Brains are one of the most important organs. They provide power and instructions to the entire body, and they allow us to interact with the world. Therefore, it is important to detect changes in brain activity as quickly as possible. One dangerous change that can lead to permanent damage is neural silence—that is, when a part of the brain doesn't show any signs of activity.
A region of silence in the brain can be a sign of stroke or a brain tumor. These silences can also result from brain damage or tissue death. Locating the regions of neural silence could ensure patients are diagnosed earlier and allow their treatment to begin sooner.
SilenceMap successfully localized the region of silence of three patients. SilenceMap significantly outperformed the state-of-the-art source localization algorithms.SilenceMap researchers, Carnegie Mellon University
Alireza Chamanzar, an electrical and computer engineering Ph.D. student, has recently published a paper about an algorithm—a computer program that analyzes patterns in data—he created to noninvasively locate silent regions of the brain. The algorithm, called SilenceMap, uses data from an EEG, a device that collects brain activity from electrodes placed on the patient’s head.
Doctors usually order an MRI to analyze brain activity and find a neural silence. MRIs, however, aren’t portable, cannot be used for continuous bedside monitoring, and aren’t accessible to all patients. Some hospitals, especially rural ones, don’t have MRI machines while patients with metal implants, such as pacemakers, are unable to get the test. To combat these issues, Chamanzar recommends using EEGs. Like MRIs, EEGs are non-invasive, meaning no surgery has to be done. Unlike MRIs, EEGs are easy to conduct and portable.
Locating regions of silence pose unique challenges. Usually, doctors aren’t looking for silences—they’re looking for signals. To do this, they look for consistent, large signals in the low hum of the background brain activity. But this process doesn’t work for finding regions of silence. For silences, doctors have to find where there isn’t background noise—think of it like trying to find the only person not talking in a crowd. SilenceMap must use novel techniques, including graph signal processing and artificial intelligence to find the regions of silence. According to the authors, it has been successful.
“SilenceMap successfully localized the region of silence of three patients,” the authors wrote. “SilenceMap significantly outperformed the state-of-the-art source localization algorithms.”
This research will help patients who don’t have access to MRI machines, especially those who haven’t arrived at a hospital. EEGs can also provide continuous monitoring in a way that MRI machines can’t.
Patients suffering from a stroke or a traumatic brain injury (TBI) would receive a faster and more accurate diagnosis with SilenceMap. Stroke and TBI must be treated quickly—these injuries can be fatal. In fact, the longer a patient waits before treatment, the more likely they are to suffer long term effects, such as disability.
Even though SilenceMap has made great strides, there is still future research to be done. Chamanzar and his team must improve the algorithm to allow for multiple or dynamic regions of silence and to increase precision.
The paper was accepted into Nature Communications Biology in October 2020. Pulkit Grover, an associate professor of electrical and computer engineering, and Marlene Behrmann, a professor of psychology, were listed as co-authors. The work has been supported by the Chuck Noll Foundation for Brain Injury Research, CMU BrainHub, the Pennsylvania Infrastructure Technology Alliance, the NSF STTR program, the Center for Machine Learning and Health at CMU, under the Pittsburgh Health Data Alliance, a CMU Swartz Center Innovation Fellowship and a grant from the National Institutes of Health.