Biomedical Sciences and Visualization

Human Brain Through Multimodel Imaging

Spatial integration for identifying the source of electroencephalogram (EEG) data provides an important tool in neurology. The inverse solution provides a good method for determining the source of specific waveforms. Most methods for determining the dipole source rely on models that assume that the head and brain are perfect spheres. Although these methods provide tractable solutions computationally, they also present many inadequacies.

This research is a collaborative effort with the Alcohol and Drug Abuse Research Center of McLean Hospital at Harvard Medical School. The specific aim of this project is to integrate precise neuroanatomincal data obtained from magnetic resonance imaging with functional information obtained from EEG. Studies to date have included the location of P300 waveforms and the source of sleep spindles and K complexes.

Dipole source estimation, providing a location, direction, and magnitude, is determined using the "inverse" solution method available with commercial software (Biologic). After EEG sensors are removed, fat soluable vitamin E capsules are affixed to the electrode site using collodion. These fiduciaries will be used to correlate the functional and structural data sets.

P300 P300
Figure 2: Example image showing P300 waveform dipole source
(right) Control. (left) After the induction of Ethanol.

Magnetic resonance data acquisition is performed on a GE Sigma 1.5 Tesla Whole Body Imager with head coil. Consecutive axial images were obtained with T1 and T2 weighting (using multiple-echo sequences on the latter) with the following settting: TR/TE/flip angle = 33ms/min=5ms/400; FOV = 22 cm, image resolution of 256X256 with an in-plane voxel dimension of 0.86 X 0.86 mm; slice thickness = 1.7 mm for a total of 124 slices; scan time = 9 min. Both MRI and EEG data sets are then transferred electronically to the Ohio Supercomputer Center via the Internet by using the file transfer protocol, ftp.

A common reference frame was developed using the voxel locations most representative of the nasion/inion and preauricular points, (FPZ,OZ, T3 and T4). An orthonormal reference frame based midway between the closest approach of FPZ/OZ and T3/T4 was constructed. The Y-axis (superior/inferior) of the reference frame is perpendicular to both FPZ/OZ and T3/T4, and the X and Z axes are approximately parallel to FPZ/OZ and T3/T4, respectively (see Figure 1). Finally, the required transformation was derived by taking the inverse of the transformation that converts the voxel reference frame to the dipole reference frame. All calculations and visualizations were performed on a Silicon Graphics INDIGO. The visualizations provide direct visualization of both the structural and functional information. In the P300 study, they suggest that the hippocampus is the most likely source of the P300 (see Figure 2). In the sleep study, they suggest that sleep spindles are generated from the thalmus, while K-complexes arise from much deeper nuclei, most likely the pons (see Figure 3).

negative-K negative-K
Figure 3: Examples from sleep study. (left) Two negative-K complexes. (right) Negative Spindle. (mpeg files)

This research is supported by grants DA03994 and DA00115 from the National Institute on Drug Abuse. Additional thanks to Dr. Scott E. Lukas, Chief of Neuropsychopharmacology at the Alcohol and Drug Abuse Research Center, McLean Hospital, Harvard Medical School.

Further Reading
Integration of P300 Evoked Potentials with Magnetic Resonance Images (MRI) to Identify Dipole Sources in Human Brain
Lukas SE, Sholar M, Stredney DL, Torello MW, May SF, and Sheepers F, Society of Neuroscience, November 7-12, 1993, Washington, D.C.

Apparent Source of EEG Sleep Spindles and K-Complexes: Correlations with Anatomical Sites Identified Using Magnetic Resonance Imaging (MRI)
Lukas SE, Sholar MB, Stredney DL, Torello MW, May SF, and Sheepers F, Sleep Research Society, American Sleep Disorders Association APSS 8th Annual Meeting, June 4-9, 1994, Boston, Massachusetts