2.1 Introduction

It stretches human imagination that in a profoundly deaf person speech perception can be partially re-established by a device implanted by a surgeon that activates fewer than 10 effective electrodes placed over a biological structure containing nearly 30,000 neurons. It is even more astonishing if we consider that these neurons are modulated by multiple other neurons, and that the target region cannot be visually distinguished from surrounding neural structures.

To understand why auditory brainstem implants (ABIs) function at all, we will focus on two major aspects here:

• 
  

  Why do present ABI actually work?

  
  
• 
  

  How can these neurotechnical interfaces be improved based on what is known about the microanatomy and physiology of the auditory system?

The neural processing of two of the basic physical properties of sound—volume and pitch levels—starts as far peripheral as the cochlea itself. Pitch is transferred from a specific frequency into a spatial location along the hair cells of the basilar membrane and their associated neurons.

The brain deciphers pitch by determining which fibers of the cochlear nerve are maximally active at a given point of time.

Volume or loudness is coded in the cochlea by the firing frequency of these primary sensory neurons. The brain interprets volume as a function of both the number of axons firing and their frequency.

The other function of sound perception, localization, requires bilateral input from both ears, and this function cannot be restored with any unilateral hearing prosthetic.

In this chapter, I will discuss the basic morphological and physiological aspects of each anatomical structure in the brainstem portion of the hearing pathway where the electrodes of current ABIs located as well as the adjacent sites of the hearing pathway (Fig. 2.1). I will delineate opportunities and obstacles for the development of effective ABIs presented by these biological specifics.

2.2 The Cochlea and the Cochlear Nerve

While patients to be implanted with an ABI usually do not have a functional cochlear nerve, it remains important to consider this nerve in order to understand the concept of tonotopy. The cochlear nerve relates cochlear tonotopy to the brainstem. The fibers of the cochlear nerve originate from the nerve cell bodies of the spiral ganglion, in the modiolus of the cochlea. The neurons of the spiral ganglion are the first (of four) order neurons between the cochlea and the cerebrum.

The inner and outer hair cells in the spiral organ (of Corti

Alfonso Giacomo Gaspare Corti, Italian anatomist, 1822–1876.

) are the sensory receptor cells for the mechanical impulses that are primarily generated when sounds hit the eardrum and are transmitted via the middle ear structures to the cochlea. A single inner hair cell is innervated by numerous myelinated nerve fibers, whereas a single—unmyelinated—nerve fiber innervates many outer hair cells. A particular region with a number of hair cells which supply input to a particular afferent nerve fiber is considered to be the “receptive field” of this fiber. High frequencies are represented at the base turn of the cochlea while low frequencies have their receptive field at its apex (Fig. 2.2). This arrangement of neurons in the auditory system is “cochleotopic” or “tonotopic.”

There are different types of hair cells corresponding to two different types of auditory nerve fibers: Inner hair cells convey their signals through type I fibers of the cochlear nerve. Type I fibers are myelinated, constitute 90 to 95% of the fibers in the auditory nerve, and are the primary afferent input to most of the cell types in the cochlear nucleus. ^{1 , 2}

Outer hair cells are innervated by type II fibers, which are unmyelinated and project to nonprincipal cell regions of the cochlear nucleus. ^{3 , 4} Responses to sound have not been recorded from type II fibers and it is not clear what role they play. ⁵ Outer hair cells appear to participate in the mechanical response of the basilar membrane; loss of outer hair cells leads to a loss of sensitivity to soft sounds and a decrease in the sharpness of tuning (i.e., the frequency selectivity) of the basilar membrane.

About 95% of the cochlear nerve fibers innervate the inner hair cells. The cochlear nerve trunk exits the modiolus to run into the internal auditory canal (IAC). The peripheral segment of the cochlear nerve joins the vestibular nerve to form the vestibulocochlear nerve (VIII cranial nerve) at the lateral part of the IAC. In cadaver studies, the two portions of the nerve have been found to have a distinct and quite constant spatial relationship: in the IAC the cochlear portion lies anterior to become located inferior to the vestibular nerve in the middle of the nerve and remain there until its entrance into the brainstem. ⁶ In a magnetic resonance (MR) study, the spatial relationship between the cochlear nerve and the superior portion of the vestibular nerve was more variable. In general, the prevailing—cadaver study based—textbook notion that the three nerves rotate around each other needs to be corrected. It is more accurate to describe the cochlear nerve and inferior vestibular nerve as coursing beneath the superior vestibular nerve in either the IAC or cerebellopontine cistern. ⁷

Glial tissue surrounds the nerve in the inner ear canal. Close to its exit, the porus acousticus, the glia encasement changes into by Schwann

Theodor Ambrose Hubert Schwann, German anatomist and physiologist, 1810–1882.

cells (“Obersteiner-Redlich zone”). It is here where most vestibular schwannomas originate.

The primary terminal of the cochlear nerve is the cochlear nucleus in the medulla oblongata. The length of the vestibulocochlear nerve, from the glial–Schwann junction to the brainstem, is 10 to 13 mm in human males and 7 to 10 mm in females. ⁸

The pars cochlearis of the human cochlear nerve contains 32,000 to 41,000 fibers with a diameter of 2 to 3 µm, with nearly all myelinated for fast conduction. ⁹ The fibers reach the ventral cochlear nucleus (VCN) at its ventromedial surface. At the root entry zone, each axon splits into an ascending and a descending fiber. ¹⁰ Ascending fibers run dorsolaterally into the VCN, and descending fibers run caudal and dorsal straight through this portion and diverge into the dorsal cochlear nucleus (DCN). In contrast to other mammals, the human DCN is entirely penetrated by these descending fibers, forming a rather homogeneous plexus with the neurons. ¹⁰ Therefore, the DCN in humans does not have a laminar structure as in the cat. ¹¹

2.3 Cochlear Nuclei

2.4 Human Research

Functional data about the auditory brainstem nuclei in humans are scarce. In the 1970s Dublin attempted to establish a “cochlear nucleogram” to chart tonotopy in the cochlear nuclei. He measured the loss of spheric cells in the VCN and of spiral ganglion hair cells in postmortem specimens and correlated these with the estimated audiogram in sensorineural hearing loss to determine a “best fit” frequency map (“audiohistogram”). His data revealed an intranuclear gradient with low frequencies represented ventral and high frequencies dorsal. ^{19 , 38 , 39} These and other findings basically matched the tonotopic map that was found in animal research. ^{11 , 18 , 40}

The next step in charting tonotopy in humans came with the advent of ABIs when pitch ranking became possible by selectively activating stimulating electrodes along the brainstem surface of the cochlear nucleus. Pitch ranking showed the lower frequencies to be represented more medial (deeper inside the so-called “lateral recess” of the fourth ventricle) and caudal, and higher frequencies more lateral and cranial. ⁴¹

The significance of these data, however, is arguable since the inner (medial) electrodes primarily stimulate the DCN and electrodes located on the outer (lateral) portion of the carriers overlap and may stimulate primarily the VCN. Moreover, the location of the electrodes varies inter-individual and, as we will see, most of the VCN cannot be reached by surface implants at all. In addition, one must bear in mind that the distribution of the electric field around the electrodes is inconsistent, so that it is not easy to calculate at what depth neurons underlying specific electrodes would be excited or inhibited by the electric stimulation. Again, it is notable that threshold-distance measurements were similar in human and animal research. ^{42 , 43} When it comes to charting tonotopy, however, the possibility that perceived pitch will also be a function of depth penetration of the electrical field generated by surface electrodes, is detrimental.

The key point demonstrated by these studies is that axons conveying higher frequencies penetrate to terminals deep inside the VCN while axons representing lower frequencies terminate more superficial. ^{10 , 18 , 19 , 44} In the human DCN, fibers carrying higher frequencies terminate more ventral than fibers carrying lower frequencies. ^{10 , 18}

Another line of functional data on the human cochlear nucleus stems from degeneration studies. Axons in the cochlear nerve have a trophic function for the second-order neuron of the auditory pathway that originates in the cochlear nucleus. It has been shown that following destruction of the cochlea resulting in deafness, neurons in the DCN degenerate much less than those in the VCN. ^{11 , 45} This suggests that most of these neurons do not primarily receive their input from the cochlear nerve.

In a group of deaf patients suffering from sensorineural hearing loss, Jean Moore and colleagues found a 20 to 30% shrinkage of cells in the VCN in postmortem studies. They found the same effect in a deceased patient with a cochlear implant, suggesting that electrical stimulation alone cannot substitute for the trophic function of afferent nerve fibers. ⁴⁶

Interestingly, the authors saw the same cell shrinkage in higher centers of the auditory pathway, namely the olive and the IC. The necessity of trophic input from the cochlear nerve, however, is limited, since 70% of the cochlear nucleus cells do not show signs of degeneration even a long as 10 years after the onset of deafness. Again, data from human and mammalian animal research overlap considerably. ⁴⁷ Still, data from both sources are of limited use for the surgeon tasked to place an implant as close to the auditory brainstem nuclei as possible.

2.5 Anatomy and Surgical Approach

2.6 Intrinsic Anatomy of the Cochlear Nuclei

It has been suggested that the volume of the cochlear nucleus almost triples over the first five decades of life, and then decreases again by about one-third until toward the end of life. ⁵⁹ Seldon and Clark found a larger volume of the human cochlear nucleus on the right side. ⁴⁵ This may be attributed to a preference of the right ear in right-handed people. ^{60 , 61 , 62} The fact that neurons in the hearing cortex of the dominant left hemisphere have been found to be larger than on the right side would be well in line with this finding since a majority of fibers of the auditory pathway cross over to the contralateral side. ⁶³

The inferior cerebellar peduncle, a large neural fiber connection between the pons and the cerebellum, borders on and partially overlaps the most ventral portion of the cochlear nucleus (Fig. 2.4). ^{10 , 11 , 18} The floccular peduncle, which originates from the cerebellar flocculus, forms part of the lateral border of the VCN. From the caudolateral portion to its ventrolateral aspect, the VCN is crossed by myelinated fibers (“peripheral astrocytic border”) of the pontobulbar body. This nucleus is a neuronal relay site with connections to visual and auditory regions of the cerebellum, the trigeminal system, and to the spinal cord. ^{18 , 64 , 65}

For several reasons, the real dimensions of the human cochlear nucleus are difficult to assess. In histological sections, the borders of the VCN and DCN are not clearly delineated. Also, histological fixation ultimately leads to shrinkage of specimen in the range of 10 to 17%, more pronounced in longitudinal than in transversal direction, so that measurements need to be multiplied by a correction factor. ^{46 , 54 , 59 , 66} Until recently, the intrinsic morphological dimensions of the cochlear nucleus in the human brainstem and measurements of protruding structures have only been examined in case studies. ^{18 , 21 , 49 , 53 , 67} On magnetic resonance imaging (MRI) the nucleus is hardly detectable, but its length and width on a horizontal plane have been estimated to be 8 × 3 mm. ⁶⁸

Previous attempts to determine the dimensions of the cochlear nucleus have resulted in varying results. Measured rostrocaudal dimensions have ranged from 2.3 to 4.5 mm. ^{13 , 17 , 45 , 52 , 69} Moore and Osen were the first to investigate the shape and the spatial orientation of the human CNC. ¹⁸ They noted that its rostrocaudal (“up to down”) axis is tilted to the brainstem axis by an angle of 30 to 35 degrees. Other authors attempting to better refine this have come up with sometimes paradoxical results. ^{70 , 71 , 72} Other attempts were made to measure the distance of the nucleus from the surface of the brain over its length. ²⁰ Needless to say, measurement of this complex structure is far from a simple issue.

In order to attempt to synthesize the complex issue of dimension, spatial orientation, and depth of the cochlear nucleus, we undertook a three-dimensional study of 20 brainstem specimens (33 nuclei). ⁵⁴ The most striking finding in this analysis was that the inward rotation of the VCN to the longitudinal axis of the brainstem is so pronounced that the rostral half of the VCN has a distance of several millimeters to the surface of the brainstem (Fig. 2.5).

The overall extension of the CNC as a whole was 8.01 × 1.53 × 3.76 mm (length × width × height) with respect to the intrinsic axis of the complex, with standard deviations between 0.2 and 1.21 mm (Table 2.1). The VCN becomes wider in more rostral (upper) slices of the brainstem and its longitudinal axis (length) was longer in more caudal (lower) sections.

Table 2.1 Means and standard deviations of the maximum dimensions measured in each histological slice of the human cochlear nuclei (mm). The variables were assessed in along the intrinsic axis of the cochlear nuclei

	Length	Width	Height
DCN	3.42 ± 1.21	0.68 ± 0.20	1.90 ± 0.66
VCN	4.59 ± 0.89	1.53 ± 0.64	3.18 ± 0.69
CNC	8.02 ± 1.05	1.53 ± 0.64	3.76 ± 0.89
Abbreviations: CNC, cochlear nucleus complex; DCN, dorsal cochlear nucleus; VCN, ventral cochlear nucleus.

Individual variability of the dimensions of the cochlear nucleus is high. In this study, the smallest and largest values for maximum dimensions and maximum surface depths varied by factor of 3 (Table 2.2). Of note, there was no single CNC that was smallest or largest in all three dimensions.

Table 2.2 Means and standard deviations of the smallest and largest maximum dimensions of the human cochlear nuclei as measured in about 825 histological slices. Because measurements were performed in axial sections, height was calculated by multiplying slice thickness (320 µm) by number of sections on which the nuclei were visible. Therefore, standard deviations are not given for this variable

	Length		Width		Height
	Smallest	Largest	Smallest	Largest	Smallest	Largest
DCN	1.46 ± 1.00	4.24 ± 1.67	0.29 ± 0.19	0.81 ± 0.19	0.77	3.10
VCN	1.92 ± 1.11	5.42 ± 1.75	0.780 ±.31	2.10 ± 0.91	2.32	4.26
CNC	3.38 ± 1.05	9.66 ± 3.42	0.78 ± 0.31	2.10 ± 0.91	2.32	7.36
Abbreviations: CNC, cochlear nucleus complex; DCN, dorsal cochlear nucleus; VCN, ventral cochlear nucleus.

In each CNC, width and length of the VCN and DCN correlated with each other, that is, if the DCN was smaller in these dimensions, the VCN was too. Interestingly, the height of the DCN correlated inversely with the length of the VCN and that of the whole CNC.

In this particular study, there were no significant side differences with respect to length and height of the cochlear nucleus. On the contrary, most data regarding these variables and the surface depth of the nuclei on either side of the brainstem correlated linearly with each other. The only exception was the mean length of the DCN in the sagittal plane: it was greater on the left side than on the right. With advancing age, the height of the CNC was reduced, a finding that correlates to the pattern of degenerative shrinkage without reduction in the number of cells described in the literature. ^{47 , 56}

There are few data on three-dimensional shape of the human cochlear nuclei. At the HEI, Terr and Sinha have produced two models, one demonstrating the CN in relation to the brainstem and the other showing the CN completely dissected out of the brainstem (Fig. 2.6). ⁶⁵ Of course, measures of variance are unobtainable with this method, but still the models helped to design the first ABIs. In our molded three-dimensional renderings of the CNC, it has a distorted X-like or bootlike shape viewed from the side and looks like a wedge from above (Fig. 2.7). ^{54 , 55 , 73}