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

Powered by
© Fyper VOF
Conference Websites
10:40   Neurophysiology: Biological Neural Networks
Chair: Alfred Schouten
15 mins
Yan Zhao, Hui Chen, Xiaorong Liu, John Troy
Abstract: The receptive field (RF) organization of retinal ganglion cells (RGCs) determines how they preferentially respond to visual stimuli. In the mouse retina, an increasingly popular model for vision research, most RFs are thought to be composed of an excitatory center and a concentric antagonistic surround [1]. However, a thorough systematic characterization of these RF organizations in the mouse is lacking. To advance our understanding about its structure and function, we developed a method to characterize both the center and surround mechanism of many RFs simultaneously at the level of spike output. We performed in vitro multielectrode array (MEA) recordings on whole-mount retina preparations, using spatiotemporal Gaussian white-noise (GWN) checkerboard stimuli to map numerous RFs via spike-triggered average analysis [2]. Typically in the mouse, GWN is not effective at eliciting spikes attributed to the surround mechanism, due to a low probability of simultaneously stimulating multiple sub-areas of the annular surround. To tease out the surround mechanism from that of the center, we used several different checker dimensions ranging from 50μm in diameter to fullfield stimulation to elicit different contributions from the center and surround. Since RGCs exhibit spatial tuning, stimulation of RGCs with different stimulus sizes resulted in an area response function, in which the RGC spike rate varies as a function of the stimulus size. By modelling the area response function as the integral of a difference of Gaussians [3], the spatial properties of center and surround are extracted from the parameters of the 2-dimensional Gaussians. Consistent with previous reports, 89.8% (N=557) of the RGCs had center-surround organization, with diameters of 253.0+8.6μm and 1008.3+31.9μm (mean+SEM), for the center and surround respectively. ON cells exhibited the largest center diameters (320.5+12.5μm), followed by ON-OFF cells (281.3+14.5μm) and OFF cells (263+11.2μm). Interestingly, the classification of 43.0% of RGCs, as determined by the response dominance index, changed as the stimulus size increased. The area-response GWN method for MEA recordings has numerous applications including physiological classification of mouse RGCs, developmental studies of the maturing retina, a better understanding of contrast sensitivity and adaptation, and the study of retinal degeneration in models of eye diseases.
15 mins
Marijn van Dongen, Wouter Serdijn
Abstract: Power efficiency and reliability are two key aspects for implantable neurostimulator devices. An integrated circuit (IC) stimulator design is proposed to address two important issues: 1. switched-mode operation, to allow for power efficient stimulation, and 2. Minimization of the number of external components to significantly increase reliability. The proposed stimulation scheme employs very high frequency current stimulation, in which current pulses with a pulse width in the order of 10ns are injected into the tissue with a frequency of 10MHz. The dynamic properties of the tissue [1] will filter these high frequencies and the resulting electric field can closely resemble the electric field resulting from classical stimulation strategies. The system is designed to have an implantable battery of 3 to 4V as its only energy source, while it is capable of delivering stimulation amplitudes up to 10V. The properties of the system allow for implementing the required safety mechanisms commonly found in neural stimulators, such as charge cancellation. Thanks to the high frequency switched-mode operation, the power efficiency can be increased up to 80%, reducing space requirements for the battery. Furthermore the system uses a single inductor as its only external component, thereby further increasing the implantability of the system. Tissue simulations confirm the ability of this high frequency stimulation to generate an electric field similar to classical stimulation strategies. A prototype has been developed and successfully tested on tissue models and a volunteer. Work is being carried out towards a fully integrated design.
15 mins
K. J. van Dijk, D. G.M. Zwartjes, M.L.F. Janssen, A. Benazzouz, Y. Temel, V. Visser-Vandewalle, Tjitske Heida, Peter Veltink
Abstract: ABSTRACT Parkinson’s disease is an age-related neurodegenerative disorder and is characterized by gradual deterioration of motor symptoms. Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is now widely used in neurosurgical therapy, because it markedly improves the Parkinson’s disease symptoms. Tracing studies have shown that the STN is segregated into multiple functional areas. We hypothesize that DBS should be targeted in the area responsible for motor function to reduce unwanted side effects and improve the positive DBS effects. Therefore, we are interested in localization of the functional areas of the STN. In our study, we determine the functional distribution of the synaptic inputs within the STN of anesthetized rats. To identify the functional areas, we stimulate the cortex, namely the motor cortex (MC) and cingulate gyrus (CG), and measure the subthalamic response. In this response signal, we can distinguish unit activity and the local field potential (LFP). Unit activity reflects the action potentials generated by neurons, while the LFP is a consequence of synchronized synaptic activity [1]. In other words, the unit activity is a measure of the output of a single neuron and the LFP the input of the underlying neuronal population. By carefully measuring the LFP response on 320 different points, within and around the STN, we are able to reconstruct the current source density (CSD) in the STN and surrounding volume. The CSD is used to track the sinks and sources of the neuronal input [2]. Cortical neuronal input from MC and CG are conveyed to the STN through two different pathways. The monosynaptic pathway directly enters the STN, while the polysynaptic pathway first passes through the striatum and globus pallidus (GP) before entering the STN. Previous electrophysiological studies have shown a typical multiphasic response in the STN after cortical stimulation, i.e. two excitatory periods, which are interrupted by a brief inhibition, and are followed by a long inhibitory period. Part of this cortically evoked response is explained by the influence of the different pathways [3]. CSD analysis of our recorded LFP data showed that the input from the MC and GP were locally distributed in the STN. The sink corresponding to the excitatory MC input was located medial to the source corresponding to the inhibitory GP input. Furthermore, the CSD analysis showed a strong local source evoking the long inhibitory period after MC and CG stimulation. These sources did not show a distinct separation of the functional areas in the STN, because both the MC and CG stimulation evoked the source on the same location. Although we did not see a distinct separation of the functional areas in the late response, the method in this study was well able to discriminate the sources and sinks evoked through the monosynaptic and polysynaptic pathway after MC stimulation. Therefore, in the future this is a potential approach to improve the localization of the STN motor area during DBS electrode placing in humans. REFERENCES [1] Buzsáki, G., 2004. Large-scale recording of neuronal ensembles, Nature Neuroscience, 7, pp.446–451. [2] Łęski, S. et al., 2007. Inverse current-source density method in 3D: reconstruction fidelity, boundary effects, and influence of distant sources. Neuroinformatics, 5(4), pp.207–22. [3] Magill, P.J. et al., 2004. Synchronous unit activity and local field potentials evoked in the subthalamic nucleus by cortical stimulation. Journal of neurophysiology, 92(2), pp.700–14.
15 mins
Wannaya Ngamkham, Marijn van Dongen, Wouter Serdijn
Abstract: Cochlear Implants (CI) are prosthetic devices that restore hearing in profoundly deaf patients by bypassing damaged parts of the inner ear and directly stimulating the remaining auditory nerve fibers in the cochlea with electrical pulses. Since the stimulators are implanted inside the body, these devices must be as small as possible. This means avoiding the use of external components, while simultaneously keeping the power consumption as low as possible to avoid the need for big batteries. A current source based stimulator architecture was chosen. First of all the voltage compliance of the current source must be as high as possible to allow for the lowest supply voltage possible. Furthermore the output impedance of the current source needs to be high to guarantee a well defined output current to handle a wide range of electrode-tissue impedances. This is particularly important to achieve charge cancellation in order to prevent electrolysis. This work presents the design of a constant current mode, biphasic neural stimulator circuit for cochlear implants that meets the demands mentioned above. Using a double-loop negative-feedback topology, the output impedance of the current generator is increased, while only a single transistor is needed at the output stage. This means that the compliance voltage of the current source is maximized, which allows the circuit to convey more charge into the tissue. The circuit can provide a biphasic stimulation scheme from a single-ended supply with an amplitude range of 10uA up to 1.05mA for a wide range of electrode-tissue impedances (RL=1k ohm~10k ohm, CL=1nF~10nF). The stimulation current is set by scaling a reference current using a two stage binary-weighted transistor DAC configuration (3 bits HV transistor DAC and a 4 bits LV transistor DAC) to improve the rise and fall times of the stimulation pulses and minimize the area of the circuit. Simulation results, using AMS 0.18m high-voltage CMOS IC, show that the charge error within a cycle (of 600us) is only 0.02%, equivalent to a DC current error of 3nA at the maximum stimulation current with a load of 10k ohm+10nF. The charge mismatch was found to be well below the safety limits. The die area of the chip is 200um × 200um, 5 times smaller than an electrode contact, which is useful for the future to bring the stimulator circuit close to electrode array to increase the number of channels and reduce the area of the device. Measurements of the manufactured chip are currently in progress.
15 mins
Chi Wing Wu, Senad Hiseni, Wouter Serdijn
Abstract: In medicine, there is a growing awareness that drugs cannot cure everyone and often produce harmful side-effects. Advances in technology, together with favorable results from actual monitoring, treatment and clinical trials, have led more physicians to recognize the potential clinical value of neurodevices. Neurodevices are implanted inside the body by a neurosurgeon and used to deliver electrical stimulation safely and in a controlled (albeit primitive) way for the clinical treatment of a great variety of nervous system diseases [1]. In order to stimulate the tissue effectively, voltages up to 20 V may be required. For such voltages, battery technology itself cannot provide power efficient implantable solutions. Therefore, a fully implantable, compact and efficient voltage up-converter is required. Moreover, the converter should be compatible with other medical devices like MRI scanners. Two main principles can be distinguished to implement voltage up-converters. Both types make use of the fundamental properties of reactive components, by switching them on and off to generate a high voltage. The first one is inductor based, and the second one is capacitor based. Advantage of the capacitor based type is that it is relatively easier to make it MRI compatible. The proposed voltage up-converter is realized by utilizing a charge pump technique. To increase the energy efficiency of the conversion, the output voltage of the converter can be modified by enabling and disabling consecutive stages of the converter. Using a Cockcroft-Walton [2] inspired charge pump structure, voltage drops across the switching transistors are kept minimal. The result is that transistors with low break-down voltage can be used, keeping chip area and breakdown risk minimal. The high density on-chip capacitors of the various stages are optimally distributed, in such a way that the startup and recovery time are optimized and the number of external components is kept minimal. To verify its performance, a voltage up-converter using the proposed techniques is designed to be implemented in 0.18 μm CMOS AMS technology. The design is verified by means of simulations in Cadence using RF Spectre. The simulations showed that the worst case circuit recovery time is 9 ms. The efficiency of the up-converter is 75 % and the voltage ripple remains below 10 % of the maximal output voltage.
15 mins
Bettina Schwab, Richard van Wezel, Tjitske Heida, Stephan van Gils
Abstract: The pathology of Parkinson's disease (PD) is characterized by modified behavior of neuronal networks in the basal ganglia after depletion of dopamine. PD states show bursting neural activity and high synchronization among neurons as well as altered oscillations in local field potentials. These network modifications are thought to be directly related to PD motor symptoms such as tremor, akinesia and bradykinesia. Computational models of the basal ganglia can reproduce physiological and pathological neuronal behavior depending on certain parameter values [1]. However, it is still a matter of debate what triggers and stabilizes the pathological behavior and how deep brain stimulation (DBS) influences it. In particular, computational models suffer from a lack of experimental data on structure and function. They commonly estimate the degree and course of chemical synapses and do not take electrical synapses (neuronal gap junctions) into account. Gap junctions (GJs), constructed out of connexins (Cxs), are direct electrical connections between cells. In different parts of the brain, neuronal gap junctions have been shown to be able to lead to synchronization, bursting and oscillations. They are remodeled in a number of different diseases and can change their conductance under the influence of neurotransmitters such as dopamine [2]. Current research also demonstrated activity-dependent plasticity of GJs [3]. In our experiments, we look for the occurence of Cx36, a neuronal Cx, by immunohistochemistry and confocal microscopy in rat tissue of the subthalamic nucleus (STN), globus pallidus external segment (GPe) and internal segment (GPi). We found spots of Cx36 in all three nuclei at a level that is roughly comparable to the Cx36 level in the striatum, where coupling of neurons via gap junctions is well described in literature. We investigated the impact of gap junctions in a computational model of the basal ganglia, the Rubin-Terman-Model [4] including STN, GPe and GPi. Gap junctions between neighboring neurons were implemented as ohmic resistors inside the three nuclei. At low conductances of the gap junctions, the network showed irregular (physiological) states. After uprising gap junction conductance, STN and GPe showed bursting and the firing rate of the GPi increased. In all neurons, synchronization was enhanced. These states are in high accordance with experimental measurements of neural activity in PD patients and animal models of PD. Our results strongly suggest a role of neuronal gap junctions in PD pathology. The modulation of gap junctions by dopamine might be a candidate for the remodeling of oscillations, synchronization and bursting. A better understanding of these basic processes can in later stages lead to improved therapies such as refined stimulation protocols for DBS or novel medications. Currently we are extending our experiments to human post-mortem tissue of PD patients and controls. The aim is to quantify differences in Cx36 expression and adjust the computational model to these changes.