Science Communications - Publicity for Technologya

brain gain
How Pittsburgh's Brain Scientists Are Decoding Our Body-Mind Network

Dr. Walter Schneider spoke with ebullience as we made our way through the labyrinthine maze of corridors, passages, exits, stairways, entrances, sidewalks, loading docks and elevators that connect his office in the University of Pittsburgh's Learning Research and Development Center with UPMC's functional magnetic imaging center across the street. I couldn’t help but wonder if finding my way back to his office would be one of the tests for the brain scan I was about to undergo.

“I still find it amazing that nature has given us a totally non-invasive way of detecting brain activity right inside our bodies that is completely safe and non-toxic," he said as we crossed the street. Professor Schneider, who is senior scientist at the University of Pittsburgh’s Learning Research and Development Center, was referring to the weak but detectable electrical and magnetic signals that emanate from our brains and that have precipitated the emergence of the science of functional brain imaging.

As its name suggests, functional brain imaging attempts to crack the brain-body-mind code by correlating pictures of human brains with sensory, emotional, cognitive and motor activity. In other words, it tries to explain how our brains make us tick.

The field is enabled by advancements in non-invasive imaging tools that can be ethically used on healthy subjects because, unlike many other imaging techniques, they do not require the injection, ingestion or exposure to potentially harmful substances such as radioactive tracers or X-rays. All of the tools exploit the fact that brain activity is marked by electromagnetic signals emitted by the brain’s 100 billion neurons when they fire electrical sparks across any of its 100 trillion synapses to communicate with their neighbors.

Non-invasive brain imaging technologies include: Electroencephalography (EEG) which uses electrodes affixed to the scalp to measure the natural electrical pulses emitted when neurons fire in the brain; Magnetoencephalography (MEG) which senses the tiny, naturally generated magnetic fields in the space around the head; and Functional Magnetic Resonance Imaging (fMRI) which detects the unique resonant frequencies of oxygenated and deoxygenated blood when atomic spins are manipulated by the magnets in an MRI machine. Two technologies derived from MRI, diffusion tensor imaging (DTI) and diffusion spectrum imaging (DSI), both detect water diffused in tissue to determine its physical characteristics. Positron Emission Tomography (PET) is sometimes used to detect specific chemical activity, but with limited frequency due to the technology’s requisite radioactive tracers.

Each method makes trade-offs between image clarity and speed of capture, otherwise known as spatial and temporal resolution. Dr. Schneider’s recent improvements in fMRI analysis have resulted in vast increases in spatial resolution. MEG increases temporal resolution by registering data within two-tenths of a second of a brain event. Numerous analytic techniques are employed to identify functional areas of brain activity and correlate them with the neural pathways that connect them.

Some methods and combinations of methods are better suited to detect neural events in certain areas of the brain than in others. Because the images show only changes in the brain’s physical structure and not the causes of those changes, associations between body/mind activity and physical changes in brain states must be reverse-engineered and compared to sets of normed sample images.

Among the most prominent researchers in Pittsburgh’s brain imaging community are: Schneider who focuses on graphically representing the brain’s five hundred or so functional areas along with their connections and relative strengths; The University of Pittsburgh’s Dr. Mary Phillips, who is working to bring brain imaging science into clinical practice; and Carnegie Mellon University professors Marcel Just and Tom Mitchell who are attempting to crack the brain’s cognitive code by correlating brain activity with word cognition.

Having recently learned about Pittsburgh’s brain pioneers, I volunteered my brain as a living laboratory specimen in a mercurial flight of journalistic enthusiasm for this story. This was the day for my brain scan.

“Nothing I do today in terms of functional brain imaging was possible 15 years ago,” Schneider continued as we walked. Schneider and colleagues published one of the first papers on functional brain imaging in 1992. “We get the time course with EEG or MEG. With diffusion spectrum imaging (DSI) we can make a complete map of a person's brain connections. We can find activity, time course, connection and codes from patterns in the process.”

The time course to which Schneider refers is the key to determining the direction of neural synaptic activity. By measuring the miniscule differences in time between one set of synaptic firings and the next, the origin, path and destination of an entire sensory, emotional, cognitive or motor sequence can be determined from beginning to end. Until very recently, the task of mapping the brain’s neural connections eluded brain scientists due to the sheer number of the estimated 100 billion neurons and 100 trillion synapses along with the fact that their intersecting paths tended to obscure any continuity of direction. Only within the past year has Schneider integrated a mathematical method of determining a neural pathway’s directionality by analyzing the electromagnetic polarity where one pathway crosses another. “Now we can show each of the links in the circuits along with each of the nodes and their edges. That can tell us how many connections there are between the pieces,” Schneider said.

Although brain activity is defined by synaptic firings, monitoring individual neurons and synapses would provide too many tiny images to correlate brain function with mental or physical activity. In response to that problem, an artificial construct called a voxel is employed to make the task of gauging brain activity manageable. Voxels might be thought of as three–dimensional pixels. Measuring 3x3x6 millimeters, each of the brain’s approximately 14,000 voxels contains 1.5 to 2 million neurons. On the clinical side of the brain imaging street, one of Schneider’s colleagues at the University of Pittsburgh, Dr. Mary Phillips, employs the full battery of brain imaging tools in a sub-specialty of functional brain imaging called psychiatric neuroimaging. Phillips’ primary area of interest centers on the interface between normal and abnormal brain function, with a sharp focus on how we perceive and react to the facial expressions of others. She correlates brain data both forwardly and backwardly: images to functionality and functionality to images. “We know what normal and abnormal brains look like,” Phillips said. “Given a picture of someone's brain function we are asking what is the likelihood that they're going to get depression or bi-polar disorder or some other neurological disorder?” Then, taking the opposite approach, “We're also taking pictures of patient's brains that have been diagnosed with disorders and asking the reverse question: what's the likelihood that they are depressed or are likely to become depressed?”

Phillips, a practicing MD psychiatrist, with academic appointments both here and in the United Kingdom, uses fMRI, EEG, MEG, DTI and DSI along with PET (Positron Emission Tomography), which employs a radioactive tracer. “The future is not about using just one technology, but using combinations of technologies,” Phillips said.

Whatever the technology, functional brain imaging works by correlating mental or physical activity with emissions from the brain – electrical with EEG; magnetic with MEG; radioactive with PET and radio frequency with fMRI.

Coincidentally fMRI is descended from magnetic resonance imaging (MRI), a technology developed by Nobel Prize winner and University of Pittsburgh Ph.D., Paul C. Lauterbeur, who improved an earlier technology called nuclear magnetic resonance (NMR) by figuring out how to create continuous-tone graphic images, rather than linear histograms, with it. Lauterbeur's efforts were driven by the fact that when manipulated by a set of magnetic fields, the polar axes of atomic nuclei inside a human body act as tiny radio transmission antennas emitting a unique signal for every different material. For functional brain imaging, the level of oxygen in the brain’s blood indicates neural activity or lack thereof.

The phenomenon is known among the cortical cognoscenti as Blood Oxidation Level Dependency (BOLD). BOLD starts with the idea that, brain activity occurs when neurons fire electrical pulses across the synapses between them. As with many types of firing, synaptic firings require oxygen. But because neurons don’t store extra oxygen, our blood has to deliver it on a just-in-time basis. It is delivered on the backs of the iron molecules in hemoglobin.

At the molecular level, it arrives as the oxygen-rich compound, oxyhemoglobin, which becomes oxygen-poor deoxyhemoglobin once the neuron fires across a synapse. Fortuitously for magnetic resonance technology, oxyhemoglobin and deoxyhemoglobin exhibit two exotic species of magnetism, each of which occurs only at very small size scales. Oxyhemoglobin is diamagnetic, which means it is weakly repelled by external magnetic fields. Conversely, deoxyhemoglobin is paramagnetic, which means it is weakly attracted to external magnetic fields. So when oxyhemoglobin and deoxyhemoglobin are influenced by the alternating magnetic fields of an MRI machine, they emit different resonant frequencies – radio waves – the key to determining which parts of the brain are active and which are inactive at any given moment.

Working with BOLD and other technologies, Schneider and his colleagues across Pittsburgh are pioneering a new paradigm of brain science that enables the precise identification of the brain’s operative structures, and more significantly, the precise mapping of the neural pathways between those structures. The result is the development of a map of an individual’s brain showing the sizes, shapes, distances, connections and activity levels between some 500 discrete areas in which our senses, emotions, thoughts, words, speech, decisions and actions reside. The map is the brain’s analog of the body’s genome. Schneider calls it the connectome. On that day the data for my connectome would be gathered. After being screened for my medical history to determine my suitability as a subject for the four procedures I would undergo while inside the fMRI machine while wearing a head-coil that would easily have served as a costume in an episode of Star Trek, Dr. Schneider explained the procedures; for the first three minutes, I would undergo a structural brain scan to determine what my brain looked like; then for about 20 minutes I would respond to a series of visual stimuli, such as icons of right and left hands and feet, thumbs and tongues to which I was to respond by moving that appendage. This would show where my body parts and motor drivers live inside my brain. Then I would view black and white images of faces in various states of joy, anger and surprise, to which I was to press a keypad when the expressions matched. This would show where I make decisions and how I respond to facial expressions. Finally, I would be asked Jeopardy questions like, "a female who shares the same mother and father as you," to which he instructed me to think, not say, "What is a sister?" This would point to my language center.

Although in the past brain function has been correlated with specific functional regions, more recently researchers have concluded that brain phenomena like words, images, thoughts and ideas reside in multiple interactive regions across the brain according to a subject’s experience with them. For instance, the idea of a banana may reside simultaneously in the motor cortex as a cylindrical object that must be grasped and peeled with the hand, in the visual cortex as a yellow fruit, and in the gustatory cortex as something sweet and delicious to eat. When viewed under such a model, the network of connections between the brain’s operative regions more accurately indicates brain function than its simple topography. Presumably my connectome would tell me what was going on inside my brain.

Upon being inserted into the barrel of the fMRI machine in a supine position, I breathed a cautious sigh of relief when my latent claustrophobia failed to rear its ugly head. Having been informed that any movement in excess of three millimeters one way or the other would disqualify me from having a digital image of my connectome made, I realized that I had just drawn what would be my last deep breath for more than an hour. As I lay in the fMRI barrel, I had a surreal sense of not being able to determine where the images reflected just inches before my eyes by the exquisitely polished mirror affixed to my head coil stopped and the real-world room that housed the machine began. While I responded to the words and pictures, the machine clanked an arrhythmic beat as its magnets manipulated my brain’s nuclear axes to point either up toward the top of my head, or down toward the soles of my feet for a brief instant. Most of the up-axes would cancel out most of the down-axes. But a few would not be cancelled. Then a set of smaller magnets would throw the uncancelled axes out of kilter for a brief moment, then switch off causing them to relax and fall back into alignment with the others. As the nuclei relaxed, they gave off some of the energy they had absorbed from the magnet that threw them out of kilter. That energy manifested itself as a set of very weak radio waves, whose frequencies the fMRI machine captured to determine the relative concentrations of paramagnetic oxyhemoglobin and diamagnetic deoxyhemoglobin.

After an hour of responding silently to visual and auditory stimuli, I emerged from the machine slightly disoriented, but none the worse for wear. An overnight trip to Pitt’s supercomputer would transform my functional brain data into the multi-colored connectome that showed up in my email three days later and which appears on the cover of this magazine. The image depicts about 10 percent of the 80,000 or so neural fibers detected during my journey into the fMRI machine. Connectome images are frequently reduced in fiber density in order to facilitate clarity. Red indicates high levels of brain activity, yellow - moderately high, and blue-low. The prominent red channel from the lower right to the mid-central section of my brain indicates the neural superhighway that connects my visual cortex with my language area. The image correlates well with my looking at images and thinking of related words while in the machine.

Functional brain imaging has found practical application in presurgical planning for patients with brain tumors. By understanding with precision where the functional areas of a person’s brain are located and how they are connected to each other, surgeons can avoid damaging areas essential to a patient’s lifestyle or occupational activities while excising a tumor.

Walter Schneider’s recent breakthrough in sorting out intersecting neural pathways has given brain researchers a powerful new tool for decoding brain activity. “We now have the technology to represent the brain’s 500 areas along with their connections and relative strengths, and I could not have said that a year ago,” he said.

In the longer term, Dr. Phillips sees three future paths for psychiatric neuroimaging. “the first is diagnostics and we're closest to that,” she said. “The second is therapeutics. By combining fMRI with PET, we can index neurochemical activity, which will allow us to get to the chemical basis of abnormal neural activity and fine-tune pharmaceutical therapy. Third, with intervention, we can look at the brain and ask the question, ‘Is this person likely to acquire this disorder?’ The earlier we step in, the more likely people are to have a better life.”

At the far reaches of brain imaging research, CMU Professors Marcel Just and Tom Mitchell, are engaged in yet another variant of functional brain imaging called cognitive brain imaging. The pair, whose research has been covered by Science magazine and 60 Minutes, has developed a computer program that employs a set of 60 common nouns and 25 verbs as stimuli for the generation of a normed set of fMRI brain images. By matching the words with their corresponding brain images and comparing the co-occurrence of the words in the normed set with a data base of one trillion words collected by Google from the World Wide Web the program is able to predict the fMRI brain images that would be produced by new words, not contained in the database. The significance of Just’s and Mitchell’s work reaches across the mind/brain paradigm into that of artificial intelligence.

“People have been communicating by means of speech for thousands of years,” Just said. “Now we can capture the regularities of speech communication and correlate them with the regularities of brain imaging to crack the code of human brain activity. That's the Rosetta Stone.”

This article first appeared as a TEQ cover story.

© Copyright 2008, Thomas P. Imerito / dba Science Communications

Bookmark and Share

©2009 Science Communications