EEG Findings in Traumatic Brain Injury

This brief summary will discuss the various EEG findings seen in head injury when it results in a brain injury, though any given head injury may or may not result in traumatic brain injury.  When an injury is incurred by the brain there are a few varieties of findings seen in the EEG, ranging from spectral changes associated with either white or gray matter damage, to the changes in “connectivity”, seen as changes in coherence or correlation measured across the cortex, or between more distant functionally related areas.

Damage is seldom restricted to merely being exclusively either white or gray matter, and mixed findings are seen commonly.  There are studies showing the correlation of quantitative EEG findings with quantitative MRI findings that are instructive in identifying the nature of the effect on the EEG of the different types of damage.

The EEG changes following brain injury are spectrally different between white and gray matter damage, which helps when evaluating the nature of the damage with the EEG.  The white matter is a high speed relay system that innervates the cortex, both with primary sensory input relayed from the thalamus, and with cortical-cortical input via various fasciculi.

When the cortex has decreased innervation, delta content emerges, according to the IFCN’s position paper on the basic mechanisms of cerebral rhythmic EEG**.  Thus, traumatic brain injury resulting in white matter damage is associated with slower spectral increases in the areas cortically where decreased innervation is present.  These slow spectral increases are seen primarily as delta, and may also be seen as a slower band including theta, especially with larger increases in the slow spectra.

White matter also carries signals across the cortex, and from the cortex through subcortical structures to other cortical locations, resulting in the neural network’s “connectivity”.    There has been a small case series showing that in some direct frontal injuries, there is a decrease in correlation from the left to the right frontal lobe, seen as decreased spectral correlation, also referred to as co-modulation (M.B. Sterman and D. Kaiser’s SKIL software).  This is identical to the changes seen with damage to the anterior portions of the corpus callosum following surgery.  This data was presented by Dr. Sterman, and published by the Journal of Neurotherapy as a technical paper describing their co-modulation metric.

Coherence changes may also be seen with head injury, with both hypercoherence and Hypocoherence reported, depending on the nature of the specific case’s damage.  Isolated areas may become hypercoherent due to the lack of input, though separated areas will be hypocoherent due to the damage to their connective network.

Damage may be seen in gray matter, which is highly “plastic”, unlike white matter, where damage persists.  The neural plasticity allows for regeneration of the cortical gray matter following injury, so the spectral changes associated with gray matter damage may change over time, from the more acute stages, through a transition phase into a static phase, which may allow for re-integration into functional relationships with neural network activity.

The immediate changes seen spectrally with gray matter injury is a decrease in the function of the thalmo-cortical neural network activity, seen spectrally as decreased alpha and beta, as well as decreased gamma in the affected gray matter.  These changes last for the period of the healing, commonly seen across a period from 6 months to a year.

As the gray matter heals, but is not integrated into the neural network function, the idling rhythm in alpha may return and even be seen as an excessive value in database comparisons, since the cortical area is not “working”.  The beta and gamma remain low during this phase, since they are not seen at normal levels in the idled cortical areas.  Beta is generated in local gray matter network activity, and gamma is seen in functionally bound and active networks only.

Once the neural network of the local gray matter is re-integrated into the functional processing, the alpha will then be reduced, and the faster activity seen associated with local function will also be seen as returning to more normal levels.  This may not happen spontaneously, and may require specific interventions, such as neurofeedback, physical therapy, and/or various cognitive-behavioral interventions.

The work of Dr. Kirtley Thornton showed that the gamma and beta remain low, even when the alpha return has occurred.  These faster patterns returned following successful clinical therapy to re-integrate the neural tissue into the functional neural network of the cortical gray matter and white matter.

Some software provides multivariate discriminant analysis, differentiating normal controls from mild traumatic brain injured clients.  These were collected retrospectively, with clients in a specific state of the dynamically changing gray matter’s plasticity, within a 9 month range in one product that is commercially available.  Their prospective use clinically, like all other classification systems, provide false positive and false negative results (type 1 and type 2 errors).

When used indiscriminately, discriminants provide a significant “red herring” problem clinically.  They are not appropriately used as a screening test for individuals, but rather they are only appropriate when used to answer a specific clinical question: “Has my client who has had an actual head injury actually suffered a brain injury?”

I personally do not find them useful clinically, since they do not provide a full evaluation of a client’s brain’s specific injury, and have an unacceptable false negative rate in know head injured clients.  The dismissal of clinically significant findings by the relatively blind use of a head trauma discriminant would tell 20%-30% of those who have had a real brain injury that they are “normal”.  This is not acceptable in the real world when a better clinical judgment would be provided by a careful analysis of the EEG and qEEG by experts in this specific application area.  We have also found a 50% false positive rate when applied to a general clinical population (though this is not the intended use of the discriminant).

The neurological professional groups are divided on the use of traumatic brain injury discriminant classification, with the American Academy of Neurology (AAN) refusing to accept discriminant use clinically, but the 1994 position paper of AMEEGA, published in EEG and Clinical Neurophysiology, provides for the clinical acceptability of the technique in the hands of experts.  ECNS (EEG and Clinical Neuroscience Society)  has reiterated the AMEEGA position paper, and the AAN position paper has had specific responses to it from those who use discriminants.

I find the detailed evaluation of the client’s EEG and qEEG, and an understanding of the dynamics of the brain’s response to trauma, provide a superior working understanding of the client’s specific injury.  This is far superior compared with a simplistic sorting into classifications of “normal” or “TBI” by software that admittedly misses 20% of the actual cases of brain injury and 50% of other clinical cases would be classified positive for brain injury in the absence of any history of head injury in an open clinical series.

Therapeutic intervention is not specified by TBI discriminants, nor is it reasonably possible to customize a therapeutic approach using discriminants due to their sensitivity to artifact.  By contrast, the EEG and qEEG data can be used for both understanding the brain injury, as well as to help the clinician customize a therapeutic approach to the specific neural network areas injured traumatically.

**   M. Steriade, P. Gloor, R.R. Llinas, F.H. Lopes da Silva, and M.M. Mesulam (1990)
Report of IFCN Committee on Basic Mechanisms:  Basic mechanisms of Cerebral Rhythmic Activities,
Electroencephalography and Clinical Neurophysiology; 1990, 76: 481-508


Nuwer, M, et al, Routine and Quantitative EEG in Mild Traumatic Brain Injury; Clinical Neurophysiology, 116 (2005) 2001-2025

Thatcher, R.W., Camacho, M,, Salazar, A, Linden, C., Biver, C. and Clarke, L.: Quantitative MRI of Gray-White Matter Distribution in Traumatic Brain Injury. Journal of Neurotrauma, Volume 14, No. 1, 1-14, 1997

Thatcher, R.W., Moore, N, John, E.R., et al.: QEEG and Traumatic Brain Injury: Rebuttal of the American Academy of Neurology 1997. A Report by the EEG and Clinical Neuroscience Society, Clinical Electroencephalography, 30(3): 94-98, 1999

Leave a Comment