Courtesy of Sylvain Bouix, Psychiatry Neuroimaging Laboratory, Brigham & Women's Hospital
The Head's Kaleidoscope
Exciting Advances in Brain Imaging
by John S. Liu
Think back to phrenology, the art of reading your personality, abilities, and character traits through bumps on your head. Though mostly relegated to the "what were they thinking?" vault of scientific history, the practice bears an eerie resemblance to the exploits of scientists using functional magnetic resonance imaging (fMRI)—this era's most popular brain imaging technique. Some of what phrenology purported to do, fMRI actually does, and with scientific rigor at that. Developed in the 1990's, fMRI offers a quantifiable, noninvasive method for detecting localized changes in brain activity. Unlike traditional MRI and computer tomography (CT), which yield static pictures of the brain, fMRI allows us to pin dynamic processes—thoughts, emotions, and experiences—to concrete neural tissue, all in real time.
The advent of fMRI has helped countless investigators map, Lewis and Clark style, the lay of the cerebral land. What neuropsychologists had to infer from brain trauma patients ('lesions in the left hemisphere seem to lead to deficits in language'), researchers could now probe in healthy volunteers. Moreover, fMRI has opened the door to studying processes absent in species that don't mind having their skulls pried open—most notably language, but more recently morality as well. Indeed, important psychological discoveries have been made in figuring out what cognitive abilities emanate from which wrinkles of gray matter.
Even outside the scientific community, fMRI images, depicting conflagrations beneath the skull, seem to captivate. What is it about fMRI that has people so enamored? Yale psychologist Paul Bloom suggested in Seed that one reason may be the bells and whistles that come along with doing an fMRI experiment: "It has all the trappings of work with great lab-cred: big, expensive, and potentially dangerous machines, hospitals and medical centers, and a lot of people in white coats." But if this seems rather superficial, there may be a deeper reason as well, striking the core of our mental lives. "We intuitively think of ourselves as non-physical," Bloom writes, "so it is a shock, and endlessly interesting, to see our brains at work in the act of thinking."
Metaphysics aside, claims from fMRI scientists can still make us feel a little uneasy. You've heard it all before—Region X is responsible for greed, Region Y for lust. Investigators have pointed to regions that explain why some prefer Coke over Pepsi, and even searched for the "God-spot," the putative 'antenna' that receives signals from the sky. Has phrenology returned, some might ask?
Though it may seem like callous denigration, the charge does bring up legitimate concerns with the method's power to reveal the inner-workings of the mind. FMRI measures brain activity by detecting increases in blood-flow to specific parcels of tissue. The idea is that brain regions working hard during a specific task need more oxygenated blood to fuel them, and this increase is registered by fMRI machines as elevated activity. The problem is that the brain is not a constellation of soloists; it's more of an ensemble, which sounds in harmony. Locating key players and their functions, then, tells us little about the big picture—that is, how these regions work together. Even if one area lights up during sexual arousal, for example, we won't know what input triggered this activity, or the next region involved in the circuit.
Despite such shortcomings, we probably shouldn't be so dismissive as to call fMRI phrenology's grandchild. As David Dobbs observed in Scientific American, the technique is still young. Indeed, it seems "only natural to plot a simple map of cities before delineating the intricate road systems that link them." FMRI may identify important nodes the network, but other approaches are needed to determine how they interact.
Luckily, scientists are working in parallel to fill in the gaps left by fMRI investigation, and the prospects look good. Meet diffusion tensor imaging (DTI) tractography, a new technique that does what fMRI alone cannot: map out the connectivity between brain regions. DTI tractography works by detecting and interpreting the simple diffusion of water. Since nerve fibers are shaped more or less like tubes, water molecules in the brain diffuse more readily along nerve fibers (in parallel) than they do across them (transversely). DTI tractography reads this directionally-biased diffusion of water and outputs a map of nerve fibers stretching from one brain locus to another. Combining these maps with fMRI data allows us to see not only that Region X and Region Y are activated during, say, a psychotic episode, but also that Region X sends projections through Region Z to reach Region Y. It then starts to become clear how all these regions work in conjunction.
But perhaps we want to zoom in some. Perhaps we want to know not only how Region X connects to Region Y, but also how individual nerve cells within Region X connect to one another. After all, neural computation emerges first from discourse between small groups of neurons within brain modules. In light of this fact, a fundamental limitation of fMRI and related technologies is its spatial resolution. As writer David Dobbs puts it, "because each voxel [the unit of signal used to produce the image, several cubic millimeters] encompasses thousands of neurons, thousands or even millions may have to fire to significantly light up a region; it is as if an entire section of a stadium had to shout to be heard." Unlike data gleaned from fMRI or DTI, knowledge of the microscopic connectivity and communication between neurons addresses the precise ways in which information is coded and processed by the brain. Illuminating these more basic circuits is the aim of an exciting new frontier in imaging.
In a 2007 Nature publication, Harvard faculty members Joshua Sanes and Jeff Lichtman unveiled a technique that allows us to see the brain at an unprecedented level of detail and clarity. Their "brainbow" is a stunning image of distinct neurons and glia illuminated in over 90 vibrant colors. Sanes and Lichtman achieved their psychedelic portrait of the mouse brain by expressing combinations of different fluorescent proteins—variants of those purified from jellyfish—in the neurons of transgenic mice. This was something never before possible with traditional fluorescence imaging, which often employed only a few colors. With the palette now expanded, individual neurons can be distinguished from one another and their efferents traced, shedding new light on the intricate connectivity between single cells. Clearly, these images have a lot to tell us: "Knowing the circuit diagram of the brain, like for any other computing system such as a radio or a computer, is very helpful in understanding how it works," Sanes said about the new technique. But "it's going to take many different approaches to figure out the complete story."
A connectomics reconstruction of the rabbit retina. Computer algorithms scan collections of electron microscope images to develop a 3D wiring diagram of neural circuitry.
Courtesy of Sebastian Seung, Massachusetts Institute of Technology
A major drawback of the brainbow approach is that it produces only a flat picture. Like the rest of our bodies, however, the brain is not two-dimensional, but rather a three-dimensional mesh composed of billions of neurons, each cell making potentially thousands of connections with its comrades. In the emerging field of "connectomics," scientists seek to describe the brain in terms of its connections, and further, to understand how information is conveyed and processed by these pathways. Visualizing these 3D networks is no easy task, but computational neuroscientist Sebastian Seung at MIT has been developing a technique that is up for the challenge. In Seung's method, computer algorithms scan chunks of tissue, slice-by-slice, following neuronal outcroppings through two-dimensional electron microscope images to trace the circuitry weaved throughout the tissue. Data collected from this process allows investigators to construct vast wiring diagrams depicting the brain's elaborate connections.
But such a task may be one that is doomed never to be complete. For the brain, it seems, never stands still. The program of connectomics suggests that the brain is hard-wired; but its connectivity is dynamic in response to stimulation. Networks are continually modified, rearranged, and pruned. Were the brainbow to reflect this, it would be ever-shifting, like a whirling kaleidoscope. Still, we're a long way from scalp massages.