In the late 1980s when I first began conducting brain-behavior research that used neuroimaging techniques, I was examining the location and extent of brain damage in individuals with aphasia relative to their performance on verbal memory tasks (Beeson et al., 1993). To determine the location of the brain damage, we examined research participants' CT scans because the use of magnetic resonance imaging (MRI) was not yet commonplace. Comparing lesion location across study participants required spending hours manually tracing the region of damage onto paper copies of a brain template. To examine the overlap of the lesions across participants in our research study, each "standardized" lesion was traced onto a transparency sheet using a variety of fill patterns (e.g., stripes), cut out with scissors, overlaid on the standard template, taped in place, and then photocopied to generate a composite image. This was fairly state-of-the-art at that time.
Today our lab continues to pursue lesion-deficit research with the benefit of current neuroimaging techniques and software advances. High-resolution MRI brain scans are used to acquire 1-mm-thick three-dimensional brain images and several available software programs render 3-D brain volumes. To determine the precise location and extent of damage, the lesion is demarcated on every slice and a lesion volume can be generated. To compare effectively the location of the lesion across individuals and studies, the brains can be digitally modified to fit a standard brain template. This automated normalization procedure stretches and shrinks each unique brain to match the template. Ultimately, images can be produced that demonstrate lesion overlap in large patient cohorts, providing a powerful means to determine which cortical regions are likely to be essential for specific language processes.
The 3-D brain images generated from MRI can be examined to determine the status of the brain tissue (damaged or not) at specific locations. By convention, the brain is parsed into 1-mm cubes (voxels) that are identified based on the directional location relative to the "center" of the brain using x, y, and z coordinates, where x = left-right, y = front-back, and z = superior-inferior. When individual brain images are modified into a standard space, the x, y, and z coordinates for one individual are roughly in the same anatomical location for another individual. The 3-D images not only allow us to examine common and unique regions of brain damage across individuals, but also allow us to examine the regions of damage for a cohort relative to their language behavior. For example, an analysis procedure referred to as voxel-based lesion-symptom mapping (VLSM) calculates the statistical relationship between performance on a given task (such as naming or reading) and the status of a given brain voxel (damaged or healthy). VLSM maps can be generated that show brain areas that are statistically related to a task of interest.
Diffusion Tensor Imaging
The lesion-based approaches to the study of brain-behavior relations for language have focused primarily on the status of gray matter (i.e., the cortical and subcortical tissue made up of neuronal cell bodies), yet most lesions also disrupt white matter connections (i.e., the mylineated axons of neurons that form the fiber tracts interconnecting processing centers within the nervous system). With the development of diffusion tensor imaging (DTI), it is now possible to investigate the status of neural connections in the brain, not just the cortical endpoints. This procedure detects the rate and direction of water molecule movement, which provide an indication of the integrity of the white matter tracts. DTI images are generated from the analysis of specially weighted MRI brain images and are color-coded by convention to indicate the direction of the fiber track (e.g., anterior-posterior, inferior-superior, left-right). The resulting tractography images can be quite beautiful.
DTI is commonly used to examine the integrity of white matter tracts in individuals with stroke as well as other neurologic conditions. The use of DTI in lesion analysis is still in the early stages but holds great promise to help advance our understanding of brain-behavior relationships as it provides a means to evaluate the status of language networks, rather than simply considering the integrity of brain regions in isolation.
Acquired language impairment is not always associated with obvious focal damage. The progressive aphasia syndromes, for example, are associated with gray and white matter atrophy that may or may not be evident to the eye. The decrease in brain volume can be measured using voxel-based morphometry (VBM), in which high-resolution MRI scans are evaluated relative to a cohort of healthy brains. Statistical analyses on a voxel-by-voxel basis ultimately provide brain maps indicating regions of significant atrophy. Most VBM research is conducted using cohorts, but analysis of an individual patient is possible and can reveal subtle volumetric changes that would be difficult to identify otherwise (Good et al., 2001; Gorno-Tempini et al., 2004).
The development of functional imaging approaches that offer a view of brain activity was one of the most exciting advances in brain imaging over the past 30 years. Functional magnetic resonance imaging (fMRI), which detects neural activity based on changes in blood oxygenation, is currently the most common functional imaging approach used by language researchers (Crosson et al., 2007; Thompson & den Ouden, 2008). This noninvasive procedure requires an individual to perform specific tasks, such as naming pictured objects, while lying in the MRI scanner. Functional images are collected repeatedly while tasks of interest are performed, and the data are reconstructed to reflect blood flow changes over time. It is particularly important that the images are spatially aligned correctly, so head movement must be minimized during data collection. Experimental designs are employed that accommodate the constraints of the loud scanner and restrictions on the freedom to move, while still engaging the language processes of interest. Flexibility continues to increase with the development of peripheral devices that can be used in the scanner, such as MRI-compatible microphones, response boxes, and goggles that project images from the computer screen, as well as advances in statistical analyses of the acquired images.
The fMRI results are typically displayed as statistical maps overlaid on brain templates, so that the cortical regions involved in a particular task look like "hot spots." Brain regions activated during specific language tasks can be compared to the lesion locations associated with impairment of the task of interest. The convergence of findings obtained from both lesion-deficit and functional neuroimaging approaches can offer strong support for hypotheses regarding brain-behavior relations.
An important use of fMRI is to examine regions of the brain that support language skills following brain damage. Studies have shown that individuals who have incurred damage to critical left-hemisphere language regions often demonstrate reliance on the brain regions immediately adjacent to damaged region in the left hemisphere (e.g., Crosson, 2007), but activation in the right homologue of the damaged area also is common (e.g., Crinion & Price, 2005). Additionally, fMRI research has shown that activation patterns change over time as a function of natural recovery as well as in response to behavioral treatment (e.g., Saur et al., 2006). In fact, the extent of neural plasticity demonstrated by functional imaging research has been somewhat surprising. The use of fMRI and other functional measures of brain activity to characterize the neural changes associated with intervention is a focus of research in a number of aphasia research labs. An interesting observation regarding this emerging literature is that the focus of research questions has shifted from whether or not behavioral treatments can affect change in individuals with aphasia to what regions of the brain support the changes brought about by behavioral intervention.
Other Dynamic Measures of Brain Function
A variety of other approaches are available that provide insight regarding brain function, and additional measures are rapidly developing. One technique involves the imaging of blood perfusion during the acute stages of stroke, which tracks brain changes during the initial stages of recovery (e.g., Hillis, 2007). Positron emission tomography (PET) provides another means to examine regional cerebral blood flow. PET has the advantage of fewer constraints on movement than fMRI (i.e., movement artifact does not interfere with PET imaging so, for example, connected speech tasks can be reliably imaged), but also the disadvantage of requiring the injection or inhalation of a radioactive agent.
In contrast to approaches based on blood flow are a number of techniques that measure electrophysiological indicators of cognitive activity. Magnetoencephalography (MEG) records the rapid changes in magnetic fields in the brain that reflect neural activity. With special equipment, some labs combine MEG with fMRI so that the resulting data include the high-temporal resolution of MEG and the high-spatial resolution of fMRI. There are also ways of recording brain waves (EEGs) from the scalp during language tasks using high-density arrays (head caps with many electrodes). These combined approaches hold considerable potential to provide a more comprehensive understanding of brain activity during language-processing and cognitive tasks.
Approaches to the study of the neural substrates of language have advanced dramatically during the last three decades. This research is accomplished necessarily through multidisciplinary teams representing, for example, neurology, neuroscience, cognitive psychology, neuropsychology, and biomedical engineering, as well as speech-language pathology and audiology. Our discipline contributes substantially to the theoretical motivation driving many of these multidisciplinary efforts, especially those designed to examine the neural correlates of the comprehension and production of spoken language and written language. Recent advances using neuroimaging techniques provide exciting opportunities to better understand the neural basis of intervention approaches that contribute to the long-term recovery and rehabilitation of our patients.