Project topics and materials
Get Complete Project Material File(s) Now! »
Context of the problem
Cochlear implants have been developed to help rehabilitate profoundly deaf persons by providing them with a measure of sound perception through electrical stimulation of auditory nerve fibres (ANFs). The cochlear implant injects electric currents directly into the cochlea by means of an electrode array to stimulate the ANFs directly. The challenge facing researchers is how to convey meaningful speech information to the brain via electrical stimulation. Modern implants are multiple-electrode implants, typically utilizing 16 to 22 electrodes, to take advantage of the tonotopic organization of the ANFs (Loizou, 1998).
However, while the average performance of cochlear implants has improved over the last three decades, large variability in speech performance across individual implant users is still a major problem (Shannon, Fu, Galvin III and Friesen, 2004). This can in part be ascribed to dissimilar neural excitation spread patterns, both intrasubject and intersubject, as a result of variability in factors such as implant type, degree of degeneration of the auditory nerve fibre population across human subjects, electrode geometry, intrascalar electrode location and stimulation strategy (Nadol Jr, 1990; Schuknecht, 1993; Zimmermann, Burgess and Nadol Jr, 1995; Nadol Jr, 1997; Arts, Jones and Anderson, 2003; Cohen, Richardson, Saunders and Cowan, 2003; Abbas, Hughes, Brown, Miller and South, 2004; Abbas and Miller, 2004; Van Wieringen, Carlyon, Laneau and Wouters, 2005; Fayad and Linthicum Jr, 2006). Even though potential implantees undergo pre-operative auditory testing, the successful outcome of the implantation is not known until after the implant has been switched on (Niparko, 2004). Ideally the electrodes should be situated closest to the sites of surviving ANFs, since this leads to reduced power consumption in the implant, lower stimulation thresholds, narrower neural excitation spread patterns and an increased dynamic range (Townshend and White, 1987; Shepherd, Hatshushika and Clark, 1993; Rebscher, Snyder and Leake, 2001; Abbas and Miller, 2004; Leake and Rebscher, 2004; Glueckert, Pfaller, Kinnefors, Rask-Andersen and Schrott-Fischer, 2005a). A telemetric measuring system for cochlear implants, called Neural Response Telemetry (NRT) by Cochlear Limited, is available to measure the electrically evoked compound action potential (ECAP) of the ANFs (see for example Abbas, Brown, Shallop, Firszt, Hughes, Hong and Staller, 1999; Dillier, Lai, Almqvist, Frohne, M¨uller-Deile, Stecker and van Wallenberg, 2002). ECAP data can be used to obtain an objective estimate of the dynamic range, and thus the extent of neural survival, through measurement of the behavioural threshold and most comfort level (MCL) of implantees (see for example Abbas et al., 1999; Franck and Norton, 2001; Dillier et al., 2002). It is also used to examine the extent to which psychophysical measurements reflect the amount of neural excitation spread (Cohen et al., 2003).
The primary means of modelling the electrically stimulated human auditory system is using mammalian research animals, especially cats and guinea-pigs (see for example Javel, Tong, Shepherd and Clark, 1987; Abbas and Miller, 2004). Since these animals’ cochleae have larger dimensions than those of smaller rodents, multi-electrode arrays similar to those implanted in humans, can be used. Computational models are used in combination with animal studies to enhance understanding of the underlying physiology of electric hearing (Abbas and Miller, 2004). Several ANF models have been developed (Bruce, White, Irlicht, O’Leary, Dynes, Javel and Clark, 1999c; Frijns, Briaire and Schoonhoven, 2000; Matsuoka, Rubinstein, Abbas and Miller, 2001; Rattay, Lutter and Felix, 2001b; Briaire and Frijns, 2005; Macherey, Carlyon, van Wieringen and Wouters, 2007). The ANF models are frequently used in combination with volume-conduction models of the cochlea to predict neural excitation profiles (Frijns et al., 2000; Hanekom, 2001b; Rattay, Leao and Felix, 2001a). An advantage of the computer models is that it is possible to isolate and manipulate critical model para-meters, as well as modelling of the differences in cochlear anatomy and ANF physiology between species, which is not always feasible with animal studies (Morse and Evans, 2003). Most of these models are at least partially based on animal data. However, the differences in cochlear structures between animals and humans, differences in the number and percentage myelination of auditory nerve fibres and innervation patterns of both inner and outer hair cells across species, may be physiologically significant and care must be taken when extrapolating the animal results to predict results in human implantees (Nadol Jr, 1988; Frijns, Briaire and Grote, 2001). Computer models can only approximate a real neural system, owing to the complexity of the latter. The modeller therefore has to make certain simplification assumptions when abstracting the real system and this place some limitations on the model as to the realistic correctness and completeness of the physiology and anatomy represented (Morse and Evans, 2003).
Research gap
Human ANF models have been developed by Briaire and Frijns (2005; 2006) and Rattay et al. (2001b). These models are partially based on human morphometric1 data, while the ionic current dynamics are still those of rat and squid respectively. However, the Briaire and Frijns (2005) model cannot fully account for the ECAP morphology2 observed in humans, while Macherey et al. (2007) argue that the ion channels of the squid based model, on which the Rattay et al. (2001b) model is based, are not sufficient to account for non-monotonic excitation behaviour experimentally observed. Hence, a more comprehensive computer model is needed, based on human cochlear dimensions and peripheral ANF characteristics, incorporating simulation of temporal characteristics.
Even though the physical structure of human ANFs has been investigated (refer to Section 2.2), the properties and types of ionic membrane currents of spiral ganglion cells have been characterised in murine (Mo, Adamson and Davis, 2002; Reid, Flores-Otero and Davis, 2004; Hossain, Antic, Yang, Rasband and Morest, 2005; Chen and Davis, 2006) and guinea-pig (Bakondi, P´or, Kov´acs, Szucs and Ruszn´ak, 2008), but not in human. Since the human ANF is of the peripheral sensory type, the possibility exists that similar ionic membrane currents to those found in a peripheral sensory fibre might be present. Ionic membrane current data from single human myelinated peripheral nerve fibres have been recorded by Reid, Bostock and Schwarz (1993), Scholz, Reid, Vogel and Bostock (1993), Schwarz, Reid and Bostock (1995) and Reid, Scholz, Bostock and Vogel (1999), but only the Schwarz et al. (1995) data have been used to develop a human nerve fibre model. However, to date none of these data have been applied to simulate human ANFs (Section 2.1.1). The development of a general human peripheral sensory nerve fibre can hence serve as an intermediate step to develop a human ANF model, until ionic membrane current data from human ANFs become available.
1 INTRODUCTION
1.1 PROBLEM STATEMENT
1.2 RESEARCH OBJECTIVE AND QUESTIONS
1.3 HYPOTHESIS AND APPROACH
1.4 RESEARCH CONTRIBUTION
1.5 OVERVIEW OF THE STUDY
2 BACKGROUND REVIEW
2.1 PHYSIOLOGICALLY BASED MODELS
2.2 PERIPHERAL AUDITORY NERVE FIBRE
2.3 SINGLE-FIBRE VERSUS GROSS ENSEMBLE AUDITORY NERVE FIBRE STUDIES
2.4 GROSS ENSEMBLE STUDIES: PREDICTING NEURAL EXCITATION SPREAD INSIDE THE COCHLEA
2.5 TEMPORAL CHARACTERISTICS
3 HUMAN RANVIER NODE MODEL
3.1 INTRODUCTION
3.2 MODEL AND METHODS
3.3 RESULTS
3.4 DISCUSSION
3.5 CONCLUSION
4 GENERALISED HUMAN SENSORY NERVE FIBRE MODEL
4.1 INTRODUCTION
4.2 MODEL AND METHODS
4.3 RESULTS
4.4 DISCUSSION
4.5 CONCLUSION
5 TYPE I HUMAN AUDITORY NERVE FIBRE MODEL
5.1 INTRODUCTION
5.2 MODEL AND METHODS
5.3 RESULTS
5.4 DISCUSSION AND CONCLUSION
6 EVOKED COMPOUND ACTION POTENTIAL WIDTHS
6.1 INTRODUCTION
6.2 MODEL AND METHODS
6.3 RESULTS
6.4 DISCUSSION
6.5 CONCLUSION
7 INFLUENCE OF PULSATILE WAVEFORMS ON THRESHOLD PREDICTIONS
7.1 INTRODUCTION
7.2 MODEL AND METHODS
7.3 RESULTS
7.4 DISCUSSION
7.5 CONCLUSION
8 GENERAL DISCUSSION AND CONCLUSION
8.1 RESEARCH OVERVIEW
8.2 RESULTS AND DISCUSSION
8.3 CONCLUSION AND FUTURE RESEARCH DIRECTIVES
REFERENCES
