New Computational Method Matches Brain Lesions to Impairments

With further development, this tool could not only allow clinicians to make more precise prognoses, but could also help them provide individual patients with personalized treatment plans, investigators say.

This so-called “glassbrain” visualizes the brain’s structural connectivity network and the effects on the network of a particular ischemic stroke lesion, plotted in green. Each sphere represents a different gray matter region, with blue indicating that the stroke had no effect on the gray matter region’s connectivity to the rest of the brain and red indicating that the stroke had some effect on the gray matter region’s connectivity to the rest of the brain. Pipes between the spheres indicate that a structural connection exists between the gray matter regions, with connections not affected by stroke drawn in black and those effected by the stroke in red.
Image credit: Dr. Amy Kuceyski

Dr. Amy Kuceyeski, an assistant professor of mathematics in the Department of Radiology and at the Feil Family Brain and Mind Research Institute at Weill Cornell, pioneered this computational Network Modification (NeMo) Tool, which calculates how much damage a patient has sustained to their brain’s white matter, the tissue that connects different brain regions. When using this tool to evaluate patients with either multiple sclerosis (MS) or stroke, Dr. Kuceyeski and her team, who imported data into the NeMo tool that was derived solely from earlier MRI images of the brain, found that white matter damage can be used to predict physical and cognitive impairments a patient will experience. This research, which was published in two separate journals in November 2014 and February of this year, shows that using the NeMo tool could provide clinicians with an easy way to deliver more accurate prognoses and rehabilitation plans for their patients.

“The brain is mysterious, so anything that can be done to quantify the relationship between the behavior at the organism level and the anatomy is pretty interesting,” said Dr. Kuceyeski, a former Leon Levy Neuroscience Fellow and lead author on both studies. “If we can improve the accuracy of a prognosis or help clinicians make more targeted treatment decisions, that would be amazing. That’s where the research is going.”

The ability to trace impairments to their root location provides a wealth of knowledge about the way our brains function – knowledge that is crucial in delivering targeted treatments in a clinical setting. Current research is looking at how doctors might be able to examine a patient’s initial brain scan and use the NeMo tool to deliver a more accurate prognosis for the next six to 12 months. With an accurate long-term prognosis, tailored rehabilitative plans can be developed for each patient.

To use the NeMo tool, researchers first analyzed each patient’s MRI images. They superimposed the damaged brain area onto an atlas derived from healthy subjects, and the computer output a score that indicates what percentage of the connective tissue – or white matter – was lost for a given functional area, known as gray matter. The researchers then tested the subjects’ cognitive and physical abilities to glean which impairment corresponded to which brain location. This ultimately allowed researchers to make connections between the location of a patient’s initial lesions and their impairments.

Dr. Kuceyeski and her colleagues used the tool in two different studies. The first, published November 2014 in the American Journal of Neuroradiology, investigated processing speed in patients with MS who were receiving care at the Judith Jaffe Multiple Sclerosis Clinical Care and Research Center at Weill Cornell and NewYork-Presbyterian Hospital.

Processing speed, which controls attention and short-term memory, is important to carrying out many everyday tasks. To test this ability, the scientists administered the Symbol Digits Modality Task (SDMT), which asks patients to decode symbols into their number equivalents by using a legend. Using data collected by Dr. Susan Gauthier, an associate professor of clinical neurology and of clinical neuroscience at Weill Cornell, the researchers measured patients’ ability to complete this task, and also looked at the amount and the location of their brain damage. Using the NeMo tool, they found that processing speed was affected by damage to connective white matter towards the back of the brain that joins visual integration areas. This result shows that these tracts play an important role in processing speed, and that damage to these pathways will likely inhibit the same function in future patients.

Dr. Amy Kuceyeski
Photo credit: Weill Cornell Art & Photography

The second study, published in Human Brain Mapping in February, focused on patients with ischemic stroke who were being treated by physicians in the Division of Rehabilitation Medicine at Weill Cornell. Using data collected by Dr. Michael O’Dell, a professor of clinical rehabilitation medicine, and Dr. Joan Toglia, a senior lecturer in rehabilitation medicine, the investigators reviewed patients’ ability to complete 18 cognitive and physical tasks using tests that assessed everything from their motor ability to their attention to their ability to live independently. Along with these findings, researchers also analyzed damage to the connective tracts between different brain regions in an effort to form correlations between brain damage and the individual’s impairment. Because many of these functions – like activities of daily living – employ multiple brain regions, it’s not as easy to predict how a brain lesion will impact a patient. Nevertheless, the researchers explored these more elusive functions and mapped them to their corresponding brain locations.

The investigators found that the NeMo tool predicted impairments, on average, two to three times more accurately than predictions based on patient information — age, gender, etc. — and lesion size alone. Additionally, findings about processing speed mirrored the MS study, with damage to the connective tracts of visual areas once again corresponding to processing speed impairments. Obtaining the same outcome in a different population demonstrated that the NeMo tool was reliable and effective, and gives further confidence in the structure-function mapping of processing speed.

“The location in the brain of a patient’s disease or impairment really matters,” Dr. Kuceyeski said. “There was no quantitative map to go between where the brain lesion is and what impairment the person is going to have. The whole idea behind this work was trying to quantify those relationships with the ultimate goal of improving clinical care.”

Weill Cornell Medical Colleg