in the US or EU.”

The device will eventually be developed as an app-based screening tool for mobile health, akin to tools such as the hand-held ultrasound scanner, called Butterfly iQ.

Dutta concedes that MindEye has high sensitivity, but low specificity, a deficiency pointed out by Hutton. “My concern is not for the sensitivity but for the specificity, as many drugs and other illnesses also affect eye-tracking performance,” Hutton says.

Can MindEye detect dementia even before the onset of MCI? Yaakov Stern, a professor of neuropsychology at Columbia University, New York, says, “We now know that the Alzheimer's process begins way before MCI.” Das says though in theory MindEye can diagnose Alzheimer’s before the onset of MCI by measuring the saccadic eye movements, this needs to be corroborated with more data.