A Simpler Window On The Brain

Imagine donning something as lightweight as a sweatband or a bicycle helmet and getting information about which parts of your brain are active. That's not as far-fetched as it might sound.

Devices for making such measurements already exist and are being used in research labs to tell whether someone is lying or how a patient is responding to stroke rehabilitation. These applications hinge on monitoring the brain by functional near-infrared spectroscopy (fNIRS). Near-IR light sources and detectors in the headgear track brain activity by detecting blood flow to regions of the brain with high energy demand.

The current gold standard for mapping changes in the brain is functional magnetic resonance imaging (fMRI), which has exquisite spatial resolution. But as patients who have been immobilized inside the huge magnet used for MRI can verify, it doesn't map brain function during everyday activities. The slightest movement can render the brain maps useless. Plus, fMRI has poor temporal resolution.

However, fNIRS avoids some of these weaknesses and may eventually provide an alternative to fMRI, particularly in applications requiring less expensive and more portable forms of analysis.

Scientists first suggested the use of near-IR spectroscopy for brain monitoring 30 years ago, but it wasn't until the early 1990s that the first measurements of brain activation were made by this technique. Britton Chance, professor emeritus of biophysics at the University of Pennsylvania, pioneered brain monitoring with fNIRS. He has continued to push the technology in collaboration with a team at Drexel University. Despite this long history, fNIRS has so far remained a research tool, but it now shows promise as a clinical method as well.

The method pinpoints brain activity by measuring blood flow and tissue oxygenation. "When a region of the brain is activated, those neurons are active and require more energy," says David Boas, director of the Photon Migration Imaging Lab at Harvard Medical School, one of the groups developing fNIRS. "Blood flow increases locally to that region to support increased energy demands."

What's behind fNIRS is hemoglobin chemistry. The technique measures the near-IR absorbance in blood of hemoglobin with and without oxygen. The total amount of hemoglobin is proportional to the blood volume and thus reveals information about blood flow. The difference between the two forms of hemoglobin indicates the amount of oxygen being used by tissues for metabolism; fNIRS records both the blood flow and the tissue oxygenation.

Although near-IR radiation penetrates only about a centimeter into the brain, this seeming limitation proves not to be much of a problem. "The functioning part of the brain is mostly the cortex, and most of the cortex is close to the skull," Boas says.

Probes using several lightweight configurations of light sources and detectors are available, giving researchers access to different regions of the brain. Boas's team uses headgear made of plastic strips with holes for positioning optical fibers at regular intervals over much of the head. They are collaborating with TechEn, of Milford, Mass., to make commercially available research probes. The team that Chance collaborates with at Drexel University, led by biomedical engineering and electrical engineering professor Banu Onaral, uses a probe that sits across the forehead. The Drexel researchers thereby avoid interference from the patient's hair but they are limited to taking measurements in a small portion of the front of the brain, the prefrontal cortex region.

Key questions for clinical deployment of fNIRS are what end point is being measured and whether it actually helps the patient, Onaral says. Her team and others are developing many applications of fNIRS, ranging from monitoring responses to medical treatments to designing better human-computer interfaces.

A promising area is in the treatment of stroke and head trauma, Boas says. In the case of stroke, doctors have historically had few options. "Now, treatments are becoming available, but you really need imaging feedback," he says. The initial diagnosis still requires imaging methods with high spatial resolution like MRI and computed tomography, but those methods aren't good for ongoing monitoring. "Optical imaging can continuously monitor the patient for any changes in brain oxygenation that would result from expansion or resolution of the stroke," Boas says.

Beyond the purely physiological, there are also behavioral aspects to stroke rehabilitation that could be measured. Maria Schultheis, a clinical neuropsychologist at Drexel, collaborates with Onaral's biomedical engineering group to develop fNIRS as a quantitative measure of the effectiveness of stroke rehabilitation. Most rehabilitation relies on behavioral feedback to gauge success. Schultheis wants to correlate such behavioral measures with fNIRS data showing which practices hold patients' attention and help them relearn vital tasks.

The Drexel team is particularly excited about the possibility of using fNIRS to monitor the awareness of patients during surgery. Currently, anesthesiologists keep tabs on a patient's status by monitoring muscle movement, which can fail to provide advance warning that a patient is waking up. However, brain oxygen levels increase when a patient emerges from anesthesia.

Team member Kurtulus Izzetoglu is studying the use of fNIRS sensors in the operating room. The challenge is validating clinically relevant signals that could help anesthesiologists. Surgical team members need to know if a patient is waking up so they can administer more anesthesia if necessary. And even if patients don't fully awaken, physicians want to limit their awareness of the surroundings during surgery. Preliminary data show fNIRS gives the surgical team a heads-up two to three minutes before a patient starts to come out of anesthesia.

In the behavioral realm, researchers are using fNIRS to detect when someone is lying. The Drexel team, in research spearheaded by psychiatry professor Scott Bunce, finds that tissue oxygenation increases when a person knowingly lies. The team is interested in this application both for law enforcement purposes and for the identification of malingerers.

Researchers also are finding that fNIRS reveals techniques that successfully help people learn. For instance, Patricia Shewokis, a professor in the College of Nursing & Health Professions who studies the acquisition of cognitive and motor skills, collaborates with the Drexel team to determine what types of exercises are effective during training and learning of specific skills. They monitor the brain function of people playing a computer game to figure out effective ways to organize practice. They find that randomly ordered exercises benefit experienced users, whereas beginners learn more from repetitive drills-a pattern of learning especially apparent during initial learning stages.

Whether the learning is being done by children or adults, fNIRS can help researchers see what's going on. At Harvard, for example, Boas and his coworkers, in particular Maria Angela Franceschini, are using fNIRS to monitor brain development and response to spoken language among infants. "We know that the brain matures very rapidly in the first years of life," Boas says. "There are significant increases in metabolism in the brain" that correlate with mental activity and can be measured by fNIRS.

At Tufts University, Boston, meanwhile, biomedical engineering professor Sergio Fantini and computer science professor Robert Jacob are using fNIRS to improve the experience of computer users. Using machine-learning techniques to analyze fNIRS data, they characterize the brain activity associated with different tasks. The long-term objective is to use fNIRS data to develop more efficient human-computer interfaces.

Although it is starting to show promise in the clinical realm, widespread acceptance still lies in the future. "It's a long path to introduce any new technology into the clinic, especially in a case where it's doing something that no other technology has done before," Boas says. "The clinical applications will be realized down the road."