Cadaveric Feasibility Study of da Vinci Si–Assisted Cochlear Implant With Augmented Visual Navigation for Otologic Surgery

Importance  To our knowledge, this is the first reported cadaveric feasibility study of a master-slave–assisted cochlear implant procedure in the otolaryngology–head and neck surgery field using the da Vinci Si system (da Vinci Surgical System; Intuitive Surgical, Inc). We describe the surgical workflow adaptations using a minimally invasive system and image guidance integrating intraoperative cone beam computed tomography through augmented reality.

Objective  To test the feasibility of da Vinci Si–assisted cochlear implant surgery with augmented reality, with visualization of critical structures and facilitation with precise cochleostomy for electrode insertion.

Design and Setting  Cadaveric case study of bilateral cochlear implant approaches conducted at Intuitive Surgical Inc, Sunnyvale, California.

Interventions  Bilateral cadaveric mastoidectomies, posterior tympanostomies, and cochleostomies were performed using the da Vinci Si system on a single adult human donor cadaveric specimen.

Main Outcomes and Measures  Radiographic confirmation of successful cochleostomies, placement of a phantom cochlear implant wire, and visual confirmation of critical anatomic structures (facial nerve, cochlea, and round window) in augmented stereoendoscopy.

Results  With a surgical mean time of 160 minutes per side, complete bilateral cochlear implant procedures were successfully performed with no violation of critical structures, notably the facial nerve, chorda tympani, sigmoid sinus, dura, or ossicles. Augmented reality image overlay of the facial nerve, round window position, and basal turn of the cochlea was precise. Postoperative cone beam computed tomography scans confirmed successful placement of the phantom implant electrode array into the basal turn of the cochlea.

Conclusions and Relevance  To our knowledge, this is the first study in the otolaryngology–head and neck surgery literature examining the use of master-slave–assisted cochleostomy with augmented reality for cochlear implants using the da Vinci Si system. The described system for cochleostomy has the potential to improve the surgeon’s confidence, as well as surgical safety, efficiency, and precision by filtering tremor. The integration of augmented reality may be valuable for surgeons dealing with complex cases of congenital anatomic abnormality, for revision cochlear implant with distorted anatomy and poorly pneumatized mastoids, and as a method of interactive teaching. Further research into the cost-benefit ratio of da Vinci Si–assisted otologic surgery, as well as refinements of the proposed workflow, are required before considering clinical studies.

Cochlear implant (CI) surgery, which was pioneered in the 1960s, has been performed in more than 320 000 individuals worldwide. The cochlear implant has been approved by the US Food and Drug Administration for congenital and acquired sensorineural hearing loss since the 1980s. Although there have been many innovations to the implant device, in particular with the development of multichannel electrodes, the surgical approach has not deviated significantly since its inception. Typically, the CI surgery involves a standardized approach through a mastoidectomy, facial recess drill out, posterior tympanostomy, and cochleostomy with electrode insertion.

The accuracy of cochleostomy placement and CI insertion angle has been shown^1-3 to be critical for device function and clinical functional outcomes. Appropriate cochleostomy placement is important to allow for cochlear electrode array insertion in the scala tympani and not the scala media or vestibuli. This can be challenging given the depth of the cochlea, which is 30.0 mm within the temporal bone, and the facial recess width of 2.5 to 3.0 mm with a target area for cochleostomy of 1.0 mm for appropriate insertion.⁴ The current cochleostomy surgical technique requires freehand drilling with a 0.6- to 1.0-mm-diameter burr through the promontory into the scala tympani without intracochlear trauma. Recent studies,^5,6 however, suggest that a significant proportion of CI surgeons do not adequately position the cochleostomy, typically described as inferior or anterior-inferior to the round window. Not surprisingly, less-experienced surgeons are more likely to have inadequate exposure of the round window through the facial recess.

Novel technology integrating preoperative/intraoperative image data has the potential to significantly improve the accuracy of the implantation approach through improved facial recess widening and cochleostomy placement. Image guidance can be done through registration of standard preoperative computed tomography and/or cone beam computed tomography (CBCT) coupled with mapping of the fixed temporal bone landmarks and target trajectories directly to the patient’s operative workspace. Other groups⁷ have described a stereotactic frame carefully mounted to the patient’s skull and surrounding mastoid, which facilitates a surgical drilling trajectory that enables a percutaneous approach to implantation. In fact, the use of optical tracking systems (eg, StealthStation; Medtronic) for image-guided, endoscopic, endonasal skull base surgery is well described^8-11 in clinical use and robot-assisted otoneurosurgery,^12-17 as well as computer-assisted implant placement for CI surgery,^4,13,18-20 and is an ongoing active area of research.

The goal of this study was to test whether a well-developed and industry-standard system (da Vinci Surgical System; Intuitive Surgical, Inc) could facilitate cochlear implant. The study used a da Vinci Si system with custom tool adapters for master-slave–assisted cortical mastoidectomy, posterior tympanostomy, and cochleostomy. Two cadaveric case studies were conducted using the master-slave system to test the feasibility of a proposed clinical workflow. With the second case, we introduced additional steps incorporating intraoperative CBCT-based image guidance through stereo video augmentation with direct overlay of critical anatomy (facial nerve, cochlea, and round window).

A da Vinci Si system was used in bilateral cadaveric cochleostomies and mastoidectomies. Institutional review board review from Children’s National Medical Center was waived for this study. Case 1 was a donor cadaver left temporal bone, with the CI surgery being performed with the master-slave system. Case 2 was a donor cadaver right temporal bone, master-slave assisted with augmented reality. To dissect the temporal bone, a custom drill adapter was fabricated using a 3-dimensional (3D) printer (Objet Eden500V; Stratasys, Ltd). This secured the attachment of an osteon pneumatic drill (CONMED) with a 30° offset on the shaft of an 8-mm da Vinci Si tool. The offset strategically positioned the da Vinci Si arm away from the workspace while allowing the drill shaft effective parallelization with the axis of the endoscope. High-definition (1080 interlaced scan at 60 Hz) 3D visualization, rendered through the surgeon’s console using a 12-mm (0°) da Vinci Si endoscope, had a minimal pixel resolution of 0.4 mm/pixel at a corresponding near-field measurement of 7 mm and provided appropriate viewing for the surgeon. Digital magnification ranged from ×1 to ×4 (cochleostomy), and the scale of the master-slave manipulators was fixed at a ratio of 3:1.

The workflow steps involving cases 1 and 2 are outlined in Figure 1. The second case extended the workflow steps of the first with integration of CBCT-based image guidance through stereo video augmentation of segmented critical anatomic structures.

Soft-tissue work to expose the mastoid was performed without the da Vinci Si. The surgeon exposed the mastoid bone and retracted the ear forward. The following workflow was used for case 2, and the workflow for case 1 included all of the components listed below except for those identified with an asterisk (Figure 1).

1.* Expose mastoid bone and retract ear forward to fix 3 self-drilling zinc pan head Phillips screws 6 mm long (No. 6 drill bit used to initiate entry; custom drill adapter shown in Figure 2 inset), placed with their centroid above the inner ear to an exposed right mastoid (Figure 3A).

2.* Acquire preoperative CBCT scan (Siemens Powermobil; Siemens; head protocol, 109 kilovolt [peak], 290 mA, 0.48 × 0.48 × 0.48-mm³ voxel size) with the isocenter positioned around the cochlea.

3.* Use ITK-Snap (http://www.itksnap.org/) ²¹ to manually segment the facial nerve, basal turn of the cochlea, and the round window (Figure 3B), as well as the divots of 3 fiducial screw heads from preoperative CBCT.

4. Position the components of the da Vinci Si as shown in Figure 2, with the base of the patient side cart on the opposing side of the patient target.

5. Use the printed adapter to attach the drill at a 30° offset onto the shaft of an 8-mm da Vinci Si tool that is inserted into the primary robotic arm while either a da Vinci Si suction/irrigator tool (EndoWrist; Intuitive Surgical, Inc) is placed or a standard suction/irrigator device is fixed to the secondary arm.

6.* Conduct manual point-based registration of endoscopic view (video) to the CBCT²² by identifying the fiducial screws.

7. Perform mastoidectomies, posterior tympanotomies, and cochleostomies using the da Vinci Si.

8.* Perform manual refinement of cochleostomy and insertion of an implant wire phantom into the cochlea at the basal turn (Figure 4).

These 2 case studies demonstrate the feasibility of master-slave–assisted cochleostomy and mastoidectomy with augmented reality integrated as image guidance on a da Vinci Si. Setup for the system after exposing the soft tissues and placing the fiducials, which included capturing the CBCT images, segmenting the critical anatomic structures, and configuring the augmented reality capability, required approximately 110 minutes. This information is summarized in the Table. The mean surgical time for surgery per side was 160 minutes. This surgical time included the time required to bring the da Vinci Si system to the field, mount the tools and drill to the robotic arms, and perform all aspects of the drilling and electrode array insertion. More important, there was no violation of critical structures and, in both cases, there was full insertion of the phantom electrode array. Accuracy of the augmented reality overlay was confirmed on the second side by uncovering the mastoid segment of the facial nerve. The position of the nerve was found to correspond accurately with the overlaid image. We did not assess the position of the phantom array within the cochlear compartments with histopathologic testing. Similarly, the angle of insertion and the position of the cochleostomy were not confirmed with histological analysis.

The da Vinci Si–assisted drilling of the mastoid bone was conducted on the cadaver’s left ear without breaching of the facial nerve to expose the middle ear. Successful cochleostomy, which was confirmed with postoperative CBCT, was also performed at ×4 digital magnification with the da Vinci Si system.

Similar to case 1, case 2 achieved successful cochleostomy following a mastoidectomy on the right cadaveric specimen. Additionally, postoperative image data confirmed the successful placement of an implant wire phantom in the cochlea (Figure 4). Registration of endoscopy to CBCT using 3D-augmented reality of segmented models of vital anatomic structures (Figure 3C) was also visually confirmed after exposure of the facial nerve.

Each full surgical procedure was completed within a mean of 160 minutes. The custom drill adapter, which has a 30° offset and is coupled to the working da Vinci Si robotic arm, allowed for enough degrees of freedom to navigate around corners with the drill and complete all of the necessary motions required for drilling the temporal bone in a fluid natural fashion, not unlike freehand drilling. There was no injury or surgical breach of the tegmen tympani, sigmoid sinus, or facial nerve. After the facial recess/posterior tympanostomy was opened, the 3D endoscope camera provided an adequate view of the middle ear structures, including the stapes and the round window niche. Because of the restrictions related to magnification (dependent on the distance of the camera tip), this view of the middle ear was thought to be inferior to the standard stereo vision conferred by a binocular microscope. This will be addressed in subsequent studies by adjusting scope selection and the camera focal length. Finally, new suction and drilling equipment will be engineered. In the present case studies, the articulating suction irrigator was not small enough to insert through the recess. As a result, the surgeon had to physically attach and couple a 3F Frazier tip to a robotic side port instrument to suction in the middle ear. Another limitation was the size of the robotic arm tools (grasping forceps) that are not currently designed for otologic surgery in the middle ear, thereby making it difficult to complete middle ear work through the facial recess using the master-slave system. Manual refinement was needed to finalize the insertion of the phantom electrode. Despite these minor limitations, we were able, in both surgical cases, to complete the entire procedure without difficulty, including the cochleostomy, using the da Vinci Si system.

The da Vinci system has shown promise in surgery for many clinical indications, including prostate cancer, colorectal disease, renal disease, cardiac disease, and head and neck cancer. Although the da Vinci system has been primarily applied for handling soft tissue in anatomic regions that are difficult to access, the robotic arm has great potential for lateral skull base procedures, in particular, cochlear implant. Coupled with image guidance, we believe that the use of the da Vinci system has the potential to offer the surgeon enhanced surgical accuracy, allowing for more precise cochlear implant, particularly in patients with cochlear malformations, poorly pneumatized mastoid bones, or congenitally absent temporal bone landmarks, as well as those who have undergone previous ear surgery.

Research in computer-assisted approaches for cochlear implant includes custom robotic systems^12,17 and adaptations of industrial robots.^13,14,23 These groups have demonstrated the value of human-robot collaboration to filter hand tremor,¹⁷ to support motion and force scaling,¹⁵ to theoretically avoid skipped movement, and for past-pointing and precise microdrilling.^24,25 In the proposed approach using the da Vinci Si system, we take advantage of tool position stability as well as motion scaling and visual magnification. Other studies have focused on image guidance through path planning, which is especially significant for percutaneous approaches and implant electrode placement^20,24,25 with steerable arrays that provide force feedback. The approach described herein used augmented reality as a means of guidance. Rather than tracking a surgical tool to a registered radiologic image, the software allows the image of the critical structures to be “injected” and overlaid onto the live 3D endoscopic image in real time. To accomplish this, it is necessary to first segment the structures of interest from a CBCT scan, map these images to fiducials that are placed in the surgical field of view, and then overlay a 3D image of the critical structures onto the actual 3D endoscopic view that the surgeon sees in the console.

In contrast to percutaneous robotic methods and similar to traditional CI, the proposed approach of stereoendoscopy maintains the advantage of possible manual correction through visual cues, allowing the surgeon to maintain a high level of engagement in intraoperative decision making and dissection and reduce the risk of error with overreliance on a preplanned surgical route. Similar to the use of stealth navigation in sinus surgery, where image guidance has a more than 90% level of satisfaction with a reported accuracy of 1 to 3 mm in 79% of cases and only a few minutes needed for setup time,²⁶ the surgeon maintains direct visualization of the surgical workspace with enhanced navigation directly integrated to the primary field of view. Extension to the image guidance used can include path planning¹⁸ and virtual fixtures^17,27-30 to not only render critical structures but also improve safety and accuracy for surgical trajectories.

Appropriate cochleostomy placement is critical to allow for cochlear electrode array insertion into the scala tympani. Furthermore, cochleostomy placement anterior-inferior to the round window appears critical to ensure that all of the implant’s electrodes stimulate both apical and basal ganglion cells along the cochlea. However, studies⁵ suggest that a significant proportion of cochlear implant surgeons do not adequately position the cochleostomy anterior-inferior to the round window. Lack of familiarity with the facial recess can result in the surgeon leaving too much bone overlying the nerve undrilled, with an incomplete opening of the facial recess and poor visualization of the round window niche or membrane. Other potential factors that contribute to inadequate cochleostomy placement include variable round window anatomy, a poor angle of visualization approach through a restricted facial recess, and complex inner ear (cochlear/vestibular) malformations. Although in this pilot feasibility study we did not measure the angle of insertion of the phantom array, cadaveric studies are planned in which we will determine whether augmented reality can help predict the optimal angle and whether the position of the cochleostomy is adequate in all cases.

Suboptimal placement of the electrode too inferior to the round window increases the likelihood of surgical damage to auditory spiral ganglion neurons with subsequent decreased hearing performance, as well as the potential for vestibular stimulation, cerebrospinal fluid leakage, and false placement in hypotympanic air cells, rendering the implant nonfunctional.³¹ Although the incidence of each of these potential difficulties is rare, suboptimal placement of the electrode array is likely a contributing factor in cochlear implant soft failure rates, which can occur in up to 23% of implants.^32,33 Mitigating collateral forces resulting from electrode implant insertion on the cochlea is also a well-studied body of work.^34,35

Undoubtedly, careful evaluation of the relative anatomy of the facial nerve, chorda tympani, orientation of the cochlear basal turn, round window anatomy, and cochlear axis based on in-room tomographic images is beneficial to cochlear implant surgeons to reduce risk, improve safety, and minimize time under anesthesia. In addition, precise control of the surgical drill through a robotic arm interface along with image augmentation in the microscope field-of-view increases the confidence of the surgeon, reduces operative risk and operative time, and minimizes the risk of postoperative complications. A major safety benefit of this study would be to mitigate facial paralysis with cochlear implant surgery, which has devastating psychological consequences and an estimated incidence as high as 1.1%.³⁶

This initial cadaveric feasibility study shows that the master-slave approach with augmented reality is possible. The potential of this technology is vast. Indeed, if proved to be precise, reliable, and cost-effective, a surgical system that provides immediate intraoperative feedback of the anatomy would be quickly accepted by surgeons and patients, not only for cochlear implant but also for other neuro-otologic procedures, such as excision of cerebellopontine angle tumors, removal of petrous apex cholesterol granuloma, labyrinthectomy, and decompression of the endolymphatic sac in Meniere disease.

The lack of haptic feedback is a limitation of the da Vinci Si system. In standard master-slave–assisted laparoscopic procedures, surgeons rely on soft-tissue visual deformation to estimate the forces applied. Although the da Vinci Si workflow described in the present study is limited by the current lack of haptic feedback, this did not interfere with successful cochleostomy and mastoidectomy in both cases. Because the desired method of bone removal is to allow the high–revolutions-per-minute drill bit (10 000-30 000 rpm) to progressively and smoothly mill and remove bone, the available visual feedback of the interaction of the drill bit at the bony surface was found in these 2 cadaveric specimens to be adequate for completion of these cases. It is our opinion that, because we did not need to rigorously press the drill bit burr against bone and because visual feedback of the system was deemed sufficient, the need for haptic feedback was mitigated. Additionally, we thought that the stereoscopic high-definition video and the audiologic feedback of the drill pitch helped to convey drill contact forces and interactions.

The fabrication of the 30° drill guide attachment is a key component of the designed system. Without this 30° extension, there would be limited rotation of the robotic arm, which rotates the drill only in a single vertical axis and vector and decreases the degrees of freedom. Our 30° drill guide attachment to the robotic arm allows for rotation of the arm and the drill to occur in a circular fashion along a vertical axis, directing a variable vector of the drill (dependent on the rotation). This markedly improved the ability of the drill to navigate around corners and allowed for completion of the surgical drilling in a manner very similar to freehand drilling.

More studies need to be performed to refine the surgical equipment for general widespread clinical use. We plan to perform a comprehensive study to analyze and confirm the accuracy of augmented reality image overlay on the surgical view of a temporal bone anatomy. We found 3 limitations of the technology that need to be addressed before considering taking this approach to clinical use. First, a smaller profile-articulating suction-irrigation device (Figure 2) for navigation in the facial recess is necessary. Similarly, instruments small enough and specific for middle ear surgery need to be designed to operate, using the robotic arm, in the middle ear through the facial recess. At this point, microdissection in the middle ear with otologic instruments, including manipulation of the round window soft tissues, does not appear feasible given the access limitations. Second, improvement of the magnification of the 3D endoscope for improved visualization through the posterior tympanostomy is required because the view into the middle ear is better with classic binocular microscopy at this point. Third, once we have been able to complete this approach in more cadaveric specimens and have more adequately defined the accuracy, precision, and feasibility of the system, we plan to analyze the potential cost relative to the potential benefits of master-slave–assisted otologic surgery.

To our knowledge, this is the first study in the otolaryngology–head and neck surgery literature examining the use of a master-slave–assisted cochleostomy with augmented reality for cochlear implants using the da Vinci Si system. The assistance provided by this system for cochleostomy has the potential to improve the surgeon’s confidence, as well as the safety and precision of the procedure by filtering tremor, and facilitate integration of image guidance through video overlay. Guidance through augmented reality could improve outcomes in very young or otitis-prone patients with poorly pneumatized mastoids, in complicated revision cases, or in cases with complex or absent bony landmarks. Finally, image overlay facilitates intraoperative engagement of trainees and can help as a method of teaching through simulation and enhanced interaction.

Submitted for Publication: July 11, 2013; final revision received November 18, 2013; accepted November 24, 2013.

Corresponding Author: Diego Preciado, MD, PhD, Pediatric Hearing and Otological Research Center, Division of Pediatric Otolaryngology–Head and Neck Surgery, Children’s National Medical Center, 111 Michigan Ave NW, Washington, DC 20010 (dpreciad@cnmc.org).

Published Online: January 23, 2014. doi:10.1001/jamaoto.2013.6443.

Author Contributions: Dr Preciado had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Liu, Azizian, Sorger, Taylor, Cleary, Preciado.

Acquisition of data: Liu, Azizian, Sorger, Preciado.

Analysis and interpretation of data: Liu, Azizian, Taylor, Reilly, Preciado.

Drafting of the manuscript: Liu, Azizian, Preciado.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Liu, Preciado.

Obtained funding: Cleary, Preciado.

Administrative, technical, or material support: Liu, Azizian, Sorger, Reilly, Cleary, Preciado.

Study supervision: Cleary, Preciado.

Conflict of Interest Disclosures: Ms Liu is a predoctoral candidate completing a portion of her doctoral thesis work at the laboratories of Intuitive Surgical, Inc, the manufacturer of the da Vinci Surgical System. Drs Azizian and Sorger are research and development scientists employed at Intuitive Surgical, Inc. No other disclosures were reported.

Funding/Support: Funding for this project was provided in part through a gift from the government of Abu Dhabi to Children’s National Medical Center to establish the Sheikh Zayed Institute for Pediatric Surgical Innovation; in part through a fellowship from Intuitive Surgical, Inc; and in part by The Johns Hopkins University internal funds. Experimental facilities were provided by Intuitive Surgical, Inc.

Role of the Sponsor: Other than Intuitive Surgical, Inc, the sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: Linan Zhang, PhD, fabricated the drill attachment guide while a visiting researcher at Children’s National Medical Center. Dr Zhang received no financial compensation.