Independent Component Analysis Separates Spikes of Different Origin in the EEG

Low frequency mu-like activity characterizes cortical rhythms in epilepsy due to ring chromosome 20

OBJECTIVES

To evaluate the spectral and spatial features of the cortical rhythms in patients affected by ring chromosome 20 - [r(20)]-syndrome.

METHODS

Twelve patients with [r(20)] syndrome were studied. As controls we enrolled 12 patients with idiopathic generalized epilepsy (IGE) and 12 healthy volunteers (HV). Blind source separation, spectral analyses and source reconstruction were applied in all cases in order to identify reliable spatio-temporal patterns of cortical activity.

RESULTS

A theta-delta EEG rhythm was identified in [r(20)] patients, with spectral peak ranging between 3 and 7Hz and whose generators mapped over the sensory-motor cortices. A second peak laying at a frequency about double with respect to the first one was present in 6 cases. Analogue methodological approach in HV and IGE groups failed to show similar findings.

CONCLUSIONS

EEG of [r(20)] patients reveals the existence of a highly reproducible EEG pattern arising from the sensory-motor system.

SIGNIFICANCE

The recognition of this peculiar EEG pattern could help the diagnostic work-up. Additionally, our findings supports the existence of a parallelism between this EEG trait and the physiological "mu" rhythm which is generate by the sensory-motor system. Such link suggests a sensory-motor system dysfunction in [r(20)] patients.

Transfer Function between EEG and BOLD Signals of Epileptic Activity

Simultaneous electroencephalogram (EEG)-functional Magnetic Resonance Imaging (fMRI) recordings have seen growing application in the evaluation of epilepsy, namely in the characterization of brain networks related to epileptic activity. In EEG-correlated fMRI studies, epileptic events are usually described as boxcar signals based on the timing information retrieved from the EEG, and subsequently convolved with a hemodynamic response function to model the associated Blood Oxygen Level Dependent (BOLD) changes. Although more flexible approaches may allow a higher degree of complexity for the hemodynamics, the issue of how to model these dynamics based on the EEG remains an open question. In this work, a new methodology for the integration of simultaneous EEG-fMRI data in epilepsy is proposed, which incorporates a transfer function from the EEG to the BOLD signal. Independent component analysis of the EEG is performed, and a number of metrics expressing different models of the EEG-BOLD transfer function are extracted from the resulting time courses. These metrics are then used to predict the fMRI data and to identify brain areas associated with the EEG epileptic activity. The methodology was tested on both ictal and interictal EEG-fMRI recordings from one patient with a hypothalamic hamartoma. When compared to the conventional analysis approach, plausible, consistent, and more significant activations were obtained. Importantly, frequency-weighted EEG metrics yielded superior results than those weighted solely on the EEG power, which comes in agreement with previous literature. Reproducibility, specificity, and sensitivity should be addressed in an extended group of patients in order to further validate the proposed methodology and generalize the presented proof of concept.

Localizing seizure-onset zones in presurgical evaluation of drug-resistant epilepsy by electroencephalography/fMRI: effectiveness of alternative thresholding strategies

BACKGROUND AND PURPOSE

Simultaneous EEG/fMRI is an effective noninvasive tool for identifying and localizing the SOZ in patients with focal epilepsy. In this study, we evaluated different thresholding strategies in EEG/fMRI for the assessment of hemodynamic responses to IEDs in the SOZ of drug-resistant epilepsy.

MATERIALS AND METHODS

Sixteen patients with focal epilepsy were examined by using simultaneous 92-channel EEG and BOLD fMRI. The temporal fluctuation of epileptiform signals on the EEG was extracted by independent component analysis to predict the hemodynamic responses to the IEDs. We applied 3 different threshold criteria to detect hemodynamic responses within the SOZ: 1) PA, 2) a fixed threshold at P < .05 corrected for multiple comparison (FWE), and 3) FAV (4000 ± 200 activated voxels within the brain).

RESULTS

PA identified the SOZ in 9 of 16 patients; FWE resulted in concordant BOLD signal correlates in 11 of 16, and FAV in 13 of 16 patients. Hemodynamic responses were detected within the resected areas in 5 (PA), 6 (FWE), and 8 (FAV) of 10 patients who remained seizure-free after surgery.

CONCLUSIONS

EEG/fMRI is a noninvasive tool for the presurgical work-up of patients with epilepsy, which can be performed during seizure-free periods and is complementary to the ictal electroclinical assessment. Our findings suggest that the effectiveness of EEG/fMRI in delineating the SOZ may be further improved by the additional use of alternative analysis strategies such as FAV.

Imaging foci of epileptic discharges from simultaneous EEG and fMRI using the canonical HRF

PURPOSE

Simultaneous electroencephalography and functional magnetic resonance imaging (EEG-fMRI) is considered as a powerful and non-invasive method that allows definition of the irritative zone. However, the complex interictal epileptic discharge (IED) may be present in some patients, and sometimes no active foci can be localized using General Linear Model (GLM) which is a widely adopted tool in EEG-fMRI study. The purpose of this study is to develop a new scheme to improve the detectability and localize the canonical HRF localizable foci.

METHOD

Various IEDs are classified using a combination of an independent component analysis (ICA) and a temporal correlation analysis between the independent components and the raw EEG channel; and the classified IEDs are then separately used for foci localization. This scheme is tested by ten patients with variable IEDs, including two patients whose activity could not be identified by common method.

RESULT

Applying this scheme to the two patients, some foci consistent with electroclinical data were localized. When it was applied to the remaining eight patients with positive results using common method, 2-4 types of IEDs were classified, and the activity could be identified from at least one type of IED. The results were similar to that received from common method.

CONCLUSION

These results indicate that the proposed scheme could enhance the imaging of the localizable foci by isolating its IEDs. This scheme is potentially a useful tool for epilepsy clinic.

Automatic EEG spike detection

Since the 1970s advances in science and technology during each succeeding decade have renewed the expectation of efficient, reliable automatic epileptiform spike detection (AESD). But even when reinforced with better, faster tools, clinically reliable unsupervised spike detection remains beyond our reach. Expert-selected spike parameters were the first and still most widely used for AESD. Thresholds for amplitude, duration, sharpness, rise-time, fall-time, after-coming slow waves, background frequency, and more have been used. It is still unclear which of these wave parameters are essential, beyond peak-peak amplitude and duration. Wavelet parameters are very appropriate to AESD but need to be combined with other parameters to achieve desired levels of spike detection efficiency. Artificial Neural Network (ANN) and expert-system methods may have reached peak efficiency. Support Vector Machine (SVM) technology focuses on outliers rather than centroids of spike and nonspike data clusters and should improve AESD efficiency. An exemplary spike/nonspike database is suggested as a tool for assessing parameters and methods for AESD and is available in CSV or Matlab formats from the author at brainvue@gmail.com. Exploratory Data Analysis (EDA) is presented as a graphic method for finding better spike parameters and for the step-wise evaluation of the spike detection process.

ICA decomposition of EEG signal for fMRI processing in epilepsy

In this study, we introduce a new approach to process simultaneous Electroencephalography and functional Magnetic Resonance Imaging (EEG-fMRI) data in epilepsy. The method is based on the decomposition of the EEG signal using independent component analysis (ICA) and the usage of the relevant components' time courses to define the event related model necessary to find the regions exhibiting fMRI signal changes related to interictal activity. This approach achieves a natural data-driven differentiation of the role of distinct types of interictal activity with different amplitudes and durations in the epileptogenic process. Agreement between the conventional method and this new approach was obtained in 6 out of 9 patients that had interictal activity inside the scanner. In all cases, the maximum Z-score was greater in the fMRI studies based on ICA component method and the extent of activation was increased in 5 out of the 6 cases in which overlap was found. Furthermore, the three cases where an agreement was not found were those in which no significant activation was found at all using the conventional approach.

Source localization of temporal lobe epilepsy using PCA-LORETA analysis on ictal EEG recordings

Localizing the source of an epileptic seizure using noninvasive EEG suffers from inaccuracies produced by other generators not related to the epileptic source. The authors isolated the ictal epileptic activity, and applied a source localization algorithm to identify its estimated location. Ten ictal EEG scalp recordings from five different patients were analyzed. The patients were known to have temporal lobe epilepsy with a single epileptic focus that had a concordant MRI lesion. The patients had become seizure-free following partial temporal lobectomy. A midinterval (approximately 5 seconds) period of ictal activity was used for Principal Component Analysis starting at ictal onset. The level of epileptic activity at each electrode (i.e., the eigenvector of the component that manifest epileptic characteristic), was used as an input for low-resolution tomography analysis for EEG inverse solution (Zilberstain et al., 2004). The algorithm accurately and robustly identified the epileptic focus in these patients. Principal component analysis and source localization methods can be used in the future to monitor the progression of an epileptic seizure and its expansion to other areas.

Independent components of sleep spindles

Sleep spindles are considered a hallmark of stage 2 of the sleep electroencephalogram (EEG) and are used both for sleep staging and for clinical studies of pharmacological agents. Analyses of sleep spindle topography, as well as intracranial source investigations provided evidence for the existence of two distinct sleep spindle types, "slow" and "fast" spindles at approximately 12 and 14 Hz, respectively. The aim of the present study was to apply Independent Component Analysis (ICA) to sleep spindles, for examining the possibility of extracting, through visual analysis of the spindle EEG and selection of Independent Components (ICs), spindle "components" corresponding to separate EEG activity patterns, and to investigate the sources underlying these spindle components. The inverse electromagnetic problem was solved using Low-Resolution Brain Electromagnetic Tomography (LORETA). Results indicate separability and stability of sources related to sleep spindle components reconstructed from separate groups of ICs.

Independent Component Analysis in the Study of Focal Seizures

Independent component analysis (ICA) is a novel technique that can separate statistically independent elements from complex signals. It has demonstrated its utility in separating artifacts and analyzing interictal discharges in EEG. ICA has been used recently in ictal recordings, showing the possibility of isolating the ictal activity. The goal of our study was to analyze focal seizures with ICA, decomposing the elements of the seizures to understand their genesis and propagation, and to differentiate between various types of focal seizures. We studied 26 focal seizures of temporal, frontal, or parietal origin. Only seizures with suspected focal onset were included in the study. The EEG recordings were acquired by using standard video-EEG equipment, with scalp electrodes. All the off-line analysis was carried out on a PC by means of specific software developed in the Matlab environment. ICA components were calculated with the use of the JADE (Joint Approximate Diagonalization of Eigen-matrices) algorithm. The decomposition of the seizures varied according to the EEG seizure pattern. In the seizures with focal rhythmic theta slow or sharp waves, the rhythmic activity was separated into one to five components, having an initial component with a clear concordance with the focus, whereas the others had an onset a few milliseconds later and corresponded to neighboring areas. In the 6 frontal seizures with regional rhythmic low voltage fast activity, 4 to 10 components were found, practically with a simultaneous timing, having a frontal distribution. In the three frontal seizures with a diffuse attenuation of the EEG signal, it was not possible to differentiate components of cerebral origin from the components of muscle artifact. ICA is an interesting tool to study the nature of focal seizures. The results depend on the EEG pattern. In the seizures with a clear EEG focal pattern, ICA may be useful to separate components of the ictal onset from the propagated activity.