back to Monell








Depending on the question we are trying to answer, we use one of three different brain imaging methods: functional magnetic resonance imaging (fMRI), positron emission tomography (PET), or event related potentials (ERP). Each of these methods is based on the principle that we are able to correlate external events with underlying neuronal activity in a quantitative manner. This is accomplished either indirectly by assessing blood flow (increased neuronal activation means an increase in the need of glucose and oxygen to that area, hence an increase in blood flow needed to transport these nutrients) or directly by recording the small amount of energy emitted from neuronal firing. Each of these techniques provides us with different types of information.

Functional Magnetic Resonance Imaging (fMRI)
Figure: MRI Camera

Magnetic resonance imaging is a non-invasive method of brain imaging utilizing a strong magnetic field (usually a magnetic field of 3 Tesla strength) and several radio frequency pulses. It can be used to either study brain anatomy or brain function. If one wants to investigate brain function, the most common approach is to measure task-related changes in the local hemodynamic activity within each region. Task-related neural activation is accompanied by a local increase in cerebral blood flow (CBF) that is disproportionate relative to changes in oxygen consumption.

As its name suggests, the blood oxygenation-level dependent (BOLD) signal is based on our ability to detect changes in the bloods oxygen concentration that is attached to hemoglobin (oxyhemoglobin). Deoxyhemoglobin has other magnetic properties compared to oxyhemoglobin. It is paramagnetic and, as such, it acts as an endogenous contrast agent that creates microscopic magnetic-field differences in the measured brain areas. These local differences increase the rate of decay of transverse magnetization and lead to a reduction in the local MR signal. Upon neural activation, the disproportionate increase in local blood flow “dilutes” the amount of deoxyhemoglobin. More oxyhemoglobin enters the activated brain area resulting in a local increase of the oxy-/deoxyhemoglobin ratio in this area. This in turn, reduces the amount of local magnetic field inhomogeneities and leads to a higher BOLD signal in the MR images. Differences between conditions, usually between “task” and “no task” or “high task” and “low task”, are then calculated using a large variety of statistical methods with the end results often being so-called statistical parametric maps (SPM). In these maps the statistical differences in brain activation between two conditions are color coded. In other words, the colorful images produced by these types of experiments are not a direct reflection of blood flow but rather a colorful representation of how large the statistical difference is in amount of magnetic properties between two tasks (conditions).

ositron Emission Tomography (PET)
Figure: PET Camera

The underlying idea of PET imaging is rather similar to what is described above for fMRI; differences in cerebral blood flow (CBF) are measured. However, instead of measuring difference in magnetic propertied of used vs unused blood, PET imaging typically measure CBF by tracking the regional cerebral distribution of a positron-emitting tracer with a short half-life. In a common blood-flow activation study, a small amount of the positron-emitting 15O-water is injected into a vein and the distribution of this tracer is measured over a period of 60 sec by PET detectors (marked brown in left figure).

As it decays, the tracer emits positrons that annihilate by interacting with electrons. During this process, two gamma rays are emitted. The gamma rays leave the annihilation area in exactly opposite directions. Using a coincidence circuit, a pair of detectors records the gamma rays emitted along the line passing through the site of positron-electron interaction. Tomographic measurements of the number and location of positron-electron interactions are carried out for the entire brain allowing us to map the pattern of changes in regional CBF. These signals are then anatomically located by mapping them onto an anatomical image of the
PET detector
Figure: PET detectors
subject’s brain, which has been acquired utilizing an MR scanner. The functional information (CBF values) is then calculated in a similar fashion as the data obtained from fMRI. One of the main benefits of PET imaging is that one can image the whole brain at the same time with equal good signal to noise propertied in all areas. This is not possible in fMRI. However, PET imaging is more invasive, much more expensive, and also provides poorer spatial and temporal resolution than fMRI.

vent Related Potentials (ERP) - Electrencephalography (EEG)

EEG is a technique which measures electrical potential differences across the scalp that reflect the underlying neuronal activity of the brain. More specifically, EEG measures particular synaptic activity, excitatory and inhibitory postsynaptic potentials, of cortical pyramidal neurons. It is a non-invasive technique with a high temporal, but limited spatial resolution (when compared to fMRI or PET), and easy to employ in a clinical setting. When the brain processes a stimulus, two types of changes in the EEG may occur: Evoked activities, which are exactly time-locked to the stimulus and induced activities, which are changes in the EEG that are not phase-locked to the stimulus.

Event related potentials (ERP) are changes time locked to an event, i.e. the potential occurs either immediately before or after a defined stimulus. This stimulus can originate from either an internal or an external source. ERPs reflect the synchronous and phase locked activities of a large neuronal population engaged in information processing. In other words, the EEG represents spontaneous cortical activity whereas the ERP is generated as a response to specific stimuli and is averaged over a number of samples. However, ERPs can only be generated by neuronal areas that are organized in an open field. If they are organized in a closed field, like the neurons of the cerebellum or the hippocampus, no ERPs will be generated.

ERP curve
Figure: ERP curve
When analyzing the averaged, stimulus-locked EEG signal, a number of waveforms (‘peaks’) can be identified, and characterized by their polarity, order of occurrence, and origin. Components are labeled by their polarity (Positive or Negative) and relative order/time (1 or 100) such that the first positive deflection of an ERP curve is commonly labeled P1, or P100, and they are divided into exogenous or endogenous categories. The exogenous, or early, components of the ERP curve are more a reflection of the initial neural processing of the physical characteristics of a stimulus. These responses are automatic responses to the stimulus and, hence, the magnitudes of these exogenous components are not very dependent on the cognitive processing of the stimulus. The endogenous, or later, components are a more accurate reflection of the neural processing, or cognitive handling of a stimulus. These usually involve "higher" cognitive processes like attention or memory and are sensitive to changes in the meaning of a stimulus or changes in the information processing demand. However, this sensitivity should not be interpreted as a causal link between neurological substrates of cognitive processing and late ERP components, but rather as an indication of correlation between two processes resulting from the same stimulus.

One of the appeals of ERP recordings is the ability to relate specific cortical responses to discrete psychological or physiological states and events in a less expensive and less invasive manner than PET or fMRI. Moreover, whereas PET and fMRI techniques generate data that is on a temporal scale of seconds to minutes, ERP recordings are able to provide data on a millisecond scale, thereby allowing a separation, to some degree, of sensory effects from more cognitive effects. In addition to the qualitative characteristics of an ERP curve, information can also be collected from an analysis of its quantitative characteristics, such as peak amplitude, peak latency, peak-to-peak amplitude, and the area under the ERP curve.

We are using an ActiveTwo EEG system, consisting of active electrodes, and a flexible headcap. An active-electrode is a sensor with a very low output impedance; which eliminates all problems with regards to capacitive coupling between the cable and sources of interference, as well as any artifacts by cable and connector movements. By integrating the first amplifier stage with a sintered Ag-AgCl electrode, extremely low-noise measurements are possible without any skin preparation. This makes the application of these electrodes very fast, minimizes discomfort of the subject, and eliminates the risk of infection.