Article

Integrated Imaging Approach with MEG and DTI to Detect Mild Traumatic Brain Injury in Military and Civilian Patients

Research, Radiology, Rehabilitation, and Psychiatry Services, VA San Diego Healthcare System, San Diego, California, USA.
Journal of neurotrauma (Impact Factor: 3.71). 05/2009; 26(8):1213-26. DOI: 10.1089/neu.2008.0672
Source: PubMed

ABSTRACT

Traumatic brain injury (TBI) is a leading cause of sustained impairment in military and civilian populations. However, mild (and some moderate) TBI can be difficult to diagnose due to lack of obvious external injuries and because the injuries are often not visible on conventional acute MRI or CT. Injured brain tissues in TBI patients generate pathological low-frequency neuronal magnetic signal (delta waves 1-4 Hz) that can be measured and localized by magnetoencephalography (MEG). We hypothesize that abnormal MEG delta waves originate from gray matter neurons that experience de-afferentation due to axonal injury to the underlying white matter fiber tracts, which is manifested on diffusion tensor imaging (DTI) as reduced fractional anisotropy. The present study used a neuroimaging approach integrating findings of magnetoencephalography (MEG) and diffusion tensor imaging (DTI), evaluating their utility in diagnosing mild TBI in 10 subjects in whom conventional CT and MRI showed no visible lesions in 9. The results show: (1) the integrated approach with MEG and DTI is more sensitive than conventional CT and MRI in detecting subtle neuronal injury in mild TBI; (2) MEG slow waves in mild TBI patients originate from cortical gray matter areas that experience de-afferentation due to axonal injuries in the white matter fibers with reduced fractional anisotropy; (3) findings from the integrated imaging approach are consistent with post-concussive symptoms; (4) in some cases, abnormal MEG delta waves were observed in subjects without obvious DTI abnormality, indicating that MEG may be more sensitive than DTI in diagnosing mild TBI.

4 Followers
 · 
59 Reads
  • Source
    • "Numerous neuroimaging techniques have been used to investigate changes to brain integrity related to TBI including structural magnetic resonance imaging (MRI), functional MRI (fMRI), diffusion tensor imaging (DTI), magnetoencephalography, positron emission tomography, and macromolecular proton fraction (Huang et al., 2009, Bigler et al., 2010, Yurgelun-Todd et al., 2011, Lopez-Larson et al., 2013, Han et al., 2014, Petrie et al., 2014). Despite the many limitations inherent in studying TBI, progress has been made as emerging technologies allow for novel uses of neuroimaging data to study the effects of TBI. "
    [Show abstract] [Hide abstract]
    ABSTRACT: In the United States alone, the number of persons living with the enduring consequences of traumatic brain injuries is estimated to be between 3.2 and 5 million. This number does not include individuals serving in the United States military or seeking care at Veterans Affairs hospitals. The importance of understanding the neurobiological consequences of mild traumatic brain injury (mTBI) has increased with the return of veterans from conflicts overseas, many of who have suffered this type of brain injury. However, identifying the neuroanatomical regions most affected by mTBI continues to prove challenging. The aim of this study was to assess the use of mean cortical curvature as a potential indicator of progressive tissue loss in a cross-sectional sample of 54 veterans with mTBI compared to 31 controls evaluated with MRI. It was hypothesized that mean cortical curvature would be increased in veterans with mTBI, relative to controls, due in part to cortical restructuring related to tissue volume loss. Mean cortical curvature was assessed in 60 bilateral regions (31 sulcal, 29 gyral). Of the 120 regions investigated, nearly 50% demonstrated significantly increased mean cortical curvature in mTBI relative to controls with 25% remaining significant following multiple comparison correction (all, pFDR<.05). These differences were most prominent in deep gray matter regions of the cortex. Additionally, significant relationships were found between mean cortical curvature and gray and white matter volumes (all, p<.05). These findings suggest potentially unique patterns of atrophy by region and indicate that changes in brain microstructure due to mTBI are sensitive to measures of mean curvature.
    Full-text · Article · Jan 2016 · Clinical neuroimaging
  • Source
    • "But in other ways it is quite different: (1) Rather than using a single filter with unit gain for a target location, 1000+ differenced pairs of filters [2.D] are used for dummy (referee) locations, each with zero gain [2.C,1.A] at or very near the target location. (2) To decide if there is a source at a target location, a probabilistic measure of goodness of fit is constructed [2.E,2.F] from the output of 1000+ filter pairs for that target rather than using a measurement error metric [1] or a post-hoc test on the outputs of the filters for all of the targets [4] [6] [10]. (3) The time course of the activity at the target location results from a joint estimation procedure [2.G] applied to the output of all 1000+ filter pairs rather than from the output of a single filter. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Electrical current flow within populations of neurons is a fundamental constituent of brain function. The resulting fluctuating magnetic fields may be sampled noninvasively with an array of magnetic field detectors positioned outside a patient's head. This is magnetoencephalography (MEG). Each source may be characterized by 5-6 parameters, the xyz location and the xyz direction. The magnetic field measurements are nonlinear in the location parameters; hence the source location is identifiable only via search of the brain volume. When there is one or a very few sources, this may be practical; solutions for the general problem are weak. Referee consensus is a new method which enables identification of one source at a time regardless of the number and location of others. This "independence" enables solution of the general problem and insures suitability to grid computing. The computation scales linearly with the number of nonlinear parameters. MEG recordings were obtained from volunteers while they performed a cognitive task The recordings were processed on the Open Science Grid (≈150 CPU hours/sec of data). On average 500-1500 sources were active throughout. Statistical analyses demonstrated < 2 mm resolving power 1 and very strong findings (p < 0.02 400) when testing for task specific information in the extracted virtual recordings from each individual. 3D maps of differential activation, neuroelectric tomography, provide a very high resolution functional imaging modality which compares favorably with functional magnetic resonance imaging (fMRI). Referee consensus is applicable widely to measurement systems including microwave telescope imaging, seismic tomography, and financial market linkage identification. Applicability requires: (1) The measurements are linear in at least one parameter of each "source." (2) Each source is detectable at multiple sensors. (3) A sequence of measurements in time is available. 1 Linear dimensions are represented in this standard form. Volume dimensions are represented throughout in terms of the length of a side, e.g. 8mm 3 instead of ½ cc, ½ cm 3 or 512 mm 3 .
    Full-text · Article · Jan 2014 · International Journal of Foundations of Computer Science
  • Source
    • "But in other ways it is quite different: (1) Rather than using a single filter with unit gain for a target location, 1000+ differenced pairs of filters [2.D] are used for dummy (referee) locations, each with zero gain [2.C,1.A] at or very near the target location. (2) To decide if there is a source at a target location, a probabilistic measure of goodness of fit is constructed [2.E,2.F] from the output of 1000+ filter pairs for that target rather than using a measurement error metric [1] or a post-hoc test on the outputs of the filters for all of the targets [4] [6] [10]. (3) The time course of the activity at the target location results from a joint estimation procedure [2.G] applied to the output of all 1000+ filter pairs rather than from the output of a single filter. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Electrical current flow within populations of neurons is a fundamental constituent of brain function. The resulting fluctuating magnetic fields may be sampled noninvasively with an array of magnetic field detectors positioned outside a patient's head. This is magnetoencephalography (MEG). Each source may be characterized by 5-6 parameters, the xyz location and the xyz direction. The magnetic field measurements are nonlinear in the location parameters; hence the source location is identifiable only via search of the brain volume. When there is one or a very few sources, this may be practical; solutions for the general problem are weak. Referee consensus is a new method which enables identification of one source at a time regardless of the number and location of others. This "independence" enables solution of the general problem and insures suitability to grid computing. The computation scales linearly with the number of nonlinear parameters. MEG recordings were obtained from volunteers while they performed a cognitive task The recordings were processed on the Open Science Grid (≈150 CPU hours/sec of data). On average 500-1500 sources were active throughout. Statistical analyses demonstrated < 2 mm resolving power 1 and very strong findings (p < 0.02 400) when testing for task specific information in the extracted virtual recordings from each individual. 3D maps of differential activation, neuroelectric tomography, provide a very high resolution functional imaging modality which compares favorably with functional magnetic resonance imaging (fMRI). Referee consensus is applicable widely to measurement systems including microwave telescope imaging, seismic tomography, and financial market linkage identification. Applicability requires: (1) The measurements are linear in at least one parameter of each "source." (2) Each source is detectable at multiple sensors. (3) A sequence of measurements in time is available. 1 Linear dimensions are represented in this standard form. Volume dimensions are represented throughout in terms of the length of a side, e.g. 8mm 3 instead of ½ cc, ½ cm 3 or 512 mm 3 .
    Full-text · Article · Jan 2014 · International Journal of Foundations of Computer Science
Show more