Dr Eshraghi has research grants from NOHR and MEDEL GmbH, he is a consultant for MEDEL GmbH; Dr Balkany is a consultant for Cochlear Corp. and serves on Surgical Advisory Board of Advance Bionic Corp. and MEDEL Corp.; Dr Truy is a consultant for Neurelec Corp.
The cochlear implant (CI) has restored hearing in countless patients around the world. From the first implant designed in 1953 by Andre Djournoto to the more advanced implants that are on the market today, the success of the CI has relied significantly on the advances made in the field of biomedical engineering.
Success and achievements associated with the CI would not have been possible without the close collaboration between engineers and surgeons who are ready to take the risk and use these new technologies for the betterment of human kind.
This article presents the biomedical engineering principles that underlie the development of modern CIs as well as the surgical innovations that have impacted this field.
Through the direct electrical stimulation of the auditory nerve, CIs have provided profoundly deaf people with very effective means to understand speech. Although CIs provide a crude and unnatural input to the brain, these devices enabled hearing disabled adults who had speech understanding before becoming deaf to converse by telephone without having to lip-read. Today's modern CIs use multiple electrodes with highly miniaturized powerful digital processing chips, analog/digital converters, and analog electronic circuits (Zeng et al., 2008). Although CIs have historically used a variety of stimulation modes and processing algorithms, today they are all converged to similar methods of stimulation and speech processing algorithms.
The primary function of the cochlea is to provide the central auditory system with the spectro-temporal information displayed by the streaming speech. This time–frequency display (TFD) information is normally supplied by the cochlea, which acts as a biological time–frequency analyzer and coder. This information display is coded and constrained by the biological, biomechanical, and bioelectrical limitations of the auditory periphery. In CIs, spectro-temporal processing and coding are provided by the highly powerful digital signal processors (DSPs), which can supply more information on the incoming speech than natural cochleae. This information, however, needs to be delivered into the central auditory system with a limited number of electrodes using auditory neural coding.
BIOMEDICAL ENGINEERING OF MODERN COCHLEAR IMPLANTS
The basic components of a modern CI are shown in Fig. 1. The CI is essentially made up of two separate physical components: a wearable processor and the internal unit with its electrode. The external component is primarily composed of the sound input unit (microphone, amplifiers, and filters), information transmission device (RF link), and a speech processor (digital signal processor, DSP). The speech processor in the external unit extracts and encodes the spectral and temporal cues in speech and transmits this information to the internal unit. All these main components of the external unit are powered by a battery (Fig. 1). Internal part and external parts of Advanced Bionics CI, MED-EL CI, and Cochlear CI are shown in Figs. 2–4.
The implanted component is composed of the internal unit, which essentially provides current stimulation and the electrodes which are inserted in the scala tympani of the cochlea. Another important function of the internal unit is to make information available to the outside world on the status of the electrodes. This is achieved by providing a back telemetry subunit.
Wearable and implanted parts communicate with each other using a radio frequency (RF) link across the skin of the scalp. In addition to speech information transmission, an RF link also serves to provide power to the internal unit for digital processing and the current stimulation.
The primary strategy in designing the CI is to mimic the tonotopic frequency coding of the cochlea with multiple electrodes (Fig. 5). First, two formants (F1 and F2) and the fundamental frequency (F0) are the primary cues used in such processing which is accomplished by spectral analysis using band-pass filters. These frequency regions are then mapped to the electrodes implanted in the cochlea.
Frequency-place mapping is an important factor in the performance of implants and could be optimized using individually customized functional tests (Baskent and Shannon, 2003). Another factor to pay attention to during designing the CI is electrical stimulation, as electrical stimulation has a very limited range compared to the natural speech amplitude range and requires extensive compression. This is typically accomplished by rectification of the band-pass filtered signal followed by low-pass filtering and mapping using an amplitude compression function. Final intensity (i.e., loudness) mapping adjustments are typically done individually after implantation.
The final output of the CI is the current stimulation to implanted electrodes which are typically provided by short duration current pulses. Different types of stimulators are available, each working on varying principles. Single stimulators require switching networks to stimulate different electrodes sequentially, while multiple current sources do not need this additional circuitry and provide more efficient operation. Biphasic stimulation is composed of equal negative and positive pulses and is typically used to maximize fiber activation and to minimize toxic and corrosive effects of electrical stimulation. During this biphasic stimulation, pulse duration and pulse separation are generally fixed. Thus, in each electrode, the parameters used to convey speech information are pulse amplitude (A) and rate (1/T).
Although electrical currents provide a very effective stimulation of the auditory nerve fibers, they have one major drawback. Since electrodes are closely spaced, current from one electrode spreads widely to other electrodes and causes electric field interactions. Continuous interleaved sampling (CIS) strategy is generally used to overcome this problem (Wilson et al., 1991). This strategy provides brief current pulses to each electrode such that no overlap occurs and possible electric field interactions are minimized.
CI users often have difficulty in understanding speech in noisy conditions even though they perform fairly well in quiet. This poor performance in noise is mostly attributed to poor perception of pitch or fundamental frequency (F0) of speech information. Typical CIs do not transmit F0 information effectively due to many reasons. In normal hearing, pitch information is primarily coded by the basilar membrane, while place information is provided by the temporal firing rate of the auditory neurons. This information, however, is poorly represented in most CIs due to a limited number of electrodes and the use of fixed or limited varying range of stimulation and the spread of electrical currents. This poor F0 information transmission becomes a severe problem in tonal languages such as Mandarin in which pitch is used to differentiate the meaning of otherwise similar sounding words. This problem is addressed in recent studies which developed specialized processing to convey better pitch information for the users of tonal languages (Lan et al., 2004; Carroll et al., 2011). Such improvements in pitch information transmission will undoubtedly benefit performance of users of monotonal languages such as English under noisy conditions.
NEW RESEARCHES AND SURGICAL INNOVATIONS
Electro-Acoustic Stimulation or Hybrid Devices
The electro-acoustic stimulation (EAS) concept was first published in Europe in 1999 (von Ilberg et al., 1999) as an attempt to preserve residual hearing in patients who had a stable, steeply sloping bilateral high frequency sensorineural hearing loss. These patients were not successful hearing aid users but at the same time, they did not qualify for a CI from an audiometric perspective. The EAS concept is characterized by insertion of a CI electrode into the basal turn of the cochlea, up to about the 1,000 Hz region at the maximum. The CI stimulates the high frequency region of the cochlea, while the built-in hearing aid stimulates the apical region using natural hearing, assuming residual hearing is preserved postoperatively. The first attempts at hearing preservation with insertion of a CI electrode were undertaken with the MED-EL standard electrode inserted part-way to approximately 20 mm, but MED-EL later developed the lower volume FLEXeas electrode for this application. The early European patients wore a MED-EL TEMPO+ processor and an ITE hearing aid in the same ear, but today, study participants are able to use the DUET processor, which combines a hearing aid and audio processor into one unit. The FLEXeas electrode and DUET processor are currently in FDA clinical trials for patients who fit the study criteria at a variety of investigational sites in the USA (Gstoettner et al., 2009).
The development of the hybrid S (i.e., a short electrode array) began in 1996 at the University of Iowa in collaboration with Cochlear Corporation. The clinical trial started in 1999 with two shorter electrodes than the one used by MED-EL for their EAS study. Their initial study implanted a 6 mm and a 10 mm hybrid electrode. They found that patients were overall performing better in background noise and musical instrument recognition that requires more refined spectrum information (Gantz et al., 2010). They also concluded that experience with tonotopic mismatch is important for giving patients time to adapt, and that pitch perception is plastic and may be adaptable with experience. It is important to note that even though hybrid device implantation is successful in the majority patients, ten percent of patients in a study by Woodson et al. (2010) and 30% patients in a study by Fitzgerald et al. (2008) eventually lost all residual hearing.
To understand better the mechanism involved in loss of residual hearing, the pattern of hearing loss caused by electrode insertion trauma in animal models was evaluated recently. This loss of hearing has both an acute component (direct trauma) and a delayed component (cellular and molecular damage) that develops over the period of at least a week following the initial trauma event (Eshraghi et al., 2006). Eshraghi and coworkers found that insertion of a CI electrode array causes direct tissue trauma and cell losses via necrosis, but also generates major molecular events that will contribute further to a loss of residual hearing by programmed cell death, or apoptosis (Eshraghi, 2006; Eshraghi et al., 2011).
These observations have led investigators to recognize that even if there is no detectable macroscopic damage (Grade 0 on the Eshraghi trauma grading system) (Table 1) (Eshraghi et al., 2003), cochlear damage can still be detected on both the cellular and molecular levels. This cellular and molecular damage may result in a loss of residual hearing in some patients. Degree of trauma, genetic susceptibility, inflammatory processes, immunological reaction to the foreign body (the implant), and other yet unknown factors may explain the variability of outcomes within the patient population undergoing implantation with EAS or hybrid devices (Fig. 6).
Table 1. Grading system for cochlear implant electrode insertion trauma
Reproduced with permission from Eshraghi et al., 2003.
Possible damage at the molecular level that can lead to programmed cell death (ie, apoptosis).
Fracture of osseous spiral lamina or modiolus, or tear in tissues of stria vascularis/spiral ligament complex.
Recent studies using animal models have evaluated the otoprotective effect of mild hypothermia (Balkany et al., 2005; Eshraghi et al., 2005) and several drugs such as dexamethasone (DXM) and AM-111. These innovative approaches are under laboratory investigation at this time with the goal of preserving the residual hearing after electrode implantation, very important in case of electro acoustic (bimodal) stimulation.
Targeted local therapy of the inner ear has been developed for the treatment of idiopathic sensory hearing loss (Rauch, 2004; Gouveris et al., 2005; Plontke et al., 2008, 2009) and Meniere's disease (Cohen-Kerem et al., 2004) to avoid the side-effects that can result from systemic delivery. Several experimental studies have demonstrated that targeted delivery of steroids via the fluid space of the scala tympani can also protect against electrode insertion trauma (EIT)-induced hearing loss after cochlear implantation. Protection was demonstrated experimentally with a bolus injection of a steroid immediately following a cochleostomy (James et al., 2008) and with a single delivery to the cochlea prior to insertion of the CI (Salt and Plontke, 2009).
Recently developed methods for sampling perilymph from the cochlea have shown that drug distribution in perilymph is dominated by passive diffusion, resulting in a concentration gradient along the scala tympani of the cochlea when the drug is delivered via the intratympanic route into the middle ear cavity. Intratympanic applied drugs are thought to enter the fluids of the inner ear primarily by crossing the round window (RW) membrane. To increase the amount of drug reaching the middle and apical turns of the cochlea, controlled-release drug delivery systems (e.g., biodegradable biopolymers or catheters with a drug delivery pump) are being developed and perfected (Suckfuell et al., 2007; Paulson et al., 2008; Eshraghi et al., 2011). DXM was delivered to the scala tympani through microcatheters which were attached to miniosmotic pumps and were reported to conserve auditory function thresholds after electrode insertion trauma (Eshraghi et al., 2007a; Vivero et al., 2008). Intratympanic injection of gel-based DXM was introduced by Wang et al. (2009) as another method of DXM delivery. Using this technique, drug levels in the perilymph were maintained for at least 10 days. RW membrane injection of chitosan glycerophosphate (CGP)-Dex-hydrogel has also been studied as a potential delivery method (Paulson et al., 2008).
An inhibitor of the c-Jun-N-terminal peptide (JNK) pathway, i.e. D-JNK-1 (D-c-Jun-terminal peptide inhibitor one) peptide now known as AM-111 (Auris Medicus compound no. 111), was also tested recently in an animal model of CI insertion trauma. Animals were treated immediately post-trauma with locally administered D-JNKI-1 (AM-111) through a catheter/osmotic pump delivery system into the scala tympani for 1 week (Eshraghi et al., 2006). They observed that D-JNKI-1 (AM-111) treatment of animals undergoing electrode insertion trauma (EIT) prevented EIT induced elevation of ABR thresholds and this D-JNK-1 treatment conserved hearing and was stable at 2 months post EIT (Fig. 7) (Eshraghi et al., 2007). The use of an osmotic pump for 1 week post-cochlear implantation does not appear appropriate for use in a clinical setting as it might increase the risk of infection of the labyrinth and the possibility of meningitis. Recent reports have demonstrated that hyaluronic-acid-based hydrogels are efficient, stable, and sustainable vehicles for the local delivery of drugs (e.g. DXM) into the perilymphatic space of the scala tympani via the RW membrane (Forge and Li, 2000). A recent study (Eshraghi et al., 2012) utilizes a hydrogel based on sterile sodium hyaluronate for local delivery of AM-111 to the RW membrane niche 30 min before electrode insertion. The AM-111 pre-EIT treatment of the cochlea via the RW membrane/hydrogel was as effective as the intrascalar osmotic pump delivery of the peptide D-JNKI-1 (AM-111). There were no significant increases in hearing thresholds in EIT/RW Membrane hydrogel/AM-111 pre-treated cochleae. This preliminary study also showed a difference in the pathway involved in the cell death of outer hair cells and inner hair cells, which is under further investigation at this time. Both DXM and AM-111 via local treatment have excellent protective capabilities in animal models of EIT-induced hearing loss. The next step for these translational research studies is to test their efficacy in human clinical trials.
The Digisonic Binaural Cochlear Implant
In 2008, the William House Cochlear Implant Study Group critically examined putative additional benefits of bilateral implantation. The pertinent literature was reviewed in a committee of over 250 CI specialists, and a position statement was written to strongly endorse bilateral CIs in clinically appropriate adults and children as accepted medical practice. In this article, we present and discuss a new innovation that may impact the future of bilateral implantation
The majority of patients today are implanted unilaterally; however monaural stimulation in patients with unilateral implants deprives them of important bilateral auditory cues. We do know the central consequences of unilateral auditory privation, and also know that binaural auditory stimulation has a beneficial action on auditory structures in cases of congenital or long-lasting deafness (Bilecen et al., 2000).
Bilateral electrical stimulation has been proposed historically as presenting theoretical advantages: access to spatial localization (van Hoesel and Clark, 1997), and enhancement of speech recognition performance, especially in noisy situations (McCullough and Abbas, 1992). Bilateral CIs increase the number of stimulated channels, that theoretically improve the frequency selectivity, and therefore the patient's performance; but unfortunately, the number of differentiated channels in an electrically stimulated ear is limited to 10–12, on the average (Fu et al., 1998; Loizou et al., 1999). This can be explained by the electrical interactions linked to the spreading of the charges, by the capacitive effects, and also by the complexity of the spiral ganglion tonotopy (Shannon, 1983). To stimulate both ears would allow one in theory to double the number of differentiated channels and therefore the frequency selectivity (Cokely and Hall, 1991). Some authors pointed that binaural cues have a central processing (Gooler et al., 1996), suggesting that two separated devices might not be the best way to access binaural function.
Bilateral implantation offers stereophonic perception that allows sound localization and which may provide improvement of speech perception in quiet and in noisy environments. The literature is consistent with these ideas (Litovsky et al., 2004; Neuman et al., 2007; Schoen et al., 2005; van Deun et al., 2010). Speech perception benefits are mainly provided by the head-shadow effect (monaural benefit observed by adding a second implant on the side of the better signal-to-noise ratio) and a larger summation effect (effect attributed to the increased loudness associated with bilateral stimuli and to the redundancy of information in the stimuli provided in both ears).
General information given in current literature is that bilateral is not binaural: A bilateral CI does not give all the information necessary to permit patients' central binaural processing. Limitations of bilateral CIs are in summary: (i) two monaural implants that are fit separately, (ii) processing schemes distort intensity and timing cues, (iii) two implants function independently so they lack time-dependent synchronization of normal hearing listeners, (iv) processors discard low frequency information needed for binaural system, (v) they mismatch in frequency representation in the cochlea (Blamey et al., 1995, 1996; Clark, 2006; Wilson and Dorman, 2008).
The cost of bilateral electrical stimulation of deaf ears is challenging. Unilateral CIs have been proven to be cost-effective in congenitally deaf children (Tajudeen et al., 2010) and in deafened adults as well. The initial cost, including devices, surgery, and care, is important in relation to the expected efficiency. Furthermore, in many countries it is impossible to propose two implants to a patient when other deaf people will not be able to benefit from monaural cochlear implantation for financial reasons. Bilateral implants funding is very restrictive in adults and/or in children in a vast majority of countries, even in highly developed ones; restrictions having been validated by different national Health Care Agencies. These agencies concluded that it is not possible to recommend bilateral cochlear implantation in adults as a cost-effective use of NHS resources. In children, with a 40% discount on the cost of a second implant, bilateral simultaneous cochlear implantation is a cost-effective use of NHS resources.
The aims of the developed binaural device by MXM (Vallauris Cedex, France) were to provide access to real binaural integration of surrounding sounds, for about the same cost as a monaural device. To improve the checking and the balance of the signal sent to the two ears, a single processor piloting the two implants was proposed by Lawson et al. (1998).
The Binaural Digisonic CI (Vallauris Cedex, France) is a multi-channel commercially available device (Figs. 8 and 9). This implant includes a single external processor connected to two microphones placed on both ears, and to a single external antenna that transmits, by electromagnetic coupling, the acoustic signal processed by the processor to a single receptor-stimulator implanted under the skin behind the ear. This receptor-stimulator is connected to two electrode arrays, one in the ipsilateral cochlea and a second in the contralateral cochlea. The contralateral electrode array is specially designed to be introduced under the scalp from the receptor-stimulator to the contralateral cochlea via the vertex. Due to the growth of the skull in children this device is not indicated in children. Scalp disorders such as eczema, psoriasis, previous surgery, trauma, or irradiation are contraindications. The surgery was first described by Truy et al. (2002).
The summation effect provided by the binaural device was found to be 14%. This benefit is situated in the upper part of the range reported in the literature with bilateral CIs: 5.7% by Buss et al. (2008) using sentences and adaptive signal to noise ratio (SNR). The reported head-shadow effect was smaller (mean 12%) than values in bilateral implantations reported at 38% by Buss et al. (2008), and at 20% by Muller et al. (2002). This could be explained by a ceiling effect due to a fixed signal to noise ratio (SNR), characteristics of populations, and different testing materials. Computation of squelch effect, with a mean value of 14%, was comparable or slightly better to those reported with bilateral implants [11% in Buss et al. (2008)]. One hypothesis for this 14% squelch effect could be an improved binaural processing. Hypothesis in favor of binaural implants are: (i) information redundancy given by a device combining two complimentary speech inputs across ears, (ii) stereophonic processing performed by the speech processor sending simultaneously a pseudo-synchronized stimulation to each ear. Contrary to bilateral CIs, the binaural device does not introduce any left/right temporal mismatch. Future studies would get deeper in refinement analysis using finer localization testing, Interaural time differences (ITD) sensitivity, and other methods.
Improving Surgical Technique
Surgical techniques for cochlear implantation have evolved since the pioneering work of William House in the 1960s. Some milestones along the way include surgical techniques for implantation of patients with cochlear dysplasia, ossified cochleae, and early attempts to reduce electrode insertion trauma for the conservation of a patient's residual hearing.
Other techniques have also been improved: As mentioned above, the RW insertion technique is used for favorable cases to avoid trauma related to drilling a cochleostomy and to assure the retention of the electrode within the scala tympani.
While initially patients were undergoing a significant scalp shaving before undergoing a large C-shaped incision, current methods advocate a very limited shaving and the trend is to use a small retroauricular incision.
One of the more recent surgical innovations is an anatomically based method of fixation of the receiver stimulator without drilling bone (Balkany et al., 2009). The receiver/stimulators (R/S) are generally secured by drilling a custom-fit seat and suture retaining holes in the patient's skull. However, rare intracranial complications, including death, due to drilling the skull, as well as R/S device migration have been reported with the use of this standard method. In view of the various problems encountered with the traditional methods of securing CIs, surgeons have tried many alternative methods to secure CIs to skull (Djalilian et al., 2001). Roland reviewed Cochlear Corporation data in 1998 and found 22 cases of R/S migration, 13 in adults and 9 in children (Roland, 2000). More recently R/S migration has been reported in the FDA's Adverse Event Report section of the Manufacturer and User Facility Device Experience (MAUDE). Six of the 100 most recently reported adverse events were caused by R/S migration. This number excludes cases of trauma, infection, and electrode migration. In an anatomic study for the development of the Balkany pocket technique (Balkany et al., 2009) 48 formalin fixed half-heads were studied. In eight, a CI R/S dummy was inserted. A 3–4 cm incision was made and a 3-cm long, an anterior based flap was elevated. Next, a sub-pericranial pocket was made that was deep enough to insert the entire R/S and this pocket created was designed for a very tight fit. Only the electrode cables were seen exiting the pocket. The anterior margin of the pocket was 3 cm posterior to the post-auricular crease. The orientation of the pocket was approximately 45 degrees from horizontal (Fig. 10). This approach termed the temporalis pocket technique (Balkany et al., 2009) is now being tried and evaluated by other CIs surgeons.