4th Dutch Bio-Medical Engineering Conference 2013
24-25 January 2013, Egmond aan Zee, The Netherlands

Powered by
© Fyper VOF
Conference Websites
13:30   Neurophysiology: Clinical Neurophysiology
Chair: Carel Meskers
15 mins
Jan de Munck, Petra van Houdt, Albert Albert Colon, Frans Leijten, Geertjan Huiskamp, Paul Boon, Pauly Ossenblok
Abstract: Co-registration of EEG and functional MRI (EEG-fMRI) enables the detection of the brain regions involved with interictal epileptiform discharges (IEDs). Therefore, EEG-fMRI is a potential tool in the pre-operative work-up of epilepsy surgery. The goal of this study was to obtain more insight in the clinical value of EEG-fMRI. For those purposes, the EEG-fMRI results were systematically compared to the interictal invasive EEG recordings, to the seizure onset zone, and to the resection area. The EEG-fMRI data of 16 patients were included who were implanted with subdural grids. An IED-density function was extracted from the EEG indicating the number of IEDs occurring during each fMRI scan. This density function was correlated to the fMRI signals through a general linear model framework (GLM), yielding a correlation pattern of brain regions activated during IEDs, as well as an estimate of the associated hemodynamic response function (HRF). For the validation of these results, a quantitative approach was developed [1,2] revealing the spatiotemporal patterns of the IEDs present in the invasive EEG data. These spatial and temporal patterns were visualized at the same anatomical MRI as the EEG-fMRI data. Finally, the interictal EEG-fMRI results were related to the clinically more relevant seizure onset zone and resection area. In all patients, EEG-fMRI analysis yielded multiple activated BOLD regions of which at least one overlapped with the active electrodes in the invasive EEG data. In 11 of the 16 data sets, multiple EEG-fMRI areas were concordant with active ECoG electrodes. Interestingly, one of those concordant areas appeared to be related to the early onset of the IEDs, suggesting that the other concordant EEG-fMRI areas were related to propagation of the activity. No indications were found that the shape of the HRF was predictive for the role of the EEG-fMRI cluster within the interictal network. Finally, EEG-fMRI clusters included the complete seizure onset zone in 83% and the resection area in 90% of the data sets. This study shows that EEG-fMRI has substantial predictive value regarding both the onset zone and resection area and, therefore, it could play an important role during the planning of implantation strategy of the surgical candidates. REFERENCES [1] Van Houdt PJ, Ossenblok PPW, Colon AJ, Boon PAJM and De Munck JC, A framework to integrate EEG-correlated fMRI and intracerebral recordings, NeuroImage, 60:2042–2053, (2012). [2] Van Houdt PJ, De Munck JC, Leijten FSS, Huiskamp GJM, Colon AJ, Boon PAJM, Ossenblok PPW, EEG-correlated fMRI evaluated with electro-corticographical and clinical outcome measures as gold standards, NeuroImage, submitted, (2012).
15 mins
Debby Klooster, Bert Kleine, Martien van Bussel, Miriam Lammers, Machiel Zwarts
Abstract: Aim: In this project we investigate whether EMG recordings have added value in the diagnosis and classification of epilepsy with motor seizures. Motor seizures can be either myoclonias, tonic seizures, clonic seizures or tonic-clonic seizures. One of the goals is to differentiate these seizures from non-epileptic attacks and movement disorders. Methods: EMG was recorded in 34 patients referred to our epilepsy monitoring unit (EMU) for the differential diagnosis of their spells. In the EMU, EEG and video are continuously recorded for a period of one up to five days. Stellate equipment is used in combination with a 36 channel Lamont amplifier. Depending on the number of channels that were not in use for EEG, the EMG is recorded from the M. deltoideus bilaterally or from the M. deltoideus, the M. biceps brachii and the M. triceps brachii. Following the routine procedure, the patient or nurse can use a pushbutton to mark seizures or other events. After the recording, the EEG is read together with the video of potential seizures by an experienced EEG technician and then by a neurologist and clinical neurophysiologist. This way we created a database containing seizures types, and their EMG correlates in each patient. Results: Eight out of 34 patients had motor seizures during the recording. Multiple seizure types occurred in two patients. One patient showed multiple clonic seizures. One patient showed two clear tonic-clonic seizures. Tonic seizures were found in three patients and myoclonias in five patients. All major seizures were also detected by video. However, EMG improved the characterization of the onset. Additionally, epileptic myoclonus was frequently difficult to distinguish by video alone, and EMG was required to define negative epileptic myoclonus. Conclusion: The EMG can have added value in differentiating epileptic motor seizures from non-epileptic motor seizures. Furthermore, the EMG might provide additional information about the timing and the exact location of motor seizures. Future perspectives: We will develop methods for automatic seizure detection, based on EMG signals. An algorithm was created within Kempenhaeghe [1] to detect tonic seizures based on EMG signals in combination with accelerometry data. Also algorithms based on EMG alone were proposed by Conradsen et al. [2]. We plan to investigate which parameter(s) can distinguish epileptic motor activity from non-epileptic movements. Furthermore, we plan to find parameters that can help to automatically detect motor seizures. REFERENCES [1] R.A.P.P. Adriaans, Strategies for a real-time detection system of the tonic phase of nocturnal generalized tonic and tonic-clonic epileptic seizures. Master thesis report, Technical University of Eindhoven, 2012. [2] I. Conradsen, S. Beniczky, K. Hoppe, P. Wolf, H.B.D. Sorendsen, Automated algorithm for generalized tonic-clonic epileptic seizure onset detection based on sEMG zero-crossing rate. IEEE Biomedical Engineering, VOL. 59, NO 2, February 2012
15 mins
Pauly Ossenblok, Petra van Houdt, Albert Colon, Paul Boon, Jan de Munck
Abstract: EEG-correlated functional MRI (EEG-fMRI) has proven its application as a noninvasive technique for the pre-operative work-up of epilepsy surgery candidates. EEG-fMRI investigates the correlation between the occurrence of interictal epileptic discharges in the EEG (IEDs) and blood oxygenation level-dependent (BOLD) changes using a general linear model framework (GLM). However, this approach highly depends on the number of IEDs present in the EEG [1], which might not occur during the limited time of scanning or might not be visible at the scalp. In order to increase the ability to identify the epileptic network based on fMRI, we explored the feasibility of independent component analysis (ICA), a data-driven fMRI approach without the use of EEG. As a proof-of-concept study fMRI data were selected of seven patients from a previous EEG-fMRI study [2] for whom the results of the standard GLM approach were validated with the gold standard (i.e. invasive EEG recordings and surgical outcome). The fMRI data of these patients were divided into two epochs, one during which IEDs were present in the simultaneously recorded EEG (with IEDs) and a second one during which no IEDs were present (without IEDs). Both epochs were analyzed with ICA yielding a large number of independent components (ICs). To select the component related to epileptic activity, the spatial overlap was calculated for each IC with the resection area of a post-operative MRI and with the EEG-fMRI correlation pattern. The IC that showed a high degree of overlap with both was identified as the epileptic component (ICE). Furthermore, as a clinical application, EEG-fMRI data were acquired before and after withdrawal of anti-epileptic drugs (AEDs) in patients with localization-related epilepsy who are candidates for epilepsy surgery (n=5). For all patients of the proof-of-concept study an ICE could be selected both in the data with IEDs and without IEDs. For the patients in the clinical study, no IEDs were present in the EEG data recorded while taking AEDs. However, using the a priori information based on the standard GLM approach in combination with the ICA results we were able to identify the ICE for the fMRI data acquired in both conditions, with and without AED administration. This study shows that ICA enables the identification of the ICE of patients with localization-related epilepsy. These components can also be identified when no IEDs are present in the EEG. This could indicate the existence of a patient-specific epileptic resting-state network, similar to the existence of large scale resting-state networks in healthy volunteers [3], providing promising information for the clinical application of fMRI in the presurgical work-up of patients with epilepsy. REFERENCES [1] P. van Houdt, J. de Munck, M. Zijlmans, G. Huiskamp, F. Leijten, P. Boon, P. Ossenblok. “Comparison of analytical strategies for EEG-correlated fMRI data in patients with epilepsy”, Magn Reson Imaging Vol. 28, pp.1078-86 (2010). [2] P. van Houdt, P. Ossenblok, A. Colon, P. Boon, J. de Munck. “A framework to integrate EEG-correlated fMRI and intracerebral recordings”, Neuroimage, Vol. 60(4), pp. 2042-53 (2012). [3] C. Beckmann, M. DeLuca, J. Devlin, S. Smith. “Investigations into resting-state connectivity using independent component analysis”, Philos Trans R Soc Lond B Biol Sci, Vol. 360, pp. 1001-13 (2005).
15 mins
Kris Cuppens, Anouk Van de Vel, Peter Karsmakers, Bert Bonroy, Milica Milosevic, Stijn Luca, Tom Croonenborghs, Tinne Tuytelaars, Lieven Lagae, Berten Ceulemans, Sabine Van Huffel, Bart Vanrumste
Abstract: We focus on the detection of nocturnal convulsive seizures, more specific on hypermotor seizures. This type of seizures are marked by involuntary movement of the patient which can last for multiple seconds. Due to the violent movement the patient can injure himself. The early treatment may prevent consequences such as severe cerebral damage or mortality [1]. Epileptic seizure detection is traditionally done using video/electroencephalogram (EEG) monitoring, which is not applicable in a home situation. In recent years, attempts have been made to detect the seizures using other modalities. In this research we investigate if a combined usage of accelerometers attached to the limbs and video data would increase the performance compared to a single modality approach. Therefore, we used two existing approaches for seizure detection in accelerometers and video. The video detection method makes use of STIP features proposed by Laptev [2], the accelerometer approach makes use of different features from the time and frequency domain. In both approaches, an SVM classifier is used for classification. We combined both approaches (accelero and video) using a linear discriminant analysis (LDA) classifier. The seizure data is acquired at the Pulderbos rehabilitation center for children and youth where epileptic children are monitored and treated. The data is labeled based on the EEG and video. To be able to objectively compare both approaches, based on acceleration and on video, we use the same segmentation of the data for both modalities. The training and testing is done in a 10-fold randomization. This means that we randomly select a number of normal and epileptic movements for the training and test set and perform the modeling and validation 10 times using a different combination of movements, and average out the obtained results. This makes the results less dependent on the division of the data in a training and test set. The same randomizations are used in both approaches and for the acceleration/video integration. In a first test we combined the normalized features from both approaches in an early integration. In a second test, we combined the outputs of both individual classifiers from the video and accelerometer detection in a late integration using an LDA classifier. The output values of both classifiers give a probability of the sequence belonging to the epileptic seizure class, as we use the libsvm implementation for extending SVM to give probability estimates based on Wu et al. [3] and Lin and Weng [4]. The combined detection using the early integration seems to give a lower performance (sensitivity: 83.33%, positive predictive value (PPV): 96.00%) than the accelerometer detection alone (sens: 83.33%, PPV: 100.00%). This means that the video features do not have any added value in this integration. The late integration of both modalities has a small positive influence on the performance. Although there is a decline in the PPV from 100.00% to 97.50% compared with the accelerometer detection, the sensitivity increases to 86.67%.
15 mins
Arno Janssen, Dick Stegeman, Thom Oostendorp
Abstract: Transcranial magnetic stimulation (TMS) is a non-invasive technique that is used in a wide range of neurophysiologic and clinical studies to measure or change the excitability of specific brain areas. A very high and brief current is send through a coil that causes a time-varying magnetic field. This time-varying magnetic field in its turn induces an electric field in the human head as described with Faradays law of induction, which may generate neural excitation. Many cortical areas have been studied with TMS, but most of the protocols are developed for the motor cortex (M1). When M1 is stimulated a motor evoked potential (MEP) can be measured with the use of electromyography (EMG). The minimal intensity needed to evoke an MEP is called the motor threshold (MT). Most of the cortical areas outside M1 do not have a similar outcome measure like the MEP. The stimulation intensity used for these areas is commonly based on the MT in M1. There are, however, inter-individual differences in brain anatomy and physiology between cortical areas. Therefore, the intensities used for the cortical areas outside M1 are probably sub-optimal. To optimize the TMS induced effects, the inter-individual differences between M1 and other cortical areas should probably be included in the determination of the stimulation intensity. A way to verify if these inter-individual differences have an effect on the electric field is by using TMS simulations. In this study the finite element method (FEM) was used to calculate the induced electric field in a realistic head model based on geometry and conductivity, acquired from Magnetic Resonance Imaging (MRI) and Diffusion Tensor Imaging (DTI) measurements. The electric field was calculated for 15 cortical locations, which were determined based on multiple experimental studies. The results show that the magnitude of the induced electric field differs between cortical locations, as expected. The distance between the cortical location and the coil has the most prominent effect on the electric field magnitude. In addition the local anatomy and conductivity have an influence. The results indicate that an increase in intensity for cortical areas more distant from the coil (for example cerebellum) is needed to induce a similar electric field magnitude as over M1. However, a decrease in stimulation intensity for cortical areas closer to the coil would not be advisable.
15 mins
Sumientra Rampersad, Thom Oostendorp, Dick Stegeman
Abstract: Transcranial direct current stimulation (tDCS) entails sending a weak current through the head via two large planar electrodes attached to the scalp. This non-invasive stimulation painlessly induces polarity dependent cortical excitability modulations, making tDCS a promising technique for neurostimulation. Transcranial DC stimulation has been shown propitious in improving brain function in patients with neurological diseases like chronic pain, stroke, Parkinson’s disease, depression and epilepsy. Unfortunately, the effects are still too small and short-lived for tDCS to be used as a clinical therapy. Increasing the effect size of tDCS could possibly be achieved by better targeting the current, both in direction and amplitude. Volume conduction modeling studies have shown that the areas with the highest electric fields strengths do not, as is often assumed, lie beneath the electrodes. In order to find an electrode configuration that does result in maximum and more focal stimulation at the target area, we propose an inverse modeling approach. We simulate tDCS for ~7000 configurations and afterward determine which configuration leads to a maximal electric field at several target locations that are commonly used in tDCS research. A highly detailed finite element model of the complete head of a 25-year-old male was created by automatic segmentation of MR images followed by manual corrections. The model contains over 4 million tetrahedral elements and eleven tissue types. Special attention was given to the skull, the main barrier for the stimulation current, by including the spongiosa layer and skull holes. Anisotropy was derived from DTI measurements. To our knowledge, this is currently the most detailed model used for simulating tDCS. From the surface of this model, 86 equally spaced nodes were selected. For each combination of two points from this set, we placed 5 x 5 cm electrode patches onto the head model, centred on the two points, and simulated 1 mA tDCS. By comparing the direction and amplitude of the resulting electric field in the target area in all configurations, we determined the optimal stimulation configurations for motor cortex, dorsolateral prefrontal cortex, inferior frontal gyrus, occipital cortex and cerebellum stimulation. This approach can be used to optimize stimulation for any target location. We found different optimal configurations by looking at either strength or direction of the field. Comparing these configurations experimentally will not only verify our modeling approach, but also provide valuable information on the mechanisms behind tDCS.