Ruurd Schoonhoven

Potential distributions and neural excitation patterns in a rotationally symmetric model of the electrically stimulated cochlea

In spite of many satisfactory results, the clinical outcome of cochlear implantation is poorly predictable and further insight into the fundamentals of electrical nerve stimulation in this complex geometry is necessary. For this purpose we developed a rotationally symmetric volume conductor model of the implanted cochlea, using the Boundary Element Method (BEM). This configuration mimics the cochlear anatomy more closely than previous, unrolled models. The calculated potential distribution in the cochlea due to stimulating electrodes is combined with a multiple non-linear node model of auditory nerve fibres, which we recently developed. The combined model is used to compute excitation profiles of the auditory nerve for a variety of stimulus levels and electrode positions. The model predicts that the excitation threshold, the spatial selectivity and the dynamic range depend on the exact position of the electrode in the scala tympani. These results are in good agreement with recently published electrical ABR data. It is shown that the use of actively modelled nerve fibres is essential to obtain correct predictions for the biphasic stimuli typically used in cochlear implants and that unrolling the cochlear duct as done in previous models leads to erroneous predictions regarding modiolar stimulation.

Improving the accuracy of the boundary element method by the use of second-order interpolation functions [EEG modeling application]

The boundary element method (BEM) is a widely used method to solve biomedical electromagnetic volume conduction problems. The commonly used formulation of this method uses constant interpolation functions for the potential and flat triangular surface elements. Linear interpolation for the potential on a flat triangular mesh turned out to yield a better accuracy. In this paper, the authors introduce quadratic interpolation functions for the potential and quadratically curved surface elements, resulting from second-order spatial interpolation. Theoretically, this results in an accuracy that is inversely proportional to the third power of element size. The method is tested on a four concentric sphere geometry, representative for electroencephalogram modeling, and compared to previous solutions of this problem in literature. In addition, a cylindrical test configuration is used. It is concluded that the use of quadratic interpolation functions for the potential and of quadratically curved surface elements in BEM results in a significant increase in accuracy and in some cases even a reduction of the computation time with the same number of nodes involved in the calculations.

Single-fibre and whole-nerve responses to clicks as a function of sound intensity in the guinea pig

This paper describes a study of the intensity dependence of click-evoked responses of auditory-nerve fibres in relation to the simultaneously recorded compound action potential (CAP). Condensation and rarefaction clicks were presented to normal hearing guinea pigs over an intensity range of 60 dB. The recorded poststimulus time histograms (PSTHs) were characterized by the latency (tp), amplitude (Ap) and synchronization (Sp) of their dominant peak, parameters that are particularly important for the understanding of the CAP. For all fibres tp decreased monotonically with increasing intensity, in a continuous way for fibres with high characteristic frequency (CF greater than 3 kHz), and in discrete steps of one CF-cycle for low-CF (CF less than or equal to 3 kHz) fibres. An additional analysis of PSTH envelopes revealed that average latency shifts with intensity are similar for all CFs above 2 kHz. For all fibres Ap increased monotonically with intensity; the increase was stronger and maximum values were larger for low-CF than for high-CF fibres. A schematic model PSTH was then formulated on the basis of the experimental data. A sum of these model PSTHs from a hypothesized fibre population was convolved with an elemental unit response (Versnel et al., 1992) in order to simulate the compound action potential. Synthesized CAPs agreed with experimental CAPs in their main aspects.

Integrated use of volume conduction and neural models to simulate the response to cochlear implants

Abstract Cochlear implants are electronic devices intended to restore the sense of hearing in deaf people by direct electrical stimulation of the auditory nerve fibres that are still present in the deaf inner ear. Unfortunately, the clinical outcome is not very predictable. In this study a computational model is presented that can predict the neural response to an arbitrary cochlear implant. It first computes the potential distribution set up in a three-dimensional, spiralling volume conduction model of the auditory part of the inner ear (cochlea) and then applies a nerve fibre model to construct input/output curves and excitation profiles of the auditory nerve. As an initial validation the results are compared with experimentally induced electrically evoked auditory brainstem responses. In the light of the favourable results, we conclude that the model can serve as a tool for designing future cochlear implants. In combination with electrophysiological measurements in the individual patient it is applicable as an implant fitting tool.

The Forward Problem in Electroneurography I: A Generalized Volume Conductor Model

We describe a generalized volume conductor model for the compound action potential (CAP) of a peripheral nerve in situ. The extracellular single fiber action potentials (SFAPs), the constituting elements of the CAP, are expressed in terms of the intracellular action potentials and of the effect of volume conduction using a convolution formulation. The model incorporates variations in the intracellular action potential duration over the fiber population. Volume conduction is described in a generalized formalism for a class of cylinder symmetrical configurations. The CAP is finally formulated as a linear sumation of the SFAPs, with incorporation of the distribution of propagaytion velocities over the fiber population. We show that the final expressions for SFAPs and CAP can be given in a mathematically transparent form, which gives a clear insight into the mechanisms involved in the genesis of different potential waveshapes.

Group delays of distortion product otoacoustic emissions in the guinea pig.

This paper presents a comprehensive set of experimental data on group delays of distortion product otoacoustic emissions (DPOAEs) in the guinea pig. Group delays of the DPOAEs with frequencies 2f1-f2, 3f1-2f2, 4f1-3f2, and 2f2-f1 were measured with the phase gradient method. Both the f1- and the f2-sweep paradigm were used. Differences between the two sweep paradigms were investigated for the four DPOAEs, as well as the group delay differences between the DPOAEs. Analysis revealed larger group delays with the f2-sweep paradigm, but only for the lower sideband DPOAEs (with fdp < f1,f2). For the lower sideband cubic distortion product 2f1-f2, the f2-sweep delays were a factor of 1.17-1.54 larger than the f1-sweep delays, depending on frequency. The upper sideband DPOAE 2f2-f1 showed no significant difference between f1- and f2-sweep group delays, except for the highest and lowest f2 frequencies. Comparing the group delays of the DPOAEs for each sweep paradigm separately, equal group delays were found for all four DPOAEs measured with the f1-sweep. With the f2-sweep paradigm on the other hand, the group delays of the three lower sideband DPOAEs occurred to be larger than the group delays of the upper sideband DPOAE 2f2-f1. A tentative interpretation of the data in the context of proposed explanatory hypotheses on DPOAE group delays is given.

Transmitter release in inner hair cell synapses: a model analysis of spontaneous and driven rate properties of cochlear nerve fibres

The inner hair cell (IHC) synapse is one of the stages of cochlear processing that determine the relation between sound pressure level and spike rate in auditory nerve fibres. Transmitter released in the non-stimulated condition is held responsible for the wide range of spontaneous spike rates (SR) observed in these fibres. Properties of stimulated spike activity in auditory nerve fibres, including rate threshold and operating range of a fibre, are known to systematically vary with SR. This paper presents a model analysis of the relation between IHC transmembrane potential and transmitter release rate as becoming manifest in these spontaneous and driven rate properties. A previously developed computational model is used to identify those transfer properties of its synapse section which lead to reproduction of the variation of rate thresholds, shapes of rate-intensity functions and maximal driven rate with SR known from the literature. First a simple additive release model, in which driven transmitter release depends linearly on IHC potential, is elaborated. Its results lead to the hypothesis that the true release function is non-linear and variable across synapses generating different SR. An exponential release function is then introduced, with parameters varying across SR in a physiologically dictated way. This approach leads to adequate reproduction of the variation in rate thresholds and rate-intensity functions with SR. Finally, the model is applied in an inverse way to directly estimate the release function from given rate-intensity functions. The conclusion of both forward and inverse model analyses is that transmitter release is a non-linear function of IHC potential which, by the systematic variation of its parameters across SR, effectively leads to the physiological variation in dynamic range across fibres of different SR. Possible relations of these results with ultrastructural morphology and basic physiology of IHC synapses are discussed.

Round-window recorded potential of single-fibre discharge (unit response) in normal and noise-damaged cochleas

Unit responses (URs) of eighth-nerve fibres have been determined at the round window by spike-triggered averaging in both normal and pathological guinea pig cochleas. The pathology was mainly noise-induced damage. The URs have been analysed with respect to their dependence on the fibres threshold, characteristic frequency (CF) and spontaneous rate (SR). The results from normal cochleas confirmed earlier data (Prijs, 1986): the UR has a diphasic waveform and the amplitude of its negative first peak is about 0.1 microV. From the six parameters (amplitude, latency, and width of the two peaks) by which the UR was described only the amplitude of the positive peak showed a significant variation with CF: a small decrease with increasing CF (CF-range 0.1 to 20 kHz). This finding may possibly be caused by oscillations in the spike-triggered average for low CFs. URs for most low- and medium-SR fibres were found to be large (greater than 0.3 microV). However, this result is interpreted as an artefact caused by synchrony of fibre spontaneous activity. In damaged cochleas only slight changes of the UR were found: the waveform duration became significantly shorter and on some occasions the positive peak increased in amplitude, but latency and amplitude of the negative component of the UR remained unchanged.

The Prognostic Value of Electrocochleography in Severely Hearing-impaired Infants

This paper presents a longitudinal evaluation of electrocochleographic assessment in severely hearing-impaired infants. Electrophysiological data were obtained by transtympanic electrocochleography to tone-burst stimuli at octave frequencies of 500 to 8000 Hz at the age of 0-6 years in a group of 126 subjects. The results are compared with auditory thresholds determined at school age in the same children by means of pure-tone audiometry. Cochlear microphonics could be recorded in virtually all ears, although the majority of subjects had hearing losses of 90 dB and more. Compound action potentials (CAPs) showed waveforms varying from normal to a wide range of abnormalities. Audiometric thresholds correlated generally well with the compound action potential (CAP) thresholds obtained in infancy. The error in the predicted audiometric thresholds is between 15 and 20 dB, as compared with 11 dB reported for more moderate hearing losses. It is shown that, in spite of the high stimulus levels used, substantial frequency-specific threshold information is retained. Occasional large discrepancies in thresholds were often associated with markedly abnormal response waveforms. Among the many cases in which no ABR could be elicited, 68 per cent produced detectable electrocochleographic responses in the 1000-4000 Hz range. It is concluded that electrocochleography is a valuable method for the assessment of residual hearing in infants suspected of having a severe hearing impairment.

Recovery characteristics of auditory nerve fibres in the normal and noise-damaged guinea pig cochlea

Spontaneous activity was analysed in auditory-nerve fibres innervating normal and noise-damaged cochleas. Spike occurrences were conceived as point processes. Joint interval distributions and serial correlation coefficients reveal a weak history effect for succeeding intervals. The point process is regarded as a renewal and the recovery function, being proportional to the hazard function, is determined from the interval probability density function. In 29 out of 60 fibres the latter shows peculiarities which result in a deviation from a monotonically increasing recovery function. For three fibers of low characteristic frequency the interval probability function shows an oscillatory pattern and for 26 fibres this function exhibits an early, sharp peak around 1.1 ms irrespective of characteristic frequency, spontaneous rate, or cochlear damage. The recovery function is not different between fibres with normal and those with abnormally high thresholds and exhibits an exponential recovery with one time constant of average value 1.6 ms. Bursting activity is found in only one fibre from the abnormally high threshold group.