6th Dutch Bio-Medical Engineering Conference
26 & 27 January 2017, Egmond aan Zee, The Netherlands
13:30   EEG/fMRI II
Chair: Teodoro Solis-Escalante
15 mins
Olena Filatova, Lucas van Vliet, Alfred Schouten, Frans van der Helm, Frans Vos
Abstract: Motivation. Stroke is an interruption of the blood supply to a brain area, typically due to a blood vessel burst (hemorrhagic) or blockage by a clot (ischemic). In the Western World about 10 million people per year survive after a stroke and suffer from long-term disability such as limb or facial paresis, difficulty in verbal and written communication. A better insight into white matter network alterations after stroke onset could help to understand the underlying recovery mechanisms and improve interventions applied in the future. Recently, advanced diffusion MR imaging techniques [1] were developed for assessing white matter integrity. Objective. Loss of corticospinal tract integrity due to stroke is commonly associated with motor impairment [2]. Simultaneously, stroke is known to affect other white matter structures. We aim to compare differences in white matter tracts between the healthy and affected hemispheres in chronic stroke patients using diffusion tensor imaging. Methods. Eighteen stroke patients were scanned with a two b-value diffusion-weighted MRI protocol. This allowed us to exploit the standard single tensor model, as well as more sophisticated diffusion models: (1) single tensor (2) single tensor with isotropic compartment (3) bi-tensor model (4) bi-tensor with isotropic compartment. From each model we derived mean tract fractional anisotropy (FA), mean, radial and axial diffusivities. The unaffected hemisphere was taken as a reference. The outcome measures were also correlated with clinical scores of stroke recovery. Results Significant asymmetry between ipsi- and contralesional hemisphere was found in the corticospinal tract consistently for all the four models and across the diffusivity measures. For example, interhemispheric fractional anisotropy ratio in the corticospinal tract was significantly different from 1 (two-sided Wilcoxon signed rank test) with p=0.004 for model (1) and p=1.96*10-4 for models (2)-(4). The outcome measures significantly correlated with the clinical scores of stroke recovery. The lowest correlation of the corticospinal tract FA and Fugle-Meyer upper extremity assessment score is with the single tensor model (r=-0.5, p=0.033) and the highest correlation is with the single tensor model including the isotropic compartment (r=-0.79, p=10-4). Conclusion. We found systematic differences in white matter properties between ipsi- and contralesional hemisphere outside the lesion site. In the future we are planning to investigate how more elaborate diffusion models can improve prognostic models for recovery of acute stroke patients. REFERENCES [1] M. W. A. K. Caan, H. Ganesh; Poot, Dirk H. J.; den Dekker, Arjan J.; Olabarriaga, Silvia D.; Grimbergen, Kees A.; van Vliet, Lucas J.; Vos, Frans M., "Estimation of Diffusion Properties in Crossing Fiber Bundles," IEEE Transactions on Medical Imaging, vol. 29, pp. 1504-1515, 2010. [2] W. D. Byblow, C. M. Stinear, P. A. Barber, M. A. Petoe, and S. J. Ackerley, "Proportional recovery after stroke depends on corticomotor integrity," Annals of Neurology, vol. 78, pp. 848-859, 2015
15 mins
Matthijs Perenboom, Yuan Yang, Alfred Schouten, Else Tolner
Abstract: Patients with migraine are more sensitive to light, and about one third of patients experience visual premonitory signs (auras). Little is known on the mechanisms contributing to these visual system changes. Processing of visual input by the brain is a highly nonlinear operation, involving complex interactions among cortical and subcortical neuronal networks. Nonlinear visual system activity includes harmonic interactions, believed to reflect resonance of neural processing, whereas intermodulation, which reflects the contribution of multiple input frequencies to one output frequency, relates to functional integration [1]. Using a sum-of-sinusoid signal as visual input [2], it is possible to elicit a richer class of nonlinear responses than the classic pulse train stimulus, thereby providing a more complete assessment of nonlinearity. Here, we used sum-of-sinusoid light stimulation and nonlinear EEG analyses to quantify higher-order nonlinearities in visual processing in migraine. Migraine patients with aura, without aura and healthy participants (N=10/group) were subjected to bi-sinusoidal light stimulation for 320 1sec-epochs, while scalp EEG (8 electrodes) was recorded at the occipital, parietal and frontal lobes. Light stimulus frequencies were chosen to guarantee no overlap of their harmonic and intermodulation frequencies for different orders of nonlinearity. Nonlinear interactions and time delay from stimulus to cortical EEG response were analysed in the frequency domain using novel phase synchronization measures [3] and amplitude spectral measures. Higher harmonic and intermodulation interactions were detected between visual input and cortical responses. For all groups, first to fourth order phase coupling interactions were enhanced in the visual cortex compared to parietal and frontal responses. Migraine patients with aura showed a decreased time delay only at the occipital lobe compared to healthy controls and migraine patients without aura. These results indicate that visual processing is altered in migraine patients with aura compared to healthy controls and patients without aura. Furthermore, this study demonstrates the potential of quantifying nonlinear interactions and temporal dynamics in the visual system using sum-of-sinusoid light stimulation. Hereby, we are able to uncover alterations in visual processing in the context of neurological disease. REFERENCES [1] Friston KJ. Book review: Brain function, nonlinear coupling, and neuronal transients. Neuroscientist. 2001 Oct 1;7(5):406-18. [2] Victor J, Shapley R. A method of nonlinear analysis in the frequency domain. Biophys J. 1980 Mar;29(3):459. [3] Yang Y, Solis-Escalante T, Yao J, Daffertshofer A, Schouten AC, van der Helm FC. A General Approach for Quantifying Nonlinear Connectivity in the Nervous System Based on Phase Coupling. Int J Neural Syst. 2016 Feb;26(01):1550031.
15 mins
Teodoro Solis-Escalante, Martijn Vlaar, Alfred Schouten, Erwin van Wegen, Julius Dewald, Gert Kwakkel, Frans van der Helm
Abstract: Cortical lesions due to stroke can drastically impair sensory and motor function of the upper limb [1], leading to poor execution of activities of daily living and reduced quality of life. Maximal recovery after stroke can be achieved through the selection of patient-specific rehabilitation programs that promote plastic reorganization [1, 2]. Current clinical practice has a strong emphasis toward motor-related impairments, whereas sensory-related impairments have received less attention in stroke rehabilitation [3]. This study quantifies cortical responses evoked by periodic wrist manipulations in individuals with chronic hemiparetic stroke. The aim is to assess the integrity of the somatosensory system through a combination of joint manipulation and high-density electroencephalogram. In this study, a robotic manipulator applied continuous periodic disturbances to the wrist of the participants to provide somatosensory (i.e., proprioceptive and tactile) stimulation. Cortical responses to the periodic manipulation were quantified using the power of the somatosensory steady-state responses [4] and their signal-to-noise ratio. Thirty individuals with chronic stroke and nine healthy (age-matched) volunteers participated in this study. Participants with stroke were classified as having severe, mild, or no sensory impairment, based on evaluation with the Erasmus modification of the Nottingham Sensory Assessment for the upper extremity [5]. The periodic continuous manipulation was applied to the affected wrist of participants with stroke and to the dominant wrist of healthy participants. Somatosensory steady-state responses measured during resting conditions (relaxed wrist), occurred in the brain hemisphere contralateral to the stimulation side in healthy participants and chronic stroke participants with mild and no sensory impairment. In contrast, the somatosensory steady-state responses were strongly reduced or absent in participants with severe sensory impairment. These results suggest a relation between sensory impairment and the power of the somatosensory steady-state responses. This study suggests that with periodic joint manipulations it is possible to quantitatively (indirectly) assess the integrity of the somatosensory system, thus providing additional information to assist the selection of patient-specific rehabilitation programs. REFERENCES [1] P. Langhorne, J. Bernhardt, and G. Kwakkel, "Stroke rehabilitation", The Lancet, vol. 377, pp. 1693-1702, 2011. [2] M. A. Dimyan and L. G. Cohen, "Neuroplasticity in the context of motor rehabilitation after stroke", Nature Reviews Neurology, vol. 7, pp. 76-85, 2011. [3] N. Bolognini, C. Russo, and D. J. Edwards, "The sensory side of post-stroke motor rehabilitation", Restorative Neurology and Neuroscience, vol. 34, no. 4, pp. 571-586, 2016. [4] S. Tobimatsu, Y.M. Zhang and M. Kato, “Steady-state vibration somatosensory evoked potentials: physiological characteristics and tuning function”, Clinical Neurophysiology, vol. 110, pp. 1953 – 1958, 1999. [5] F. Stolk-Hornsveld, J. Crow, E. Hendriks, R. Van Der Baan, and B. Harmeling-Van der Wel, "The Erasmus MC modifications to the (revised) Nottingham Sensory Assessment: a reliable somatosensory assessment measure for patients with intracranial disorders", Clinical Rehabilitation, vol. 20, pp. 160-172, 2006.
15 mins
Konstantina Kalogianni, Jan de Munck, Guido Nolte, Alistair Vardy, Frans van der Helm, Andreas Daffertshofer
Abstract: When looking for clinical electro-encephalography (EEG) biomarkers the sources-of-interest often are activated within the same lobe. In particular, when studying stroke recovery might be predicate by changes within SI. For estimating this, a sub-centimeter spatial resolution is mandatory. Defining EEG’s ‘spatial resolution’ is all but trivial. It depends on many parameters like the choice of the inverse method, the resolution of the leadfield matrix, the configuration of underlying sources and the amount and structure of noise in the recordings. The purpose of here-presented study was to examine the factors that affect the spatial resolution of EEG and to set limits of tracking activation changes in the somatosensory cortex after stroke recovery. Our study relied on numerical simulations in which two dipole sources changed in time with waveforms resembling somatosensory-evoked potentials with peaks at ‘expected’ moments (N20, P50, P100). Inter-dipole distances were varied and Gaussian white or realistic EEG-lie noise was added to the simulated scalp recordings. We used a range of signal-to-noise ratios (SNRs). Prior to inverse modeling, pre-whitening [1] was applied in both the simulated data and the computed leadfield. We explored three algorithms for source reconstruction: a least squares method (two stationary dipole fit) [1], a scanning method (sc-MUSIC) [2], and a distributed method (sc-eLORETA) based on eLORETA [3]. The quality of source localization was assessed via the distance between the simulated point sources and the estimated ones. To quantify the resulting ‘spatial resolution’ of EEG we introduced the so-called separability of sources, i.e. the distinct separation of two dipolar sources in the brain with a specific inter-dipole distance. Our simulations revealed separability of two sources in the presence of realistic noise with SNR up to 3 if they are 11 mm or further apart, especially when using a two dipole fit (which we expected given we did simulate two dipoles) but also when using sc-MUSIC. Importantly, we found that accounting for spatial correlation of the background noise via spatial pre-whitening drastically improved the source localization accuracy significantly. This resulted in almost white noise case accuracies that here-for served as a litmus test. Those results suggest that it may be possible for stroke researchers to look at subtle, rehabilitation-related changes within the somatosensory cortex by using EEG.
15 mins
Rick van der Vliet, Gerard Ribbers, Yves Vandermeeren, Maarten Frens, Ruud Selles
Abstract: Context: It is unclear if and how transcranial direct current stimulation (tDCS) improves motor skill learning in chronic stroke patients. We investigated if tDCS affects motor skill learning when (1) tDCS overlaps with training, (2) tDCS aftereffects overlap with training or (3) tDCS precedes training without the stimulation or its aftereffects overlapping with training. Objective: We tested chronic stroke patients on a circuit tracing task for which positive effects of tDCS on motor skill learning have been found. We compared conventional tDCS (20 minutes bihemispheric 1mA stimulation) during training, conventional tDCS 1 day before training and long-lasting tDCS (10 minutes bihemispheric 1mA stimulation – 25 minutes break – 10 minutes bihemispheric 1mA stimulation) 1 day before training, with sham stimulation. As a secondary objective, we investigated whether motor skill learning is affected by carrying the brain-derived neurotrophic factor (BDNF) Val66Met polymorphism, which is related to neuroplasticity. Design, Setting, Participants: Between-subjects randomized controlled trial with 80 chronic stroke patients. Main Outcome Measure: Motor skill learning on the circuit tracing task from day 1 to day 9 of the study. Results: Even though patients strongly improved on the circuit tracing task (unadjusted model βoverall_training=0.512; 95%HDI=[0.374 0.652]), none of the stimulation protocols affected the amount of learning (adjusted model βdirect_effects=0.0933; 95%HDI=[-0.0936 0.280]; βaftereffects=0.0193; 95%HDI=[-0.152 0.222]; βintermediate_effects=-0.0939 95%HDI=[-0.275 0.100]). However, we found a significant effect of the BDNF polymorphism, with having at least one BDNF Met allele negatively affecting motor skill learning (adjusted model βBDNF_Met =-0.193 95%HDI=[-0.330 -0.0548]). Conclusion: tDCS does not affect motor skill learning in chronic stroke patients, which is at odds with most of the previous studies. However, we did find a significant effect of the presence of the BDNF Val66Met polymorphism, indicating that this could be an important predictor for motor skill learning and potentially recovery after stroke.
15 mins
Yuan Yang, Alfred Schouten, Jun Yao, Teodoro Solis-Escalante, Julius Dewald, Frans van der Helm
Abstract: Introduction: Reflexes are involuntary reactions in response to unexpected perturbations. Existing studies mainly assess linear input-output relations in the human stretch reflex, though several basic elements in the stretch reflex have been shown to be highly nonlinear [1]. We recently developed a novel nonlinear connectivity measure, namely multi-spectral phase coherence (MSPC) [2]. Using MSPC and multi-sine wrist perturbations, for the first time, we assessed the nonlinear connectivity in the human stretch reflex and the time delay [3]. Methods: Subjects (n=11) exerted a constant wrist flexion torque (1 Nm) during receiving a multi-sine joint perturbation (7, 13 and 29 Hz, period: 1s). The experiment included 60 trials of 22 s each. Brain activity was recorded by EEG using a 128-channel cap (5/10 system). Muscles activity was recorded by bipolar EMG electrodes from the flexor carpi radialis. We computed the directional nonlinear connectivity and time delay from the perturbation to EEG/EMG, respectively, using MSPC. Additionally, the bi-directional nonlinear connectivity between EEG and EMG was also calculated. The sources of the EEG-related connectivity were estimated by sLORETA algorithm [4]. Results: Directional nonlinear connectivity was significantly stronger from the perturbation to the EMG than to the EEG (P < 0.05). Directional nonlinear connectivity from EEG to EMG was very weak. The time delay from the perturbation to the EMG was 33 ± 6 ms, similar to the reported latency of the spinal reflex. The time delay from the perturbation to the EEG was 43 ± 8 ms, in line with the time course of dorsal column pathway conduction to the primary somatosensory cortex. Cortical sources of EEG-related nonlinear connectivity localized at the primary sensorimotor areas (M1 and S1), suggesting a transcortical contribution of these cortices to the stretch reflex. Conclusions: Our results suggest a dominant role of the spinal loop in the nonlinear muscle activity in response to mechanical perturbation. Although the sensory signal arrives to the sensorimotor cortices, the transcortical reflex loop only slightly contributes to the nonlinear connectivity. This study provides new evidence of nonlinear neuronal synchronization in the stretch reflex. REFERENCES [1] Gielen, C., and Houk, J. (1987), Biol Cybern 57, 217-231. [2] Yang, Y., et al (2016a). Int. J. Neural Syst 26(1), 1550031. [3] Yang, Y., et al (2016b). Int. J. Neural Syst 26(8), 1650043. [4] Pascual-Marqui R. (2002). Methods Find Exp Clin Pharmacol, vol. 24, pp. 5-12.