- Email: [email protected]
Human head measurement with a best-fit three-sphere model
Society proceedings
18P
erators (SADG). Polygraphic EEG features associated with SADG can be reliably detected in long serial bipolar chain derivations in which epileptiform discharges are stereotyped between polygraphic channels, while with difficulty in reference derivations. The author has evaluated polygraphic EEG criteria that can be applied to standard EEG laboratory recordings. In this presentation, a variety of SADG features will be demonstrated among periodic discharges (focal, lateralized or diffuse) selected from a total of 129 cases between 1990 and 1994. SADG features were unequivocal in 2 cases of herpes simplex encephalitis and 3 cases of diffuse encephalopathy secondary to global ischemic/anoxic brain insults. Many equivocal cases were encountered, including a small number of cases revealing triphasic waves of hepatic coma. SADG features could be reliably defined, when periodic complexes were longer in duration and the background was attenuated. Polygraphic tracing, when limited in spatial coverage, often failed to resolve phase relationship between periodic discharges of shorter duration. This technique of spatiotemporal analysis is refinement of methods previously described and can expand neurophysiological correlation in a variety of clinical settings. Elll:
Early surface positives on electroencephalography in human epileptiform activity - S. Koszerls4, S. Ali1,4, S.L. Moshe1,21314 and E. Goldensohn1s4 (IDepartments of Neurology, 2Pediatrics, 3Neuroscience, 4Comprehensive Epilepsy Management Center at Montefiore Medical Center and The Albert Einstein College of Medicine, Bronx, NY, USA)
The largest component of epileptiform spike potentials recorded by electroencephalography (EEG) are most often surface negative sharply contoured waves. Spike localization has traditionally been determined based on this surface negativity. A lower amplitude surface positive potential often precedes this major deflection, however, its frequency of occurrence, field characteristics and clinical significance are unknown. Using computerized surface voltage mapping techniques, we analyzed the early and late surface potential fields of epileptiform spike activity in 70 patients. We found that surface positivities preceded a higher voltage surface negativity in more than 40 percent of the patients examined. Localization of these positivities was similar to the large negative spike component but not identical. These findings suggest that early surface positivities occur frequently with epileptiform activity and may provide localization information that is clinically useful. E112:
The evolution of cortical surface potential fields during an epileptic spike - *J. Scott Semel, *Robert D. Sidman, and **Terrence D. Lagerlund (*Department of Mathematics, University of Southwestern Louisiana, Lafayette, LA 70504, **Department of Neurology, Mayo Clinic, Rochester, MN 55905)
Two referentially recorded, 3 l-channel, 2-second EEG epochs (in the ASTM El467 format) from a patient with temporal lobe epilepsy were analyzed using a mathematical technique for simulating cortical surface potential fields. One epoch contained a spike and the other contained part of a seizure. These were localized by conventional EEG to the right temporal lobe. In each of these cases conventional scalp topographical maps were compared with theoretical cortical surface potential maps for time points throughout the epoch. The theoretical cortical landscapes are much richer in detail than the corresponding conventional scalp topographies and are more effective for elucidating the underlying generators of the scalp-recorded potentials. E113: Precision of dipole localization: a comparison of EEG, MEG, and scalp current density methods - *-**Anthony M. Murro, *Joseph R. Smith, *Don King, *Yang D. Park (**VA Hospital and *Medical College of Georgia, Augusta, GA 30912, USA) We used a multi-concentric
spherical
volume conduction
computer
model to estimate the relative precision of dipole localization from referential EEG (R-EEG). MEG, and scalp current density (SCD) methods. Dipole location relative to recording sites, dipole depth, and number of recording sites significantly influenced precision of dipole localization. The model predicted that an increase from 21 traditional IO/20 recording sites to 7 1 modified IO/20 recording sites would approximately double the precision of dipole localization. Assuming an equal signal to noise factor for each recording method, SCD and MEG methods were predicted to have a similar precision but both methods were significantly better than R-EEG. For localization of multiple dipole sources, MEG was predicted to significantly improve localization over SCD or R-EEG alone. These results suggest that SCD and MEG could provide more precise dipole localization than R-EEG but MEG would have a special role for localization of multiple dipole sources. Supported by Parke Davis, Ciba-Geigy. E114: Human head conductivity measurement with a best-tit threesphere model - *‘**K. Jeffrey Eriksen and **s***Ramesh Srinivasan (*EEG Dept., Legacy Good Samaritan Hospital, Portland, OR 97210, **Electrical Geodesics, Inc., Eugene, OR 97403, ***Brain Electrophysiology Lab, Univ. Oregon, Eugene. OR 97403) The accurate estimation of electrical conductivity of the constituent tissues of the head is important for proper modeling of the EEG, and perhaps the MEG as well. We have designed a method for estimating the conductivity of the scalp, skull, CSF, and brain in vivo. The subject’s MRI is segmented into these regions and then a 4-sphere model is fit to the boundaries between regions. A 128-channel electrode net is applied to the subject, and a IO microamp, 50 Hz sinusoidal current is injected across a pair of electrodes for 200 milliseconds while recording the induced potential field on the remaining 126 electrodes (referenced to Cz, later average referenced in software). This process is then repeated for 1024 different pairs chosen randomly out of the 8128 possible. The 50 Hz component of the induced EEG is extracted and used as the input to a nonlinear optimization routine that minimizes the fit between model and measured induced fields. The result of this process is the recovery of the four conductivities representing scalp, skull, CSF, and brain. We have obtained high resolution MRls (256 x 2.56 x 128) and conductivity data on five subjects, whose conductivity estimates we will present. Drs. Eriksen and Srinivasan are employees of Electrical Geodesics, Inc. E115: Spatiotemporal analysis of EEG data using the KarhunenLoeve decomposition - *Andrew D. Krystal, **Henry Greenside, and *Richard D. Weiner (*Department of Psychiatry, Duke University Medical Center, Durham, NC 27710 and **Department of Computer Science, Duke University, Durham, NC 27708) The Karhunen-Loeve (KL) decomposition of spatially extended signals constructs an ordered orthogonal spatial basis which can be used to identify the dominant spatial contributions to the variance of the signals and as a result may be useful for analyzing the complex spatiotemporal patterns of multichannel EEG data. We summarize recent calculations of the KL decomposition of 21-channel EEG data recorded from eight subjects during the waking eyes-closed state and during electroconvulsive therapy seizures. Rather surprisingly, a single KL mode was found to dominate in all 16 EEG’s studied, representing an average of 77% of the total variance for the seizure data and 55% for the waking EEG data. The greater dominance of this mode during the ECT seizures suggests greater spatial coherence of the activity during the seizures, which was verified by studying the correlation matrices of these two states. Ninety percent of the variance in the EEG signals was accounted for by the first two KL modes during the seizures and by the
18P
erators (SADG). Polygraphic EEG features associated with SADG can be reliably detected in long serial bipolar chain derivations in which epileptiform discharges are stereotyped between polygraphic channels, while with difficulty in reference derivations. The author has evaluated polygraphic EEG criteria that can be applied to standard EEG laboratory recordings. In this presentation, a variety of SADG features will be demonstrated among periodic discharges (focal, lateralized or diffuse) selected from a total of 129 cases between 1990 and 1994. SADG features were unequivocal in 2 cases of herpes simplex encephalitis and 3 cases of diffuse encephalopathy secondary to global ischemic/anoxic brain insults. Many equivocal cases were encountered, including a small number of cases revealing triphasic waves of hepatic coma. SADG features could be reliably defined, when periodic complexes were longer in duration and the background was attenuated. Polygraphic tracing, when limited in spatial coverage, often failed to resolve phase relationship between periodic discharges of shorter duration. This technique of spatiotemporal analysis is refinement of methods previously described and can expand neurophysiological correlation in a variety of clinical settings. Elll:
Early surface positives on electroencephalography in human epileptiform activity - S. Koszerls4, S. Ali1,4, S.L. Moshe1,21314 and E. Goldensohn1s4 (IDepartments of Neurology, 2Pediatrics, 3Neuroscience, 4Comprehensive Epilepsy Management Center at Montefiore Medical Center and The Albert Einstein College of Medicine, Bronx, NY, USA)
The largest component of epileptiform spike potentials recorded by electroencephalography (EEG) are most often surface negative sharply contoured waves. Spike localization has traditionally been determined based on this surface negativity. A lower amplitude surface positive potential often precedes this major deflection, however, its frequency of occurrence, field characteristics and clinical significance are unknown. Using computerized surface voltage mapping techniques, we analyzed the early and late surface potential fields of epileptiform spike activity in 70 patients. We found that surface positivities preceded a higher voltage surface negativity in more than 40 percent of the patients examined. Localization of these positivities was similar to the large negative spike component but not identical. These findings suggest that early surface positivities occur frequently with epileptiform activity and may provide localization information that is clinically useful. E112:
The evolution of cortical surface potential fields during an epileptic spike - *J. Scott Semel, *Robert D. Sidman, and **Terrence D. Lagerlund (*Department of Mathematics, University of Southwestern Louisiana, Lafayette, LA 70504, **Department of Neurology, Mayo Clinic, Rochester, MN 55905)
Two referentially recorded, 3 l-channel, 2-second EEG epochs (in the ASTM El467 format) from a patient with temporal lobe epilepsy were analyzed using a mathematical technique for simulating cortical surface potential fields. One epoch contained a spike and the other contained part of a seizure. These were localized by conventional EEG to the right temporal lobe. In each of these cases conventional scalp topographical maps were compared with theoretical cortical surface potential maps for time points throughout the epoch. The theoretical cortical landscapes are much richer in detail than the corresponding conventional scalp topographies and are more effective for elucidating the underlying generators of the scalp-recorded potentials. E113: Precision of dipole localization: a comparison of EEG, MEG, and scalp current density methods - *-**Anthony M. Murro, *Joseph R. Smith, *Don King, *Yang D. Park (**VA Hospital and *Medical College of Georgia, Augusta, GA 30912, USA) We used a multi-concentric
spherical
volume conduction
computer
model to estimate the relative precision of dipole localization from referential EEG (R-EEG). MEG, and scalp current density (SCD) methods. Dipole location relative to recording sites, dipole depth, and number of recording sites significantly influenced precision of dipole localization. The model predicted that an increase from 21 traditional IO/20 recording sites to 7 1 modified IO/20 recording sites would approximately double the precision of dipole localization. Assuming an equal signal to noise factor for each recording method, SCD and MEG methods were predicted to have a similar precision but both methods were significantly better than R-EEG. For localization of multiple dipole sources, MEG was predicted to significantly improve localization over SCD or R-EEG alone. These results suggest that SCD and MEG could provide more precise dipole localization than R-EEG but MEG would have a special role for localization of multiple dipole sources. Supported by Parke Davis, Ciba-Geigy. E114: Human head conductivity measurement with a best-tit threesphere model - *‘**K. Jeffrey Eriksen and **s***Ramesh Srinivasan (*EEG Dept., Legacy Good Samaritan Hospital, Portland, OR 97210, **Electrical Geodesics, Inc., Eugene, OR 97403, ***Brain Electrophysiology Lab, Univ. Oregon, Eugene. OR 97403) The accurate estimation of electrical conductivity of the constituent tissues of the head is important for proper modeling of the EEG, and perhaps the MEG as well. We have designed a method for estimating the conductivity of the scalp, skull, CSF, and brain in vivo. The subject’s MRI is segmented into these regions and then a 4-sphere model is fit to the boundaries between regions. A 128-channel electrode net is applied to the subject, and a IO microamp, 50 Hz sinusoidal current is injected across a pair of electrodes for 200 milliseconds while recording the induced potential field on the remaining 126 electrodes (referenced to Cz, later average referenced in software). This process is then repeated for 1024 different pairs chosen randomly out of the 8128 possible. The 50 Hz component of the induced EEG is extracted and used as the input to a nonlinear optimization routine that minimizes the fit between model and measured induced fields. The result of this process is the recovery of the four conductivities representing scalp, skull, CSF, and brain. We have obtained high resolution MRls (256 x 2.56 x 128) and conductivity data on five subjects, whose conductivity estimates we will present. Drs. Eriksen and Srinivasan are employees of Electrical Geodesics, Inc. E115: Spatiotemporal analysis of EEG data using the KarhunenLoeve decomposition - *Andrew D. Krystal, **Henry Greenside, and *Richard D. Weiner (*Department of Psychiatry, Duke University Medical Center, Durham, NC 27710 and **Department of Computer Science, Duke University, Durham, NC 27708) The Karhunen-Loeve (KL) decomposition of spatially extended signals constructs an ordered orthogonal spatial basis which can be used to identify the dominant spatial contributions to the variance of the signals and as a result may be useful for analyzing the complex spatiotemporal patterns of multichannel EEG data. We summarize recent calculations of the KL decomposition of 21-channel EEG data recorded from eight subjects during the waking eyes-closed state and during electroconvulsive therapy seizures. Rather surprisingly, a single KL mode was found to dominate in all 16 EEG’s studied, representing an average of 77% of the total variance for the seizure data and 55% for the waking EEG data. The greater dominance of this mode during the ECT seizures suggests greater spatial coherence of the activity during the seizures, which was verified by studying the correlation matrices of these two states. Ninety percent of the variance in the EEG signals was accounted for by the first two KL modes during the seizures and by the