About the contributors
• Contributors to the work performed by the Parisian team: Yann Nguyen 1,2 , Mathieu Miroir 1 , Guillaume Kazmitcheff 1 , Evelyne Ferrary 1,2 , Daniele Bernardeschi 1,2 , Stephane Mazalaigue 3 , Armand Czaplinski 3 , Renato Torres 1 , Daniele De Seta 1,4 , Alexis Bozorg Grayeli 1 , Jérome Szewczyk 5 , Olivier Sterkers 1,2 .
1 Sorbonne Universités, Université Pierre et Marie Curie Paris 6, Inserm, Unité “Réhabilitation chirurgicale mini-invasive et robotisée de l’audition”, Paris, France; 2 AP-HP, GHU Pitié-Salpêtrière, Service ORL, Otologie, implants auditifs et chirurgie de la base du crâne, Paris, France; 3 Collin ORL, Bagneux, France; 4 Sapienza University, Roma, Italy; 5 Institut des Systèmes Intelligents et de Robotique (ISIR); Université Pierre et Marie Curie, Paris, France
Acknowledgments: The authors would like to acknowledge Collin ORL (Bagneux, France), Oticon Medical (Vallauris France), La Fondation Agir pour l’Audition, Advance Bionics (Valencia, USA) and MED-EL GmbH (Innsbruck, Austria) for their financial support.
• Contributors to the work performed by the Bern Team: Stefan Weber 1 , Kate Gavaghan 1 , Nicolas Gerber 1 , Tom Williamson 1 , Wilhelm Wimmer 1,2 , Georgios Mantokoudis 2 , Juan Ansò 1 , Christoph Rathgeb 1 , Arne Feldmann 3 , Franca Wagner 4 , Christian Weisstanner 4 , Masoud Zoka-Assadi 5 , Kai M. Roesler 6 , Olivier Scheidegger 6 , Lukas Anschuetz 2 , Markus Huth 2 , Martin Kompis 2 , Marco Caversaccio 2
1 ARTORG Center for Biomedical Engineering Research, University of Bern, Switzerland; 2 Department of ENT, Head and Neck Surgery, Inselspital, University Hospital of Bern, Switzerland; 3 Institute for Surgical Technologies and Biomechanics, University of Bern, Switzerland; 4 University Institute of Diagnostic and Interventional Neuroradiology, Inselspital, University of Bern, Switzerland; 5 MED-EL Corporation, Innsbruck, Austria; 6 Department of Neurology, Inselspital, University of Bern, Switzerland
Acknowledgments: The authors would like to acknowledge Brett Bell, Manuel Stebinger, Marco Matulic, Leatitia Perroud, Fabian Zobrist, Claude Jolly, Mauricio Reyes, Philippe Zysset and Philippe Büchler for the scientific contributions made to the presented system and related work. Additionally, the authors would like to thank the department of ENT surgery and Neuroradiology of the Bern University Hospital, particularity Nexhmedin Avdija, Susanne Hoffman, Nadja Feusi and Patrick Mühlethaler for their contributions to the presented clinical case. This work received funding from the Swiss National Science Foundation (Nano-Tera Initiative, HearRestore Project), the European Commission (under the HEAR-EU project, grant number: 304857), the Swiss Commission for Technology and Innovation, CAScination AG (Bern, Switzerland) and MED-EL GmbH (Innsbruck, Austria).
• Acknowledgments by the Vanderbildt team: The authors would like to acknowledge the following grants, NIDCD R01DC008408 Clinical Validation and Testing of Percutaneous Cochlear Implantation NIDCD R01DC012593 Safe, Rapid Access to the Internal Auditory Canal for Acoustic Neuroma.
• We would also like to acknowledge Dr Heinrich and Dr Zhang for providing pictures of their device.
Introduction
Robot-based surgery first emerged in the mid-eighties, with orthopedic and neurosurgical applications [ ]. The RoboDoc® system, reported by Taylor et al. [ ], was designed to assist the surgeon during hip replacement surgery. In neurosurgery, Kwoh et al. described the use of a robotic arm coupled to a navigation system to perform brain stereotactic biopsy, avoiding the need for a stereotactic frame [ ]. After these two pioneering robot-based systems, it became clear to the medical community that robotic technology could help surgeons reduce complication rates, enhance outcomes and potentially reduce procedure duration and cost. However, microsurgery and, in particular, otologic surgery has yet to benefit from the development of mechatronics and robot-based devices, despite the large potential for such technology to overcome surgical pitfalls. As is well known to surgeons, the biggest challenge of otologic surgery is the risk of damage to vital anatomy, which is often the target of repair; e.g. the inner ear, or facial nerve. The surgery takes place within a severely restricted space and requires the use of magnification with either a microscope or an endoscope. Further complicating the interventions are that positioning in surgery is complex and requires great spatial and tactile accuracy (e.g. ossicles show low resistance to fracture or dislocation), necessitating fine control of instruments used to manipulate the anatomy. And, especially in microscopic interventions and “key-hole” surgery, the surgical field is narrow with poor light penetration. Furthermore, surgery duration, especially for the most complex and challenging cases, can challenge the surgeon’s endurance, resulting in suboptimal performance at a time when performance is critical. All of these challenges, however, represent opportunities for the application of robotic systems.
In the last two decades, mechatronic assistance has been integrated into many operating rooms since Intuitive’s da Vinci® Surgical System came onto the market. Key features of its success are high reproducibility of motion, minimal invasiveness, and an intuitive command console allowing complex surgical movements under stereoscopic vision. While not officially designated a robot by the United States Food and Drug Administration (FDA), this surgical system has been widely used in urinary tract, abdominal, thoracic and gynecologic surgeries [ , ]. Although it has been successfully used for resection of base of tongue tumors, avoiding the need to split the mandible, its workspace dimensions and especially tool diameter are not well adapted for other ENT applications, including otologic surgery. For these reasons, multidisciplinary teams at various sites in Europe and the United States, composed of both ENT surgeons and engineers, have explored new designs specifically suited for otologic surgery. The results that have been achieved so far are described in this chapter as applied to 1) middle ear surgery, and particularly otosclerosis, and 2) cochlear implantation. We end the chapter with thoughts regarding further applications of robotic and mechatronic interventions as the technology progresses.
Robot-based devices for middle ear and otosclerosis surgery
Specific requirements of middle ear surgery, and in particular in otosclerosis surgery
Middle ear surgery is performed in a deep workspace through a narrow approach to the tympanic cavity, either via the external auditory canal, or by posterior tympanotomy. It uses long thin instruments, as delicate and precise action on the ossicular chain or manipulation of very light prostheses is necessary.
The workspace volume is too small to allow four-handed surgery, and one tool has to be dedicated to blood suction as even moderate bleeding will quickly cover the operating field. A suction pedal can be added to control vacuum strength, but even so, suction is not as efficient or versatile as microforceps.
Furthermore, the surgeon usually holds the tools collinearly to his or her axis of vision, as the access is often funnel-shaped. The field of view might be further impaired by the fact that the surgeon tends to hold the tools as close as possible to their tip, so as to maximize accuracy and minimize tremor. Thus, grasping the otological tools adequately requires dexterity and experience.
In otosclerosis surgery, stapes superstructure resection, fenestration or removal of the stapes footplate, placing and crimping an ossicular prosthesis between the incus and fenestrated stapes are key steps to restoring middle ear conduction. The procedure, often performed via a transcanal approach, is very challenging, and even more when performed under local anesthesia.
Stapes superstructure removal is usually performed after sectioning or vaporizing the posterior crura and fracturing the anterior crura. This first step is critical because, if the superstructure has not been weakened enough, mobilization may involve the footplate too. Fenestration of the footplate is now the most commonly used technique. It can be performed with a micropick, trephine, microdrill, laser or combination of the four [ ]. During this step, footplate fracture due to excessive pressure or footplate sinking are complications that can threaten preservation of cochlear function. Furthermore, fenestration has to be regular and calibrated, to allow correct motion of the prosthetic piston shaft, and centered on the posterior part of the footplate to minimize distance from the saccule. Finally, during piston crimping and placement, contact with the footplate should be prevented and incus stress should be minimized to avoid subluxation or dislocation. Crimping force should be controlled to ensure a mechanical contact and sound transmission between incus and prosthesis, but excessive pressure should be avoided in order to prevent secondary incus necrosis.
During all these steps, visual control of the footplate has to be maintained; blood penetration into the inner ear and perilymph suction should also be avoided at all cost. The functional result and rate of complications may vary with the surgeon’s experience [ , ]. To raise the success rate and diminish the incidence of complications for less experienced surgeons, technical modifications have been progressively adopted in the last three decades. Stapedotomy was proposed instead of stapedectomy, to decrease sensorineural hearing loss [ ]. The laser was used, alone [ ] or in combination with a microdrill [ ], to assist footplate fenestration and reduce the risk of footplate fracture. We, like other ENT teams worldwide, recommended robot-based assistance, to increase safety [ ]; it is these devices that will be described in this chapter.
Expectations for robot-based assistance in middle-ear and otosclerosis surgery
The objectives of robot-based assistance in middle-ear and otosclerosis surgery can be divided into primary and secondary objectives.
The accuracy of surgery can be enhanced by reducing involuntary movement.
Involuntary movements have been described in microsurgery [ ] and this description can be applied to middle-ear surgery. Tremor is a rhythmic sinusoidal (range, < 0.2 mm) movement (wrist: 8–12 Hz; fingers: 17–30 Hz), causing the tip of the tool to oscillate around the target position. Drift is a non-sinusoidal low-frequency movement of greater amplitude (> 0.5 mm), causing the tip of the tool to be translated instead of remaining still after a few seconds. Jerk is a sudden reflex or spasmodic muscular movement, related to lack of experience or to tiredness. And finally undershoot and overshoot designate lack of accuracy in trying to reach a target point during displacement.
The objective of reducing involuntary movement is first to increase the safety of the surgical procedure. Intraoperative complications such as ossicular dislocation, facial nerve lesion or tympanic perforation during tympanomeatal flap dissection are related to experience and accuracy, and their incidence could be lowered with the assistance of a robot-based arm by employing technologies such as scale-motion control, virtual fixture to prevent crossing critical zones (see chapter 5 ), or supraphysiological force sensing. Tremor and jerk are easy to eliminate with a conventional robot-based micromanipulator or a hand-held intelligent tool [ ]. Although their physiopathology is not fully understood, rates of postoperative vertigo or tinnitus could also doubtless be reduced by reducing mechanical trauma to the inner ear through ossicular chain manipulation. If the safety of the procedure can be enhanced, this alone would justify using such systems. Another objective is certainly to improve functional outcome, with, for example, better control of cholesteatoma resection, or more precise fitting of ossicular prostheses. For expert surgeons, outcome may not be improved, but technological assistance can be expected to guide less experienced surgeons toward a more skillful procedure.
Numerous secondary objectives of robot-based assistance can be envisaged. Thus, ergonomics can be enhanced by a teleoperated system: how the surgeon sits or positions his or her arm no longer depends on the patient’s position and anatomy and the orientation of the microscope or endoscope. With a teleoperated system, only the operating field and have to be sterile, and not the surgeon’s hand and commands. Better ergonomics and the ability to hold the tool without any physical effort can help reduce tiredness. Today, surgeons can use only two tools in the operating field, and one of these is dedicated purely to suction. Optimizing tool placement, displacement and geometry may allow more than two tools in the surgical field. Additionally, some tasks, such as blood suction, could be automated to allow the surgeon to concentrate on the critical aspects of surgery that require human decision-making and control.
Furthermore, a teleoperated system can facilitate the teaching process in the operating room, and consequently shorten the learning curve of young surgeons.
Intraoperative switch between two surgeons of equivalent experience or between a junior and a senior surgeon can be made without any delay. Furthermore, teleoperated systems with a dual command console can allow simultaneous access to the robot and operating field by the teaching surgeon and the trainee. In our current practice, such interaction during microscopic surgery is not possible, due to the restricted access to the operating field, preventing residents performing the most complex tasks of middle-ear surgery. Dual consoles have been developed for the da Vinci® system [ ].
Finally, from a medico-economic point of view, the rationale for the purchase of expensive devices is a reduction in the overall cost of patient management, by reducing surgery time, hospital stay and postoperative care. For rapid deployment, the devices will have to be not only safe and reliable but also easy to manipulate by all operating room staff.
In the following section, examples of robot prototypes designed or adapted to comply with the requirements and expectations of robot-based assistance for otosclerosis surgery will be presented.
Examples of robot-based applications for middle-ear and otosclerosis surgery
The Steady Hand (Johns Hopkins University)
The Steady Hand is a co-manipulated robot with 7 degrees of freedom (DOF). It was designed by Russell Taylor to assist the surgeon in microsurgical procedures [ ]. Composed of three prismatic links with an axis perpendicular to the base and a parallelogram frame yielding two rotational DOFs, its kinematics is a parallel structure with a remote center of motion (see chapter 4 ). An effector stage with an additional degree of rotation bears a linear actuator to translate an effecting tool along its axis. Two 6-axis force sensors (ATI-nano 17™) are mounted on the structure to allow co-manipulation: the first analyzes the interaction between the operator’s hand and the robot; the second analyzes contact between the tip of the tool and its environment.
This device has been used in ophthalmological applications such as retinal vessel cannulation, and in otosclerosis to perform high-accuracy footplate fenestration. Since it is a co-manipulated robot, operation and manipulation do not require long training. Displacements are scaled as 1:1, but tremor can be filtered to enhance quality. Furthermore, restriction zones with virtual fixtures can be applied, and force feedback can amplify tactile sensations [ ].
The Steady Hand was evaluated by Berkelman et al. [ ] and compared to the standard manual technique in two tasks required for otosclerosis surgery: platinotomy with micropick fenestration, and piston-crimping on the long process of the incus. First, the authors showed that the robot-based technique enhanced the accuracy of fenestration location. Secondly, fenestration could be performed in a single step, with lower peak force (< 3 N) than in the manual technique, providing multiple contacts with numerous force peaks. Forces applied by the prosthesis on the footplate model during piston-crimping were similar with both techniques (approximately 0.08 N). Based on these promising results, additional experiments were conducted by Rothbaum et al., with surgeons of various levels of experience [ ]; results confirmed the benefit of a robot-based procedure. Thus, the cumulative force required for footplate fenestration was reduced by 58% compared to the manual technique, and surgery time was shorter. The surgeon’s experience (junior versus senior) showed no impact on the forces applied. However, an unexpected result was found for targeting accuracy: junior surgeons benefitted from Steady Hand assistance, whereas senior surgeons were less accurate in fenestration positioning with the robot-based technique.
The smart micro-drill (Birmingham University, UK)
The Birmingham team (P.N. Brett and C.J. Coulson) developed an intelligent drilling tool ( Figure 11.1 ) able to detect the passage of the drill from bone to soft tissue. The robot analyzes the axial force applied on the tool and the torque of the drill. The tool is placed manually in the surgical field by the surgeon, and drilling is performed automatically by the robot. As the drill exits the bony structure, the force suddenly drops and torque rises as less bone is in contact with the drill. This information is crucial for inner-ear exposure. Thus, footplate fenestration or cochleostomy can be performed conserving the endosteum and integrity of inner ear structures. Clinical applications of this device comprise otosclerosis surgery and cochlear or vestibular implantation. The device was successfully used in the operating theater, and a series of three patients was reported [ ].
The Micromanipulator System II (MMSII; Technische Universität München, University Hospital of Leipzig)
The MMS-II was developed in Germany by roboticians from the Technische Universität München and ENT surgeons from the University Hospital of Leipzig [ ]. The system was designed to be simple but as efficient as possible. Its goal is to assist surgeons not throughout a procedure but during specific tasks such as middle-ear prosthesis placement. To shorten bench-to-OR time, the device was designed to bear the conventional otological tools found in most ENT operating rooms. The MMS-II is composed of a small manipulator with four DOF, comprising a XY-table with a thin vertical Z-axis, a mechanical articulated arm, and an axis for opening and closing attached forceps. It is controlled with two joysticks. It can be placed in a sterile drape to ensure compliance with operating room regulations. The initial version has been enhanced and given additional DOFs by mounting it on a macromanipulator (Jaco® lightweight robot by Kinova, Quebec City, Canada) [ ]. It was used in a clinical study, although the outcomes have not yet been reported. A study of accuracy was performed [ ]. Ten ENT surgeons were asked to target a precise point on the stapedial footplate. Results showed that accuracy was not improved by the robot, but learning curves were shorter when the micromanipulator was used.
The RobOtol System (Pierre et Marie Curie University, Paris)
The RobOtol project ( Figure 11.2 ) began in a context when robot-based surgery was rapidly expanding in urologic, abdominal and chest surgery but no device was commercially available for microsurgery. Analysis of the market orientation made it clear that no device for microsurgery was on the horizon. Otology, moreover, involves additional requirements, as it combines microsurgical accuracy and key-hole approaches. For this reason, it seemed obvious to us that we had to take the initiative to develop such a robot to give a chance to this project.
The RobOtol project was initiated through a partnership between three academic partners: Inserm/University Paris 7 UMR-S 867 Minimal Invasive Robot-Based Otologic Surgery, the Laboratoire de Robotique de Paris (LRP, Fontenay-aux-Roses), and the otorhinolaryngology department of Beaujon Hospital, Paris, plus an industrial partner (Collin Ltd., Bagneux, France). Thus, a multidisciplinary team composed of experts from the fields of robotics, otologic surgery and industrial partners familiar with certification and economic considerations was brought together to design a robot dedicated to otologic surgery. The goal was to fulfill the need for a new device that would comply with the specifications of middle ear surgery. Start-up was in 2006.
Specifications
The first step of the project was to draw up the specifications to guide the design of the system. Initial specifications were based on otosclerosis surgery (laser stapedotomy) performed through a transcanal approach. This procedure was chosen as a model for middle-ear surgery because it is very reproducible from one patient to another, with little inter-operator variation, unlike other middle-ear surgeries. Furthermore, most of the procedures and tool/organ interactions found in middle-ear surgery are found in laser stapedotomy, and physiological anatomy is only slightly modified by the disease. Besides, we wanted the robot-based procedure to be easy to deploy and compatible with the current technique combining hand-held tools and microscopic exposure, in order to allow rapid adoption by the ENT community and fast conversion between the robot-based procedure and the manual technique in case of issues arising with the device.
External workspace requirements
The first specification was to model the external workspace of the robot ( Figure 11.3 ); in other words, its interaction with the patient’s body and the other devices in the operating room. Thus, the microscope, patient’s head, upper thorax and operating table were considered for calculation. A combination of two robots, each bearing one effective arm, and the footprint of the robot cart were taken into account to plan device positioning around the operating table. Optimization of the external workspace consisted in simulating the robotic arms’ position around the patient and calculating the minimal distances between the patient’s body, table, microscope and robotic arms. Collisions between the non-effective parts of the robotic arms and the patient’s body during positioning were considered unacceptable. The second step of optimization simulated the impairment of the visual field of the microscope when the robotic arms cross the line of sight between middle ear and microscope ( Figure 11.4 ). Based on these two factors, a tradeoff was made between safety (minimal distance between non-effective robot parts and patient’s body) and optimal exposure. This optimization process will be detailed in a following part of this section.
Internal workspace requirements
The second requirement was the workspace needed for effective interaction between the tip of the tool and the target organ. This step was also mandatory for designing the smallest and lightest robotic arm that could reach the internal workspace. As the initial primary task of the robot was otosclerosis surgery through a transcanal approach, the internal workspace was defined as the middle-ear cleft and the speculum that has to be crossed to reach the target points. To ensure that all of this volume would be reachable in clinical practice, dimensions were acquired on 20 temporal-bone CT scans from adult patients. Maximal dimensions in coronal, axial and sagittal planes were measured, rounded up and saved to define the internal workspace of the robot. Figures 11.3 and 11.4 sums up determination of the volume of the internal workspace.
Minimal force requirement and accuracy to set actuation specifications
Choice of actuation was also critical to optimize the weight and volume of the overall device while fulfilling the ability to achieve the task with enough force. When the robot was initially designed, forces commonly applied and required to perform the different steps of an otosclerosis procedure had not yet been reported, and these original data had to be collected. Thus, a force measurement test bench was built to mount a cadaveric temporal bone on the force sensor (Nano 17, calibration type SI-12-0.12, resolution 1/320 N, ATI Inc. Apex, NC). Particular attention was paid to allowing two surgeons to perform the usual otosclerosis procedure on the temporal bones despite their being fixed on the force sensor. A set of 8 temporal bones were included in resin for fixation on the force sensor. Measurements were recorded via an analog-to-digital interface card, and an in-house program at a sampling rate of 1,000 Hz. Tool/organ interaction forces were measured for scutum resection using an otologic drill (2 mm diamond burr, Bien-air® drill, Bien-air Surgery SA, Noirmont, Switzerland) or a curette, stapedial tendon and crura section, incudostapedial joint separation, stapedotomy using a microdrill, prosthesis positioning, and crimping. Applied forces ranged from 2.5 to 3.5 N for scutum lowering with the drill and 7 to 20 N with the curette, 0.45 to 1.8 N for stapedial and crura section, 0.7 to 2.75 N for incudostapedial joint separation, and 0.75 to 1.5 N for stapedotomy drilling. Since scutum lowering with the curette involved far greater force than the other procedures, scutum lowering by a curette held by the robot was abandoned. Based on these measurements, 5 N was set as the upper limit for robot force/torque.
The next step was to determine the desired accuracy of the robot. No publication had previously set a minimal accuracy for otosclerosis surgery, or related targeting errors to functional results. Unquestionably, the accuracy of the robot-based device had to be greater than the surgeon’s accuracy (human physiological tremor has a 0.2 mm amplitude [ , ]). Minimal accuracy was arbitrarily set by the surgical team at 1% error during stapedotomy drilling. Considering a 500 μm diameter stapedotomy and a 1 mm-thick footplate, this corresponded to a linear resolution of 5 μm and an angular resolution of 0.3° ( Figure 11.5 ).
Kinematic choice and topological optimization
The global architecture of a robot, representing its “skeleton”, is known as its kinematics. The classical kinematics, serial, parallel and mixed, are described in detail in chapter 4 .
The kinematics determines the number of degrees of freedom, the shape of the internal workspace and the volume of the robotic arm. Choice of kinematics favored a robotic arm shape that would preserve the surgeon’s field of view. Thus a serial kinematic chain of 3 perpendicular linear links at the base and 3 rotary links at the distal part of the arm was selected. This chain allows 6 DOF, with 3 translations and 3 rotation axes for a passive tool. An additional rotary actuator was mounted on the top of the last segment of the arm to allow a seventh actuator that can control opening – close of forceps or scissors. All rotation axes were designed to be concurrent, so as to obtain a remote center of motion situated at the tool tip. This allows pure tool rotation without any tip translation. This configuration was set as mandatory by the surgical team, to increase safety during displacement, and in particular to lower the risk of involuntary ossicular chain contact. Thus, when the orientation of the tool within the middle ear is changed, there is no translation; this facilitates the control of the tool by the surgeon.
The last step consisted in topological optimization. The kinematics of the robot and its components was chosen at this step. However, the length of the various arm segments, their angles and the positions of joints and actuators remained to be determined. Candidate robot were therefore distinguished according to differences in these parameters. Then an algorithm was computed to systematically explore all the possible candidates based on these criteria. Each candidate had to meet the objectives and prerequisites determined in the specification process; i.e. to reach the entire workspace and produce a 5 N force in all directions. Once these objectives were met, the minimal distance from the extracorporeal environment and impairment of the microscopic field of view were assessed. These two criteria were used to classify candidates that had passed the first selection. The algorithm and selection process are represented in Figure 11.6 . The candidate robot that was finally selected was the solution yielding maximum safety with a minimal distance of 5 cm from the extracorporeal environment in all positions of the robotic arm; among the best candidates, it was the one that would ensure binocular exposure of the operating field in 90% of cases.
Description of the system
Components
Three different versions of the RobOtol have been built so far ( Figure 11.7 ). The first was made by rapid prototyping, but its plastic arm was too soft to allow precise control. Furthermore, in this version, the rotary actuators were bulky and would impair the field of view. For this reason, they were placed behind the robot and transmission was ensured by Bowden cables (see chapter 4 ). The second version of the arm was in rigid metal, which improved stability and reduced tremor at end-of-movement. However, accuracy was still unsatisfactory, as slippage was observed on transmission from the Bowden cables. Encoders were therefore added to correct this inaccuracy and to correct the command control to compensate for this slippage. Arm drift was also totally eliminated in this version, with an additional magnetic lock on the base of the arm. In the current version of the robot, progress in motor technology enabled the rotary actuators to be embarked directly.
The system is presently composed of two robotic arms that can be used together or separately. The 3 linear actuators were integrated into a XYZ cross-table. The table was built with 2 orthogonal (X-Y) precision linear stages with 70 mm travel, and a Z precision linear stage with 95 mm travel (LTM 80P-75-HIDS and LTM 80P-100-HIDS; OWIS GmbH, Staufen, Germany). Three DC micromotors (2342S024CR; Faulhaber GmbH, Germany) with an IE2-512 magnetic incremental encoder (512 lines per revolution) were used for all rotary actuators. These were connected to a Harmonic Drive gearhead (HFUC-8-100-2A-R; Harmonic Drive LLC, Peabody, MA) with a 100:1 reduction ratio.
Command
Two command modes are available to teleoperate the RobOtol System. In both modes, a dead man’s foot switch (DMFS) allows surgeon to confirm movement commands. In both command modes, a down-scale ratio between the master and slave arms can be implemented to enhance control for a more accurate movement.
The first command mode is the Phantom Omni® (SensAble Technologies, Inc., Woburn, MA). This interface was used as master control. The distal part of the interface is a stylus. The surgeon controls the arm remotely by a pen-like interface with 6 DOF. This interface improved the ergonomics by preserving the space around the patient’s ear. It works with a position-to-position command from the master to the slave arms. The command is based on a registered correspondence between the local stylus frame and the robot tool frame ( Figure 11.8 ). In this mode, the user does not control the speed and the robot automatically stops once the target position is reached.
The second command mode uses a space mouse (3Dconnexion®; Waltham, MA), with a position-to-velocity command. This registration mode allows master control configuration to be uncoupled from the robot, and allows indirect visual feedback (e.g. angled endoscope). In the velocity command mode, stylus motion codes for robot speed (as with a joystick). In this mode, the user has to release the DMFS to stop the robot.
Concept validation in cadaveric assessment
External workspace evaluation
The robot was evaluated in realistic conditions in a cadaver lab and in an operating room on an artificial dummy patient ( Figure 11.9 ). A complete set of tasks commonly performed during middle-ear surgery were assessed. In none of the positions was there conflict between the patient’s head and the robotic arm. The minimal distance between the robot and the external workspace was always that of the microscope head. Furthermore, the visual field under the operative microscope, with a binocular view required to obtain 3D vision, could be preserved by adequate orientation of the arm and tool during the tasks ( Figure 11.10 ). The surgeon would, indeed, often orientate the tool anteroposteriorly. This kind of manipulation or orientation is less easy to achieve with a manual technique. The surgeon’s hands could be placed posteriorly to the ear canal and orient the tools anteriorly into the middle-ear space.
Internal workspace and task achievement evaluation
Ability to reach specific landmarks was checked on 5 cadaver specimens by 5 different evaluators. Target points such as the four quadrants of the tympanic membrane, round window and oval window could be reached by all operators. Passive tools such as a microhook, micropick, joint knife or tab knife and active tools such as microscissors and microforceps could be mounted on the robot ( Figure 11.11A ). Furthermore, the suction tool could also be connected to the arm, and could hold a laser fiber if needed ( Figure 11.11B ). With this tool pack, procedures such as ventilation tube placement ( Figure 11.12A ) or laser sectioning of the stapedial tendon ( Figure 11.12B ) or crura or stapetotomy and piston placement and crimping could be performed in cadaver bone. The task could be performed with both command modes (Phantom Omni® or Space Mouse®), depending on the surgeon’s preference. Neither mode demonstrated superiority for task duration or success at first trial. No complications such as tympanic perforation or ossicular chain dislocation were observed during this experimental validation.
Where we are now, and future developments of the RobOtol system
The current modifications made to the RobOtol device extend its capacity to perform other procedures. The goal is for the device to play a multi-task role. From a medico-economic point of view, this will ensure faster deployment, if the device is not restricted to a single application such as otosclerosis surgery. The first new development of the robot is its coupling to a cochlear implant insertion tool and navigation device in order to automate the alignment of an insertion tool on a planned axis. This application will be detailed in the next section. The second development is the ability to transform the robot into an endoscope holder. This will be presented in the last part of the chapter. The RobOtol has obtained “Conformité Européenne” certification (CE marking in June 2016). After 10 years’ development, it is now ready to begin clinical evaluation.
Robot-based devices for cochlear implantation
Introduction
Cochlear implantation has become the standard of care for hearing rehabilitation in patients with severe to profound sensorineural hearing loss that has progressed to the point that benefit from hearing aids is limited. After activation of the device and intensive speech and audiology rehabilitation, many patients can achieve word scores greater than 60% and sentence scores greater than 80%. However, outcomes remain highly variable, with postoperative performance dependent on the etiology and duration of hearing loss, age at implantation, educational level, presence of concurrent central nervous system (CNS) disease [ ], and preoperative speech performance [ ]. Postoperatively, factors influencing outcome include device-fitting with manufacturer’s specific coding strategies, patient compliance with speech therapy, and ubiquitous hardware and software upgrades of the device [ ]. During a cochlear implant recipient’s lifetime, the actual surgical intervention accounts for only a very short period, but can have lasting repercussions if residual hearing can be preserved: hearing preservation is associated with better postoperative speech recognition in noise and better appreciation of music [ ]. For this reason, much emphasis has been placed on minimizing inner-ear trauma during implantation, including limitation of drilling during cochleostomy creation and avoidance of bone dust and blood penetration into the inner ear [ ]. Furthermore, individual cochlear anatomy is taken into consideration for array selection, using measurement of cochlear diameter [ ] to estimate two-turn scala tympani duct length. Advanced array designs, with progress in diameter and rigidity and the use of guiding stylets or insertion tools, may decrease trauma to inner-ear structures during array insertion [ ]. Such advances are especially relevant for the expanding indications of cochlear implantation in patients with residual hearing, as electrical acoustic stimulation (EAS) has been shown to be associated with overall better outcome [ ].
Prerequisites of cochlear implant surgery and expectations for robot-based assistance
Cochlear implantation can be decomposed into three key steps: 1) access to the cochlea, 2) opening of the cochlear via either the round window or a separate cochleostomy, and 3) insertion of the electrode array. Each step presents its own challenges.
First step: access to the cochlea
To access the cochlea, a mastoidectomy cavity of approximately 2–3 cm in each of the three dimensions is drilled; the smaller structures of the temporal bone with dimensions in the millimeter range (e.g. ossicular chain and facial nerve) are to be avoided. This wide exposure is useful for lighting the surgical field adequately and allowing enough space for rotation and translation of surgical instruments within the conical workspace. Dissection starts laterally and progresses medially, with significant time spent identifying surgical landmarks which visually guide progressive dissection. These landmarks are fixed in the skull and are not deformed by surgery, unlike the soft-tissue landmarks encountered during neurosurgical, abdominal and/or thoracic surgery. Furthermore, the temporal bone has marked air-bone boundaries, particularly when free of chronic infection (i.e. during most cochlear implant surgery). These two conditions—landmarks fixed in bone, and distinct air-bone boundaries—are favorable for the use of computer-assisted surgery. The first reports of IGS in the temporal bone date from the late 1990s [ ]. As it lacked the required submillimetric accuracy, navigation was used only as an assistance tool for the surgeon, to help identify anatomical structures of interest in revision surgeries and/or in case of anatomical variants. Strategies arose to improve registration accuracy by adding fiducial markers, using either a dental bite-block [ ] or screw-in-bone anchors [ , ].
The use of a robot to access the cochlea overcomes two particular human limitations relevant to temporal bone drilling. Firstly, spatial accuracy is poorer with human than machine drilling: drilling requires applying manual pressure from the hand-held drill to the bone, which can distort positional feedback. Secondly, the surgeon requires constant visual feedback during temporal bone drilling, whereas a robot uses positional feedback that can be obtained from non-visual input (e.g. infrared tracking). Visual feedback to a human operator via the position of a tracked tool on traditional axial, coronal and sagittal plane screens does not provide sufficient information to ensure that the operator can safely avoid critical structures such as the facial nerve [ , ]. However, a robot can be coupled and enslaved to an image-guided system to drill safely in close proximity to vital anatomy without visual feedback. To access the cochlea, most authors agree that drilling accuracy should have an accuracy of 0.5 mm to avoid damage to the facial nerve and access the cochlea with enough clearance to insert a 1 mm diameter electrode diameter array. To achieve this objective, issues of drilling path registration accuracy and drill and robotic arm navigation had to be solved. Specific prerequisites and the technological solutions developed by the ARTOG Center (Bern, Switzerland) [ , ], Hannover (Hannover, Germany) and Vanderbilt teams (Nashville, TN, USA) will be discussed in subsequent subsections of this chapter.
Second step: opening the cochlea
Placing cochlear implant electrode arrays via the scala tympani was initially recommended for geometric reasons. However, recent studies have shown that complete scala tympani insertion is associated both with better speech understanding [ , ] and with better hearing preservation [ ]. Access to the scala tympani can be performed through three approaches: round window approach, extended round window approach with drilling of the anterior-inferior ridge of the round window (also called peri-round window approach), and cochleostomy. Round window exposure may differ from patient to patient [ ], and pure round window insertion may not be possible in certain cases. Opening into the cochlea may benefit from robotic assistance. Firstly, if cochlear access has been made through a minimally-invasive tunnel, the cochlea may have to be opened without direct vision of the cochlea, and this will require, at the least, computer-assisted positioning. In performing the cochlear access approach, the topography of the middle ear is not predictive of sub-surface anatomy [ ], and image-guidance can therefore be useful to ensure that the scala tympani is targeted. With a robot-assisted drill, tremor can be eliminated and cochlear opening calibration can be controlled so as to match the electrode array diameter precisely, ensuring penetration without friction secondary to contact with the crista fenestra while limiting the risk of postoperative perilymph leakage. Drilling could also be controlled in order to stop drilling once the bony shell has been opened, to prevent opening the endosteal membrane. The force-feedback drill developed at the University of Birmingham (United Kingdom) has shown this to be feasible, including a dramatic demonstration of drilling an egg without violation of its external membrane layering the internal shell [ ].
Third step: insertion of the electrode array
Array insertion is currently performed manually with limited visual and tactile feedback, despite being a very delicate procedure that requires minimization of interaction forces [ ]. At present it is performed manually, but interaction forces are so small that they may not be detectable by human operators [ ]. Insertion force measurement in real conditions has only been reported ex-vivo [ ]; resistance was manually determined, and thus totally subjective. Lee et al. performed postmortem histologic studies comparing patients with partial or complete insertion. Partial insertions were not associated with osseous or fibrous obstacles in most cases [ ]. It was shown that lesions of the basilar membrane, osseous lamina or spiral ligament did not impair the chances of complete insertion. This suggests that anatomic lesions are related to insertion forces. This can be seen in the light of Ishii’s study, which showed that the mechanical resistance of the basilar membrane is 0.039 N in the basal turn and 0.029 N at the apex [ ]. Inner-ear structures may be torn or crossed without any visual or sensitive feedback to the operator. To reduce trauma to inner-ear structures generated during cochlear implantation, it would be ideal to measure real-time insertion forces during array insertion, so as to be able to modify insertion based on this feedback.
Array positioning in the cochlea also affects postoperative speech performance results [ , ] and hearing preservation [ ]. Translocation from scala tympani to scala vestibuli ( Figure 11.13 ) should be prevented to avoid jeopardizing postoperative speech performance. Electrode array modiolus distance in the basal turn and incomplete insertion are also factors affecting speech performance [ , ], although it is unclear if surgeons can perceive such translocation [ ].
