The Assessment of Electrode-Neuron Interface (ENI) in Cochlear Implant Users

Abstract

The electrode-neuron interface (ENI) plays an important in the outcome with cochlear implants as all cochlear implant-mediated signals should pass through this interface. Telemetry has enabled researchers to study factors affecting the quality and integrity of the interface. These factors may influence electrodes, the path between electrodes and auditory neurons, and spiral ganglion neuron survival. Promising studies on animals regarding such factors have opened new possibilities to identify cochlear implant channels with poor electrode-neuron interface. The results of these studies and similar research on human with focus were discussed in this chapter.

1. Introduction

Cochlear implants (CIs) have been very successful in helping many people with severe to profound hearing loss to hear sounds and understand speech. However, some CI users cannot take enough benefits from these devices. To improve speech perception for all CI users, researchers have been trying to find the causes for such variability in speech perception. The said variability was observed among CI users of different devices manufactured by different companies. Therefore, it was reasonable to look into individual differences among CI users as the causes for the variability in speech perception. For example, research showed that CI users with better performance and speech perception had lost their hearing after language development. It means that they had reliable hearing in their early life (before age 2) [1]. Also, the early detection of hearing loss and less time interval between the detection and treatment were other factors that may contribute to better speech perception after using CIs along with lower age at the time of implantation [1]. The etiology of hearing loss could affect the performance since auditory neurons are significantly impaired in some etiologies of hearing loss more than other [1]. Although these factors can explain some differences in speech perception among CI users, most of the variability remains unexplained. Therefore, other reasons for individual differences should be sought.

Since the electrodes of CIs deliver speech information directly to auditory neurons, it is believed that the integrity of the meeting point of CI electrodes with the spiral ganglion neurons (SGNs) may influence the outcome with CIs. This meeting point is called the electrode-neuron interface (ENI). If the ENI is somehow impaired, the delivery of speech information is expected to be compromised. However, it is necessary to have an overview on how a typical CI works before further discussion about the ENI, which is the main topic of this chapter.

1.1 How does a CI work?

Although there are some differences in the design of CIs manufactured by different CI companies, the basic structure of all CI devices is similar. All CI devices have two parts: an external sound processor and internal receiver-stimulators connected to an electrode array. The external sound processors consist of three components. The first component is a microphone that picks up signals and converts them from acoustic to electric signals. The second component is a digital sound processor that analyzes the electrical signals into separate frequency bands. The signals are then converted into digital format and sent to the third component which is a transmitter. The transmitter is held in place by a magnet and sends the digitized information to the internal receiver-stimulator through radio frequency signals. The receiver-stimulator is a surgically implanted device in the mastoid bone, which is aligned with the transmitter through a magnet. The internal receiver-stimulator is attached to an electrode array and the electrode array is inserted into the scala tympani during surgery. The electrode array has different electrode contacts (up to 22). The receiver-stimulator decodes the received signals from the transmitter and sends these signals to its corresponding electrode contact within the cochlea. Each electrode inside the cochlea receives information about a limited range of frequency. Each electrode contact stimulates its adjacent auditory neurons electrically in a way that basal electrodes stimulate high-frequency SGNs and apical electrodes the low-frequency SGNs. The stimulated neurons convey the information from different electrodes to higher auditory centers and eventually to the auditory cortex for integration and sound perception [2].

1.2 Mode of stimulation

Cochlear implants work according to electricity principles. In electricity, electrical currents should flow in a circuit between two points. The active electrodes are the electrodes inside the cochlea (i.e., the scala tympani) which receive electric signals and stimulate the SGNs. To have a complete circuit, the signal delivered to the active electrodes should flow to a reference electrode. The reference electrode can be inside or outside the cochlea. For example, one of the adjacent electrodes close to the active electrode can be assigned as the reference electrode and can return the electrical current back to the receiver-stimulator. This mode of stimulation in called the bipolar mode (BP). In another mode, the intracochlear electrodes, which do not receive electrical stimulation, can be electronically connected together and return the stimulation delivered to the active electrode to receiver-stimulator. This mode of stimulation is called the common ground (CG). However, if the reference electrode is outside the cochlea, it can also receive the return current and this mode of stimulation is called the monopolar (MP). The path from receiver-stimulator to an active electrode and from an active electrode to its reference electrode and back to the receiver-stimulator forms a channel. Therefore, a channel consists of an active and a reference electrode [2].

The distance between active and reference electrodes in a channel determines the spread of electrical current to auditory neurons and the SGNs. In fact, MP modes produce the lowest thresholds for detection of sounds (T-levels) and comfortable level (C-levels) for speech perception [2]. As a result, the CIs with MP modes use less energy to produce stimulation, and have longer battery life. However, channel interaction (the interaction between the electrical fields produced around each active electrode) is more probable and pronounced with an MP mode [3]. Channel interaction can distort sound perception since this interaction may stimulate different overlapping populations of the SGNs at the same time and may reduce the resolution of stimulation. Despite this problem, MP modes are necessary for higher stimulation rates which are currently used in all CI devices and may result in better speech perception. In contrast, the shorter distance between active and reference electrodes in a BP mode may stimulate a limited neuron region and may lower the possibility of channel interactions but greater current is generally required to reach T- and C-levels. Due to the wider spread of current in a CG mode, this mode is basically similar to an MP mode. However, T- and C-levels in CG modes are usually higher than those levels in MP modes. One advantage of CG modes is that they can detect electrode anomalies which are discussed later in this chapter. One disadvantage of CG modes is that electrical current will flow to all electrodes including those outside the cochlea and could result in nonauditory percepts and it should not be used as a programming mode with CI recipients [2, 3].

1.3 Biphasic pulse

Regardless of the mode of stimulation, all current CI devices use biphasic pulses to stimulate auditory neurons. A biphasic pulse is a symmetrically charge-balanced stimulus. It means that negative (cathodic) and positive (anodic) phases are equal in terms of total charge, so that no net charge remains on auditory neurons after stimulation [3] and auditory neurons and the SGNs are stimulated safely with electrical pulses [3]. Figure 1 shows an example of a typical biphasic pulse. A biphasic pulse has some characteristics. One of them is pulse width which is defined as the duration of each pulse in microseconds. The other one is current amplitude which is the amount of electrical current delivered to an active electrode. For a constant total charge, if pulse width is increased, current amplitude should decrease and vice versa [2, 3].

Figure 1.

Schematic representation of a typical biphasic pulse used for electrical stimulation of auditory neurons.

The total amount of charge delivered to neurons per phase of a biphasic current is equal to the product of current amplitude and pulse width: the greater the charge, the louder the sound. Increasing current amplitude or lengthening pulse width results in more charge and a louder sound. Biphasic pulses are presented in series to auditory neurons. They can be either negative-leading or positive-leading. If the first phase of a biphasic pulse is a negative phase, it is a negative-leading pulse and if the first phase of a biphasic pulse is a positive phase, it is a positive-leading pulse [2]. Figure 2 shows the negative- and positive-leading pulses. By default, the stimulation of auditory neurons occurs during the negative phase of a biphasic pulse in current CIs by default.

Figure 2.

Schematic representation of a negative-leading (A) and a positive-leading pulse (B).

Another characteristic of a biphasic signal is the interphase gap (IPG), which is the time interval between two phases of a biphasic signal. A biphasic pulse can vary from zero to as long as several hundred microoseconds. Different IPGs from 0 to 30 μs are shown schematically in Figure 3.

Figure 3.

Biphasic pulses with different interphase gap (IPG) lengths from zero (left) to 30 μs (right).

2. Factors affecting the ENI

Cochlear implant signals should go through a path from a CI electrode to reach its target neurons. The electrodes, path, and neurons can be impaired and such impairment potentially affects CI outcomes (i.e., speech and sound perception). For example, if CI electrodes are damaged, they cannot transmit sound signals to auditory neurons. Even when the CI electrodes are intact, but the location of the intact electrode is far from auditory neurons and SGNs, enough stimulation might not happen [2]. If an electrode and its location toward neurons are optimal but there are some physical obstacles between the electrode and neurons (such as bone growth and fibrous tissue), the strength of CI signals might be compromised or the signal might be diverted to unsuitable paths and could not reach the targeted neurons. Still when an electrode itself, its location relative to the neurons and the path are in ideal conditions, the auditory neurons and SGNs might be degenerated. Some etiologies or long duration of untreated deafness can degenerate or demyelinate auditory neurons and SGNs, and also reduce the size and number of SGNs survived after hearing loss [4]. Although there was not any significant correlation between the number of SGN numbers and speech perception [5], the quality of their functionality or the presence of auditory neurons’ peripheral processes is an essential factor for proper perception by influencing the ENI [6, 7].

To assess the ENI, a test or batteries of tests are required to reveal information about each of the aforementioned factors. A better understanding of how the ENI influences outcomes will help to reduce damage during implantation, improve technology, and develop biologically fit therapies which can ultimately improve speech and sound perception. For example, if the ENI can be tested in human CI users and especially with clinical instruments, it may become clinically feasible to optimize CI programming parameters in a way that helps with better outcome in users who cannot take enough benefits from CI devices. Telemetry, using a radio frequency connection, has enabled CI devices for bidirectional communication of data. Simply, the external sound processor sends information to the implant and receives information back from the implant [2]. Several measurements take advantage of telemetry and they are now available in clinical software which can provide insights into the ENI. These measurements and recent findings are discussed below.

2.1 Impedance

Impedance testing is one of the applications of telemetry. In impedance testing, the opposition to electrical current flow in CI channels is assessed (i.e., opposition against electrical current flowing from receiver-stimulator to an active electrode and from active electrode to auditory neurons and then to the reference electrode). One way of impedance measurement is that a small electrical current is presented to a single active electrode [8]. Then, the required voltage to make the electrical current pass through the active electrode and travel back to the reference electrode is measured. Since at the reference electrode, both voltage and current values are known, it is possible to measure impedance for the active electrode using Ohm’s law equation [Resistance (R) = Voltage (V)/Current (I)]. Higher the opposition, higher the voltage difference between the active and reference electrodes [8]. When a current is delivered to active electrodes, some factors resist the flow of an electrical current. The size and type of metal-forming wires and electrode contacts, the medium between electrodes and neurons, relative electrode position to the neurons, and the medium between active and reference electrodes all resist the flow of current in series. Therefore, their effects can be added together and presented as what is called the total access resistance. Reactance is another type of impedance that impedes current flow. Similar to a capacitor in which different plates hold positive and negative charges, the electrode surface and fluid/tissue hold opposing charges. The CI electrode surface acts as one of the capacitor plates and the fluid or tissue as another plate. They impede the flow of current by holding and storing electrical energy. Therefore, current can flow through the electrode surface to fluid/tissue with ions’ movement rather than electron movements. The total impedance, in fact, is the vector sum of total access resistance and reactance [8].

Impedance tests are able to identify any break or discontinuity in the path of current from the source to the active electrodes and from active electrodes to the return electrodes in the MP mode. Any kind of break in this path results is very high impedance. This prevents proper neuron stimulation and most of time, they are deactivated by programming audiologists [2]. Another electrode anomaly may happen when two electrode contacts or the electrode wires come in touch with each other and they receive similar currents when only one of them is stimulated. These electrodes are short-circuited and their voltage difference will be close to zero. However, they will have normal impedance in MP modes in which the voltage difference is calculated between an intracochlear electrode and an out of cochlea reference electrode. However, when one of them is assigned to be active and another one to be the reference electrode (i.e., in CG and BP modes), their voltage difference is abnormally small [8].

Even when impedance values are within normal range, it is not guaranteed that the power source can deliver enough voltage to generate the required current levels. There are conditions in which even maximum voltage available from the implant is not enough to generate the required current level on an electrode. These conditions are called out of voltage compliance. The problem of out of voltage compliance may be due to increase in the impedance between electrode and neurons [8]. When this occurs, CI users do not perceive enough loudness with increase in stimulation level or loudness may fluctuate from time to time in line with variation in electrode impedance for a given current level [2].

There is another way that impedance can be measured with. In this way, a single active electrode is stimulated with a small current but the voltage difference is calculated for all combinations of intracochlear and reference electrodes. It means the voltage difference is measured between each intracochlear electrode and the extracochlear reference electrode while only one of the intracochlear electrodes (active electrode) receives stimulation. By definition, impedance is defined as opposition to “flow” of current. Therefore, there should be only measurable impedance (or voltage difference) between active intracochlear and extracochlear reference electrode. In reality, however, when an electrode receives stimulation, an electrical field may develop around it. This electrical field may spread to adjacent nonstimulated electrodes. As a result, there might be a voltage difference between nonstimulated and reference electrodes, even when they do not receive any electrical stimulation. However, impedance values for nonstimulated electrodes should be low compared to that of active electrode. Therefore, it is possible to create a matrix of voltage difference (or impedance) for all electrodes when only one active electrode receives stimulation. This matrix of voltage difference or its correspondence (i.e., impedance matrix) has been used to study current spread within the cochlea under different names (e.g., electrical field imaging and modeling [EFIM] or transimpedance matrix [TIM]) [8]. In addition to current spread, CI stimulation does not remain restricted to the targeted neurons and stimulates other overlapping neurons as well. This type of spread is called spread of neural excitation which is a physiological phenomenon and cannot be measured with impedance or voltage matrix since all types of impedance measurements deal with physical opposition to currents and physical current spread. The spread of neural excitation needs methods in which responses can be collected from stimulated neural population, which is discussed later in this chapter.

2.1.1 Impedance measurements for the assessment of the ENI

Clinically, it is possible to identify electrodes with short- and open-circuit anomalies with impedance testing. As these two anomalies can affect perception, the electrodes with these anomalies can be deactivated by clinicians. Also, electrodes with out of voltage compliance are detectable by current clinical applications. The out of voltage compliance might be a sign that the ENI is compromised [3, 8]. However, impedance assessment and voltage matrix can reveal more information about the ENI. In one study on users of hybrid CI devices (i.e., a CI and hearing aid in the same ear), total impedance and access resistance were observed in two groups of users with stable and unstable residual hearing after implantation. Impedance values increased over time but the reactance component did not change in the group who experienced hearing loss after implantation. However, in the other group who had stable hearing after implantation, total impedance and access resistance were stable while reactance declined. The authors suggested that access resistance might be correlated with intracochlear fibrosis formation and inflammation in the ENI after surgery. The results were supported by animal study as well [9]. In addition, access resistance was reported to be able to detect translocation of the electrode arrays from scala tympani to scala vestibule [10].

Since one of the factors affecting the ENI might be the distance of electrodes from modiolus, impedance measures were found to be significantly higher for the perimodiolar array compared with straight arrays which are placed more distantly from modiolus [11, 12]. However, when the geometric electrode areas in the two arrays were normalized, there was not any difference in impedance values with the distance of electrodes relative to the modiolus. Behaviorally, electrode position within cochlea (lateral-wall versus medial-wall positioning) had significant correlation with audibility thresholds but could not predict speech perception [11, 12, 13]. While some studies reported association between speech perception and impedance results, two recently published studies on a large data set of CI users reject such a relationship [14, 15]. However, the variation in impedance values across the electrode array was significantly related to poorer speech recognition in both quiet and noise [14]. Different studies showed that the presence of obstacles in the ENI (such as fibrous and bone growth) might increase impedance and subsequently decrease cochlear implant function [16]. However, the negative effects of obstacles in the ENI are relatively unimportant in comparison to degeneration of auditory neurons and SGNs [17].

Voltage and TIM measurements have also shown a theoretical potential for evaluating electrode position [18]. However, there was not any strong association between the width of such matrices and behavioral assessments like speech perception in CI users with a full-insertion electrode array [19].

2.2 Electrical compound action potentials and neural health

One of the applications of telemetry is that, the neural responses can be gathered from auditory neurons and SGNs and are called electrical compound action potentials (eCAPs) in CI users. All contemporary CI software have the ability to generate and present electrical pulses to auditory neurons and then record neural responses [8]. These responses are then analyzed with computers and a waveform of the neural responses is depicted. Typically, a biphasic electrical pulse is presented to one active electrode and the neural responses arising from neurons are recorded via another electrode close (but not directly adjacent) to the active electrode. The separation between stimulated electrode (i.e., active electrode) and recording electrode will optimize the response amplitude while minimizing stimulus artifacts. The recorded responses are sent back to a computer via the external speech processor and a programming interface. The eCAP waveform typically consists of a negative deflection (N1), with the latency of 0.2–0.4 ms, followed by a positive peak (P2), with the latency of 0.5–0.7 ms [8].

The eCAP measurement can provide information regarding auditory neurons and SGN status: Three indices are usually calculated from eCAPs: threshold, amplitude-growth function (AGF), and amplitude. Therefore, these indices can be used as a means to assess the quality of the ENI. The threshold of eCAP is the lowest amount of current required for the recording of a measurable neural response. The response amplitude (measured in microvolts between N1 and P2) increases with increasing stimulus intensity. If the values of amplitudes of eCAP are depicted against current levels, an AGF is obtained. Also, the maximum amplitude of eCAP at or close to comfortable level can be calculated.

Since eCAPs are very short-latency responses, they may be obscured by transient electrical artifacts related to electrical pulses or amplifier saturation. To differentiate between real responses from electrical artifacts, the responses for a target stimulus (probe) are subtracted from responses for a masker in a forward-masking paradigm. Simply in this paradigm, pairs of maskers and probes are presented with a time interval and artifacts only affect neural responses to masker. However, the masker drives neurons into refractory period, a period in which auditory neurons physiologically do not respond to electric stimulation. The interval between the masker and probe is adjustable. By increase in this interval, more and more auditory neurons can recover from their refectory period and respond to the probe. More intervals between masker and probe, more contribution from the neurons recovered from refractory and therefore eCAP with larger amplitudes. After a specific interval (around 500 μs), all auditory neurons can respond to the probe and this response to the probe is an artifact-free response. The time interval required for responses from all neurons to probe may vary according to the health and integrity of auditory neurons. In fact, the recovery time represents the neural temporal responsiveness. The eCAPs measured with changing masker-probe interval or with fixed masker-probe interval have been used to study the ENI in several ways. The most relevant component of the ENI which is studied by eCAP is the neural health of auditory neurons and SGNs and the most studied aspect of neural health in human and animal has been the SGN density. The SGN density is quantified by the number of SGN cell bodies divided by the area containing those cell bodies in Rosenthal’s canal [5].

In animal studies, the application of neomycin resulted in the total loss of SGNs along with other structural and functional damages to the cochlea. The amplitude, AGF slope of eCAP, and the latency of N1 were significantly correlated with the amount of SGN density [20, 21]. In guinea pigs receiving CIs, impairment to SGNs significantly lowered the speed of temporal recovery studied with eCAP masker-probe paradigm. In addition, the slopes of AGF declined over the first weeks after implantation and then recovered and became steeper and then stabilized. These changes are similar to the changes observed in behavioral detection measurements and they were attributed to inflammation in the cochlea after surgery. However, eCAP thresholds had poor correlation with behavioral threshold [22, 23] and did not reflect neural health [20].

In human studies, longer duration of hearing loss has been found to be associated with poor neural health [24]. Studies evaluating the association between the eCAP AGF slope and speech perception scores in human CI users show inconsistent results. Whereas, some studies reported better speech perception scores with sleeper slopes [25, 26], other studies found no association between these two measures [27, 28].

In addition, the eCAP AGF slope increased with increasing distance between the electrode and modiolus. A more recent study reported that the eCAP thresholds, but not necessarily eCAP AGF slope, were correlated with electrode-to-modiolus distance [29]. Hearing loss occurs differently in animal and human. In humans, some postmortem studies reported an overall decrease in SGN density in the ears with ossification, while other studies showed good SGN density in the ears with significant ossification [30, 31]. Specific etiologies of hearing loss (e.g., meningitis and sudden hearing losses caused by viral labyrinthitis) are also associated with the accumulation of tissue and bone formation in the scala tympani, even prior to the insertion of the cochlear implant electrode array [30]. One explanation for these contradictory findings could be related to the presence or absence of immune responses along with impairment to the inner ear. Concomitant presence of immune response and cochlear impairment could affect the density of SGNs as well as bone formation inside the cochlea.

2.2.1 ENI and SOE

In the assessment of the ENI, impedance and voltage matrices could give insights regarding the position of electrodes, bony or fibrous tissue formation, and current spread inside the cochlea. However, the spread of physical current per se cannot provide information about the physiological spread of neural excitation. It is assumed that when current spread is wide, more auditory neurons and SGNs in the wider area would be stimulated and current spread and neural excitation spread are correlated. However, studies showed that the spread of neural excitation might not always be reflected in the current spread assessed by voltage or TIM [18, 32]. Therefore, both current spread and spread of neural activation should be measured in the thorough assessment of the ENI, as they both may reduce the neural resolution of CI stimulation and subsequently the perception of sounds.

Electrical compound action potential can be used to provide information regarding the spread of neural activation. Using forward-masking paradigm, an active electrode is stimulated and eCAP amplitudes are measured for that electrode and the electrodes adjacent to it. A function of eCAP amplitudes for active and adjacent electrodes is formed, which may show the spread of excitation (SOE) of auditory neurons and SGNs. In this measurement, the probe electrode is presented to the active electrode and variable maskers to adjacent electrodes [8]. The separation in the location of the probe and maskers may provide eCAP amplitude from different neuron regions around the active electrode. Ideally, the maximum amplitude of eCAP should belong to neuron stimulated by the active electrode with much lower amplitudes from neurons stimulated by adjacent electrodes. However, the overlap between different neuronal populations may create wide SOE function. The width of the SOE function is calculated and reveals information about the amount of overlap between different neuron regions in responding to electrical stimulation (i.e., spread of neural excitation) [2, 8].

Some studies have shown that current spread and spread of neural excitation were not significantly correlated [18]. For example, the results of TIM and SOE were not related, except for a middle electrode [33]. In contrast to early studies which claimed that close distance between electrodes and spiral ganglion would result in a narrow current spread, there was not any significant difference in SOE widths for perimodiolar versus straight electrode array [34]. Although, SOE could help with intraoperative identification of abnormalities of intracochlear electrodes (such as folded electrodes), the computerized tomography (CT) is the ideal test for a precise determination of the position and location of the electrodes within cochlear.

When the relationship of the SOE width and behavioral tests (e.g., speech and pitch discrimination) was studied, there was not any significant correlation between them [35, 36]. However, it should be taken into consideration that speech perception is assessed with stimulation of a wide range of intracochlear electrodes while the SOE width is calculated from the stimulation of few intracochlear electrodes and this might partially justify the lack of correlation between eCAP SOE width and speech perception outcome [35, 36]. Moreover, speech perception is influenced by central auditory processing which is not assessed by the SOE. Therefore, the eCAP SOE function should not be used as the sole objective measure for predicting speech perception in CI users. However, this measure may provide useful information about channel interaction and the possibility of neural dead region which might influence the ENI and CI outcome.

2.2.2 ENI and recovery time

Intuitively, it is believed that faster recovery from refractory period is related to more efficient neural response and better neural health than slower recovery. In fact, faster recovery from refractoriness has been reported to correlate with better speech perception scores in some studies [23, 37, 38]. However, this association was not observed in other studies [28, 39, 40]. Even when recovery time was compared between single and large numbers of neurons, slower recovery from refractory period was associated with greater temporal responsiveness, which was a manifestation of larger neural population and better neural health.

Two phases of refractory period have been identified in studies on auditory neurons and SGNs. The first phase is the absolute refractory period in which the auditory neurons cannot be stimulated even with very intense stimuli. The second phase of the refractory period is called relative refractory period in which auditory neurons can be activated by intense stimuli. However, the speed of recovery from refractoriness is affected by stimulus level, with faster recovery at higher levels [41].

The effect of advanced age on the recovery time from refractory was assessed in postlingually middle-aged adult and elderly CI users using eCAP. There was faster recovery at higher pulse rates and for longer durations of stimulation. There was no significant effect of age on the speed of recovery from refractory period [39, 40].

In another study based on measuring refractory recovery, there were no significant age group differences in refractory times, eCAP thresholds, and N1 latencies. However, the slopes of the eCAP AGF were significantly larger in the middle-aged group compared to the elderly group. Except for N1 latency, electrode location significantly influenced eCAP [39].

Several studies have investigated the refractory properties of the auditory neurons and SGNs in people with special conditions such as the auditory neuropathy spectrum disorder (ANSD) and cochlear nerve deficiency [23]. Children with ANSD had similar refractory recovery time compared with children with typical sensorineural hearing loss [37]. Children with cochlear nerve deficiency had longer absolute refractory time compared with implanted children with normal-sized auditor nerve, but two groups had similar relative refractory [23]. However, the relative refractory period was longer in people with longer duration of hearing loss [39, 42]. There was no difference in the temporal recovery of the auditory nerve between pre- and postlingual CI users [43].

2.2.3 ENI and pulse duration

Animal studies showed that auditory neurons and SGN density were sensitive to changes in parameters of stimulation signals such as increase in pulse duration [20]. Basically, the total amount of charges delivered to auditory neurons will increase when the duration of a biphasic pulse increases. If the auditory neurons are healthy, more current is integrated at the cell membrane and stronger eCAP can be recorded with increase in pulse duration. Therefore, eCAP elicited with different pulse durations has the potential to assess auditory neural health and to be a means for the assessment of the ENI.

The threshold, slope of AGF, and amplitude of eCAP were recorded after changing biphasic pulse duration and interphase gap and the results were correlated with histological and behavioral evidence in guinea pigs receiving CIs. The threshold of eCAP decreased with increase in duration and interphase gap while both amplitude and slope increased. However, the histological findings and eCAP were not correlated [20, 44].

In a human study, the pulse duration was increased in two groups of children using CIs. One group had normal cochlear nerve size and the other group had deficiency in cochlear nerve. The eCAP results showed that increase in duration from 50 to 88 μs did not significantly change cochlear nerve responsiveness to electrical stimulation in either group for a biphasic pulse in MP mode when electrical charge was kept constant [23].

2.2.4 ENI and phase sensitivity

Animal and modeling studies have shown that if the values of eCAP measures are very different for negative- and positive-leading biphasic signals, it theoretically reflects poorer neural health, whereas smaller differences indicate better neural health [45, 46, 47]. In healthy auditory neurons, both negative- and positive-leading pulses activate peripheral processes and initiate spikes at threshold level. At high stimulus level, the negative-leading pulses still stimulate peripheral processes, whereas positive-leading pulses inhibit peripheral processes and generate spikes at cell body.

In cases where peripheral processes are degenerated, the only site that can generate neural spikes is cell body by negative-leading stimuli. Compared with the peripheral axon, the cell body has a much higher threshold, which results in a higher threshold for negative-leading stimuli while the response of the cell body to positive-leading stimuli at high stimulus levels is not affected. As a result, at an equal stimulus level, negative-leading pulses are more effective at generating a neural response from intact human auditory nerve fibers, whereas positive-leading pulses are more effective when peripheral processes are absent or demyelinated [47]. Therefore, comparing the difference in eCAP evoked by two types of pulses may provide useful information about neural survival of auditory nerve fibers [48].

Changing polarity has been used to assess neural health as one of the factors affecting the ENI. They showed at a fixed stimulus level, eCAP evoked by positive-leading biphasic pulses elicited larger amplitudes, shorter latencies, and steeper AGF slopes than those evoked by negative-leading biphasic pulses [48]. These results indicated the possibility of degeneration of the peripheral processes in the deaf ears [49]. However, the association between speech perception capability and polarity sensitivity has not been evaluated in a human CI user. Higher threshold levels were reported for poor neural health in human when they were tested with focused stimulation and the thresholds were not influenced by the position of electrodes inside the cochlea and resistance against current flow [50]. The CI adults with long duration of deafness and poor neural health had better sensitivity to anodic pulses but this was not observed in children using CIs [45].

2.2.5 ENI and IPG change

It is believed that increase in the interphase gap of a biphasic signal allows auditory neurons and SGNs to better recover from depolarization if auditory neurons are in healthy condition. With increase in IPG, lower threshold, steeper AGF slope, and higher amplitude are expected for eCAP. Smaller changes in these measures are observed when auditory neurons and SGNs are degenerated. The IPG effect is thought to be dependent on membrane characteristics, and thus reflect the temporal response properties of the auditory nerve [51]. Given that the IPG effect is a measure performed within the same channel or electrode, it should be less influenced by nonneural conditions that vary across the electrode array (e.g., electrode impedance, fibrous growth, and electrode position) [52]. Compared with fixed IPG, eCAP recorded for different IPG can reveal more information about auditory neuron health.

In animals, the sensitivity to IPG has been shown to be predictive of SGN density. For example, guinea pigs with lower densities of SGNs showed larger IPG effects on the eCAP threshold, amplitude, and N1 latency while in human, the IPG effect was observed for the slope of the AGF [20, 44]. However, IPG effect assessed in animals showed different results from those measured in human CI users.

In a human study on two groups of child CI users with normal-sized and deficient cochlear nerves, the IPG effect showed itself to be in larger amplitude and slope of AGF of eCAP and smaller on the eCAP threshold and N1 latency [53]. However, there was smaller increase in amplitude and slope of AGF eCAP along with higher eCAP threshold in children with nerve deficiency. The effect of increasing IPG from 7 to 30 μs has been studied in another study on CI users. It was shown that increasing IPG generally yielded increased eCAP amplitude and steeper slopes of AGF. However, this effect varied across subjects and electrode locations [24].

It is assumed that SGN density may vary as a function of age at implantation and hearing loss etiology. This assumption was studied on two groups of people who were implanted at early age and adulthood while IPG increased. The group who were implanted at early age had larger changes in eCAP amplitude. However, the AGF slope and eCAP threshold did not differ between the two groups. Irrespective of IPG, child-implanted participants had larger eCAP amplitudes and steeper AGF slopes than the adult-implanted participants. However, vowel recognition performance was not significantly correlated with any of the eCAP measures assessed in this study.

2.2.6 ENI and mode of stimulation

The MP mode creates a relatively broad spread of current. More restricted current spread can be achieved with focused stimulation modes, in which current delivered to an active electrode is returned to the source through “adjacent intracochlear electrodes” or a “combination of adjacent intracochlear and extracochlear reference electrodes.” These focused stimulation modes likely stimulate a more localized region of the auditory neurons and SGNs and can be more sensitive to localized neural degeneration of auditory neurons and reduction in SGNs. Therefore, these focused stimulations have potential for the assessment of neural health and the ENI.

Focused stimulation modes have been used in electrophysiological and behavioral assessment of the ENI. Younger CI users are assumed to have better auditory neurons and SGNs than older CI users. When eCAP was used to capture neural responses to focused stimulation in these two groups, young CI users had steeper eCAP AGF slopes and larger amplitudes and they showed more sensitivity to changes in IPG. In another study, child-implanted CI users had steeper AGF slopes and larger amplitudes compared to adult-implanted CI users. The IPG effect for eCAP amplitude was significant but not for slope or threshold [54].

In another study, behavioral thresholds using focused stimulation were determined across channels. The high- and low-threshold channels in focused stimulation were considered as channels with poor and good ENI, respectively [6]. Using eCAP measurements, the high-threshold channels had higher eCAP thresholds, steeper AGF slopes than low-threshold channels. In another study, subjects with low-focused behavioral thresholds showed large eCAP amplitudes. Higher speech perception scores had significant correlation with low-focused thresholds [55].

Also, the polarity of pulse signals was changed (polarity effect) in focused stimulation studies and the results indicated that the polarity effect was not related to non-neural factors such as electrode-to-modiolus distance, electrode location inside the cochlea, or resistance. In fact, positive polarity effects, which may indicate SGN degeneration, were associated with relatively high focused behavioral thresholds. Overall, these results provide support for the theory that the polarity effect may reflect neural integrity in CI listeners.

2.2.7 ENI and other behavioral tests

In addition to objective measures, some behavioral testing measures have been tried for the assessment of the ENI. One of them is the multipulse integration (MPI) test, in which the difference between thresholds at different rate is calculated. MPI refers to a decrease (improvement) in the psychophysical detection threshold with increasing pulse rate of CI stimulation while the duration of pulse trains remains constant. MPI slopes are calculated as the amount of threshold decrease per increase in pulse rate. This slope is calculated for pulse rates over a range from a few up to about 1000 pulses per second (pps) [56].

Studies on animals have shown that the slope of the MPI function differs between animals with lower and higher SGN densities [17, 57, 58, 59, 60, 61], but only in animals with preserved IHCs [10]. In another behavioral test, temporal integration was assessed with the determination of detection thresholds for constant pulse rates varying in duration. The results showed correlation with cochlear health [58].

The behavioral threshold changes can be measured for changing in pulse phase duration in experiments, which are referred to as strength-duration function tests. The slope (changes in thresholds for changing duration) of the function is considered as a predictor for variations in cochlear health [62]. The characteristics of strength-duration functions can be influenced with the degeneration of the peripheral processes. If these processes are degenerated, the site of neural spike generation is shifted from peripheral neurons to more central axonal regions that have better myelination [63]. It was shown that psychophysical strength-duration function slopes were significantly shallower in guinea pigs implanted in an ear with residual hearing compared with those implanted in a deafened ear that was treated with neurotrophin. In animals deafened with neomycin and treated with neurotrophin, which typically had no surviving inner hair cells, the slopes of the psychophysical strength-duration functions were correlated with spiral ganglion neuron (SGN) density, being steeper in cases with higher SGN densities. However, in animals implanted in a hearing ear, which typically had surviving inner hair cells, slopes of the strength-duration functions were not correlated with SGN density. In addition, the strength-duration slopes were not predictive of duration of any hearing loss or speech recognition performance.

3. Conclusion

Many factors exist that influence the outcome with CI devices by influencing the ENI. Some of them, which could be assessed clinically, were discussed in this chapter. However, there are other factors, such as residual hearing preservation, the type and dimension of electrode arrays and their insertion depth, which are important. In addition, since speech performance needs the processing of sounds in the central auditory system, this factor and cognitive ability of a CI user impact speech perception with CIs. These factors were out of the scope of this chapter, as the focus of this chapter was on the assessments which can be performed with clinically available applications.

There was some discrepancy between animal and human findings, which may stem from different etiologies and pattern of hearing loss in human CI users. There were also differences in the research methods between the two species. It seems that a battery test of current clinical application can be used in the assessment of the ENI. However, the findings of these tests should have well-established association with behavioral findings which needs more controlled studies.