Robot-Assisted Cochlear Implant Surgery

Abstract

Since the inception of cochlear implantation, there has already been a lot of research into improving its technological aspects, whereas the surgical placement has enjoyed a golden standard for a long time. Since the advent of robotic surgery, there has now also been the development of robot-assisted cochlear implant surgery. This chapter will discuss the opportunities and challenges that robotic-assisted and image-guided cochlear implantation faces. The required accuracy and sensitivity to not harm inner ear structures during electrode insertion is already at the limits of human dexterity. With electrode arrays becoming smaller in the future, the need for robotic accuracy and reliability will become necessary. Robotic-assisted cochlear implantation is seen as a minimally invasive way of doing cochlear implantation surgery with the potential of being the golden standard in the future. An atraumatic intracochlear electrode array placement ensures that the anatomy and physiology of the inner ear structures are preserved as much as possible, thus reducing the risk of losing the rest of the natural hearing levels of the patient. This could lead to a broadening of the indication, opening the door for patients that only experience a loss at the higher frequencies. It is a given fact that robotising surgical procedures will standardise surgical outcomes.

1. Introduction

The implementation of robotics in cochlear implantation surgery was a step-by-step process. During the last 15 years, a lot of research went into the various components that could make robotic cochlear implant surgery a reality. Different aspects, such as designing the robot and developing safety mechanisms, were studied. It was only in 2017 that the first study was published that had combined all these different components into one functional robot [1]. Cochlear implant surgery consists of three main steps: middle ear access, inner ear access and electrode array insertion. The end goal of the implementation of robotics in cochlear implantation surgery is a fully robotic automation of these three steps. Autonomic middle ear access was the first hurdle that was overcome. Autonomic middle ear access was performed in six cases. The next hurdle of cochlear implantation surgery was autonomic inner ear access. This hurdle was overcome in a recent study that was able to perform inner ear access in 22 out of 25 patients [2]. The last step, autonomic electrode array insertion, is a hurdle yet to be overcome. When all the hurdles are overcome, fully autonomous robotic cochlear implantation surgery will become a reality.

A key question being asked is why we should aim for a full robotic automation of cochlear implantation surgery. Conventional cochlear implantation surgery is already well established, in which the risk of complications is minimised under an experienced surgeon. The facial recess approach, widely seen as the golden standard, has consistently achieved a facial nerve injury rate of less than 1% [3]. Still, with conventional cochlear implantation surgery, an atraumatic electrode array insertion into the scala tympani of the cochlea cannot be ensured. Stereocilia in the cochlea are very susceptible to damage. Robotic cochlear implantation surgery could predict and minimise the trauma during electrode array insertion thus preserving the residual hearing of the patient, widening the indication of surgery. Robotic cochlear implantation surgery would also have a great impact on healthcare costs and the time under general anaesthesia. The conventional method of cochlear implantation surgery can be within 1 hour within the hands of an experienced surgeon. Robotic cochlear implantation surgery would take less than 10 minutes for inner ear access without the safety protocols that are currently in place. With the further development of robotic cochlear implantation surgery, it would potentially become an outpatient procedure under local anaesthesia.

2. Robot-assisted cochlear implant surgery

Robot-assisted cochlear implant surgery (RACIS) would not have been able to develop without the advances made in image-guidance technology. In 2014, a study was published that sought to reach the round window in a minimally invasive way [4]. The ideal insertion into the inner ear is defined as an insertion through the round window with the trajectory being parallel to the basal turn of the cochlea as long as possible, in order to not hurt any inner ear structures. If one would like to follow this definition of the most efficient path to the round window, one would have to drill through the facial nerve. The ideal drilling trajectory is by definition deviating from the optimal trajectory. Clinical research though has to show that perhaps another definition is better because, inevitably, every trajectory is bound to hit the lateral wall of the inner ear. It may be shown that trajectories that are hitting the lateral wall more basal could have perhaps a better preservation of the apical structures. Currently, these data are not available.

This shows the importance of preoperative imaging. Using preoperative imaging, an attempt was made to create a trajectory to the round window where there is a safe distance between the drill and critical structures. The aim was an insertion through the round window with the trajectory being parallel to the basal turn of the cochlea as long as possible. This would, therefore, allow electrode insertion to be easier and less traumatic. Labadie et al. reported a study with nine patients where they had managed to implant eight patients [4]. In six of these patients, the electrode arrays were completely in the scala tympani of the cochlea. The electrode arrays were placed through a cochleostomy after lifting a tympanomeatal flap. During the study, there were some complications such as the tip folding-over and facial nerve injury because of the heat created during drilling. In one patient, a switch to the conventional method was necessary because of difficulties met during electrode array insertion. A frame was used during drilling. This way, deviations of the trajectory could be led to a minimum. Intraoperatively, two CT scans were taken for verifying if the intraoperative trajectory was not deviating from the planned trajectory. A second safety mechanism used during drilling was the use of intraoperative facial nerve monitoring, which is also used in the conventional method of cochlear implantation. Post-operatively, a final CT scan followed to verify if the electrode array was positioned correctly.

In 2017, the first system, termed the OtoBot system and surgical workflow for robot-assisted cochlear implant surgery, was developed [5]. The workflow is described as follows. After the indication for cochlear implantation is made, a screening CT is done. In current safety protocols, a risk mitigation of facial nerve palsy is defined as the distance between the planned trajectory and the course of the facial nerve being three standard deviations larger than the accuracy of the system. Caversaccio et al. reported the first study in men using the OtoBot system for robotic middle ear access for cochlear implantation. A total of nine patients were included in the study. The procedure had to be converted to conventional surgery in three out of the nine patients because of safety reasons. In one of the six patients where middle ear access was performed, the insertion of the electrode array in the scala tympani was not possible because of a plaque formation in the cochlear basal turn in a patient with Cogan syndrome. A complete insertion of the electrode array in the scala tympani of the cochlea was reported in two out of the six patients [6].

The patient’s anatomy has to be screened for suitability for the robot to drill a safe trajectory. Using OTOPLAN®, a dedicated planning software, a 3D reconstruction of the images and a simulation of the trajectory are made. After the anaesthesia and immobilisation of the patient, several fiducial screws are placed with imaging again after this. A simulation of the trajectory is again made after this. Once the trajectory is confirmed, robotic drilling starts in several steps. Between these different steps, imaging and facial nerve monitoring are used to check whether one is on the right trajectory. The drilling ends when an access to the middle ear is made.

2.1 Preoperative imaging and planning

A screening CT will confirm whether the patient’s anatomy allows a safe route to the round window. Using a software programme, such as OTOPLAN®, a 3D reconstruction of the anatomy is made. All the anatomical parameters, such as the size of the facial recess, cochlear length and angles of cochlear approach, are segmented. The facial recess is the distance between the facial nerve and the chorda tympani. Using OTOPLAN®, a simulation of the trajectory is made (see Figure 1). The average size of the facial recess is 2.54 ± 0.5 mm. In this space, a drill of 1.8-mm diameter has to pass at a safe distance from both the facial nerve and the chorda tympani [7]. Currently, for the risk mitigation of facial nerve palsy, a minimum distance of 0.4 mm from the facial nerve is used. This distance is defined as the distance between the planned trajectory and the course of the facial nerve being three standard deviations larger than the accuracy of the system. Ansó et al. reported that facial nerve monitoring was able to prevent the structural damage of the facial nerve at this distance with great sensitivity and specificity [7]. They have shown that a customised facial nerve monitoring device using both the active mono- and bipolar stimulations was able to guide drilling through the facial recess safely in an animal model trial. Ansó et al. also assessed the performance of the customised facial nerve monitoring device in smaller distances between 0.1 and 0.4 mm in animal experiments [8]. The study concluded that between these respective distances, facial nerve monitoring was also able to avoid structural damage. All the trajectories with distances greater than 0.1 mm were correctly classified as being uncritical to continue drilling. In distances closer than 0.1 mm, however, facial nerve monitoring was able to assess a correct distance in only four out of the seven trajectories. This study only assessed for structural facial nerve damage and did not investigate the post-operative functional status.

Figure 1.

A simulation of the optimal robotic drilling trajectory using planning software (OTOPLAN).

RACIS starts with placing the head of the patient in an atraumatic head rest. Pinning the patient’s skull as in certain neurosurgical procedures is, therefore, not necessary. After a conventional retro-auricular incision, the placement of about four fiducial screws on the surface of the mastoid follows. These will serve as artificial landmarks on imaging and are important for the registration of the images. One additional fiducial is placed to hold a patient marker. Registration comprises the robot with a fused image of both the preoperative images and the working surface in which the surgeon has to operate. This is visualised with a camera. It also verifies if the planned trajectory meets the safety settings towards critical structures. After a safe trajectory is reconfirmed on OTOPLAN®, the operation can follow.

2.2 Autonomous middle ear access

The surgical robot currently used RACIS is the HEARO® robotic system developed by CASCINATION in Switzerland (see Figure 2). The accuracy and precision was described as 0.1 ± 0.05 mm when verified with a laser scanner and 0.15 ± 0.08 mm when verified on human cadavers [1].

Figure 2.

The HEARO® robotic system. (1) robot mount, (2) headrest, (3) patient marker attachment, (4) patient marker, (5) drill and (6) drill mount with force/torque sensor [2].

The very first step of autonomous middle ear approach is the robotic drilling from the surface of the mastoid to around 3 mm in front of the estimated depth of the facial nerve. This step is done in 2-mm intervals where it is important to properly clean the drilled cavity between each step with saline water. The importance of flushing and working in intervals is necessary to avoid excess heat during drilling. Drilling is stopped 3 mm in front of the estimated depth of the facial nerve, and the first intraoperative CT is performed to evaluate the distance of the drilled trajectory but also to calculate any inaccuracies. Here, it is important that a titanium rod is placed inside the drilled cavity. This will ensure that one will be able to differentiate the drilled trajectory from the surrounding anatomy more easily. The drilled trajectory is compared with the planned trajectory during this step, and the decision to continue drilling or not is made. Intraoperative imaging is done with a mobile cone-beam CT that has a spatial resolution of at least 0.1 mm.

The second step of autonomous middle ear approach is the most critical step, drilling in the facial recess between the facial nerve and the chorda tympani. This involves drilling in 0.5-mm intervals until just beyond the facial nerve. It is crucial to maintain a minimum distance of 0.4 mm from the facial nerve and 0.3 mm from the chorda tympani during this step. Between each interval, it is necessary to confirm the safety of the drilled trajectory with neuromonitoring what will be repeated five times in the course of the second step [4].

The third step of the operation is the fastest and least critical step. Drilling will continue until one reaches the middle ear. In this step, there is the least risk of damaging critical structures (see Figure 3).

Figure 3.

The HEARO®-robotic system drilling through the mastoid with a 1.8-mm drill performing middle ear access. Facial nerve (yellow) and Chorda tympani (red) [2].

2.3 Autonomous inner ear access

Autonomous inner ear access is the next step after autonomous middle ear access and is performed with a 1.0-mm burr instead of a 1.8-mm drill. Topsakal et al. reported successful minimally invasive autonomous inner ear access in 22 out of 25 patients [2]. All the 22 cases were performed safely with no reports of complications. RACIS was aborted in three out of the 25 cases during autonomous middle ear access, after the first intraoperative CT was taken. These three cases were converted to conventional cochlear implantation surgery because either a sufficient intraoperative accuracy could not be guaranteed due to software failure (two cases) or the evaluation of the trajectory was not determined to be safe (one case).

The process of autonomous inner ear access is the milling of the bony overhang of the round window, also referred to as the canonus fossulae fenestrae cochlea or canonus in short [9]. It covers the round window for a part and serves as a protection of the round window. This access is required to be able to pass a 0.8-mm-diameter electrode array through the 1-mm-diameter canonostomy (see Figure 4), whereas in conventional surgery, a full canonectomy needs to be performed to have visual confirmation of the round window and, if possible, the course of the basal turn. A minimal invasive inner ear access and electrode insertion can be ensured with this procedure being image guided. The thickness of the canonus is calculated using preoperative imaging. Both its lateral and medial limits are also defined. The position of the mill during the milling process can always be calculated intraoperatively by using a force-torque sensor that is connected to the robotic arm that is performing the milling process.

Figure 4.

An endoscopic view of the canonus (A), the round window (B) and a partial canonectomy (C) [2].

The process of autonomous inner access is divided into four phases (see Figures 5 and 6). Phase 1 or the touchdown phase is when the drill comes into contact with the lateral wall of the canonus. Here, the force-torque sensor will measure a sudden increase in force because of a sudden increase in resistance. Phase 2 or the plateau phase is when the drill is fully positioned inside the canonus. Here, the resistance remains the same, and the force-torque sensor will always measure the same force. Phase 3 or the breakthrough phase starts when the drill has reached the medial wall of the canonus. Because of a decrease in the thickness of the bone, there will also be a decrease in the resistance that has to be overcome by the drill bit. This is also illustrated by a decrease in force. In phase 4 or the enlarging phase, the drill must still drill 0.3 mm beyond the edge of the medial wall. As a result, the diameter of the canonostomy will be equal to 1.0 mm throughout its trajectory, and the electrode array with a diameter of 0.8 mm will be able to be inserted smoothly.

Figure 5.

Inner ear access algorithm [2].

Figure 6.

Intraoperative use of the inner ear access algorithm describing the force and the depth during canonostomy; the lateral wall (LW) and the medial wall (MW).

2.4 Last step: Autonomous electrode placement

Although there are currently already some telemanipulators that can assist the insertion of an electrode array at a certain pace in standardised feed rates, such as the RobOtol® system [10] and the iotaSOFT® system [11], none of these warrants a standardised insertion angle towards the inner ear.

In this definition, an autonomous electrode placement has not yet been realised in clinical practice. In the HEARO procedure, this step is still done manually by the surgeon (see Figure 6). Because of the minimal invasiveness of RACIS, visualising the electrode array during insertion in the drilled trajectory is not possible. To make insertion possible, the surgeon will try to visualise the round window via the external auditory canal and by folding over the tympanic membrane. This way, a visualisation of the electrode array during insertion is made possible. Opening the tympanomeatal flap goes against the principle of RACIS being a minimally invasive procedure. It is, therefore, important to develop a technique that allows the autonomous placement of the electrode array. After a series of 30 cases, Topsakal et al. managed to fine-tune the insertion angles and round window approach to develop in 2022 a new clinical routine for inserting electrodes without opening the tympanomeatal flap. The design of a device that is capable of inserting the electrode at an automated speed would not be too difficult. Calibrating and defining the insertion angles before the approval of a surgeon would expect a learning curve.

Topsakal et al. reported a successful and complete insertion of the electrode array in 21 out of the 22 cases. Twenty-five patients were included in the study, but three had to be converted to conventional surgery because of safety protocols. In one case out of the 22 cases, the angle of insertion proved to be problematic with the last contact of the electrode (C12) being localised in the position of the round window [2]. In three out of the 22 cases, electrode array insertion was not possible from the first time. Because of a difficult angle of insertion in two cases, it was decided to enlarge the round window approach. In one case, an intracochlear ossification in a patient with Cogan syndrome hindered a smooth insertion. Here, it was decided to widen the inner ear space. A successful electrode insertion was eventually possible in all the three patients. During electrode array insertion, a guiding tube is used to avoid a false route to aerated mastoid cells (Figure 7). The guiding tube exits out of two half pieces allowing for an easy removal after electrode array insertion. The guiding tube still allows the surgeon to estimate the depth of electrode insertion. All the electrode array insertions in this study were done with a 28-mm straight flexible lateral wall electrode (Med-EL, Innsbruck, Austria). The guiding tubes allow for a quick and easy electrode insertion with these flexible electrode arrays. Post-operatively, a final cone-beam CT follows. A correct position of the electrode is checked this way. Electrophysiological tests are also used to check if the electrode array is fully inserted.

Figure 7.

Manual insertion of the electrode array using a guiding tube.

Rajan et al. reported in a study of 40 patients that a slower speed of electrode array insertion leads to a better preservation of the patient’s residual hearing and vestibular function [12]. Here, two speeds of insertion were compared, namely, a speed of 15 and 60 mm/min. The mean loss at low frequencies in dB was 16 ± 1.75 in the high-speed insertion group and 10.5 ± 2.5 in the low-speed insertion group. This conclusion was carried over to the development of the autonomous systems.

In ref. [13], different types of motoric insertion of electrode arrays were studied in in vitro models of the scala tympani. Here, more than 1000 insertions were studied. It was concluded that the smoothness of the insertion could be greatly improved, and the necessary forces could be greatly reduced depending on the angle at which the electrode array was inserted. This angle is the angle relative to the centre line of the scala tympani. The smoothest insertion corresponds with the electrode array running parallel to this centreline. In this study, it was also shown that a non-constant speed of electrode insertion also leads to a decrease in the amount of force required to do the insertion. The speed of insertion should decrease depending on the progression of electrode array insertion. Less force would mean that the insertion would be less traumatic.

The technical knowledge of the most optimal electrode insertion will be used to make RACIS a fully automated process from middle ear access to electrode insertion. This knowledge will help standardise results in terms of residual hearing after the procedure.

3. Conclusion

RACIS has demonstrated to be safe and reliable for cochlear implantation. It has the potential to standardise the surgical outcomes of cochlear implantation to a maximum. When mechanical trauma is also limited to the bare minimum, it may pave the way for new indications for cochlear implantation in terms of electroacoustic simulation. The thrive to improve RACIS has already contributed a lot to the healthcare of patients with sensorineural hearing loss. The tools in image guidance surgery have shown us the importance of an individualised patient care because of the differences in cochlear duct length [14].

As a minimal invasive technique, RACIS requires only a canonostomy, whereas in conventional surgery, a full canonectomy needs to be performed to have visual confirmation of the round window and, if possible, the course of the basal turn.

Every trajectory to the inner ear is bound to hit the lateral wall of the inner ear. It may be shown that trajectories that are hitting the lateral wall more basal could have perhaps a better preservation of the apical structures. Currently, these data are not available.

The indication for RACIS versus conventional surgery is still strict. Currently, with the use of a drill of 1.8 mm, it is estimated that 50–75% of patients are indicated to be operated on via the RACIS technique. With future developments and the drill getting smaller to 1.0 mm, it is expected that RACIS would be possible in up to 100% of the patients [15]. The HEARO procedure is still not a fully autonomous way of cochlear implantation surgery, with autonomous electrode array insertion being not yet clinically possible. Further studies into making the procedure more efficient and safe are to be expected.

Acknowledgments

Prof. Vedat Topsakal enjoys a senior research grant from FWO Vlaanderen number 18B3222N.

All authors would like to thank MED-EL (Innsbruck, Austria) and CAScination (Bern, Switzerland) for their support for this project.

Conflict of interest

The authors declare no conflict of interest.