The Laryngoscope Lippincott Williams & Wilkins, Inc.

, Philadelphia 2000 The American Laryngological, Rhinological and Otological Society, Inc.

Efficacy of Computed Tomographic ImageGuided Endoscopic Sinus Surgery in Residency Training Programs
Roy R. Casiano, MD; William A. Numa, Jr., MD

Objective: To determine the efficacy of computed tomographic image-guided endoscopic surgery in the hands of inexperienced surgeons. Study Design: Four second-year otolaryngology residents, with no prior experience performing ethmoidectomies, performed endoscopic sinus surgery (ESS) on formalin-fixed human cadaveric specimens with and without the aid of computer-assisted surgery (CAS). Methods: Each resident was asked to identify critical sinus, orbital, and skull base structures while performing a total ethmoidectomy and multiple sinusotomies. Their surgical accuracy (percentage of correctly identified structures), total operative time, and incidence of major complications were recorded for each side. A total of 16 sides were evaluated (8 with and 8 without CAS). Statistical significance between groups was determined by means of Pearsons 2 analysis. Results: Statistical analysis showed a significant difference (P .001) in the mean accuracy of identifying critical anatomical landmarks between the CAS (97%) and non-CAS (76.8%) groups. Although not statistically significant, operative time appeared to be longer in the group using CAS (average of 67 vs. 80 min). Three major intracranial complications were documented only in the group not using CAS. Conclusions: Although, unquestionably, a thorough knowledge of the anatomy remains essential for performing ESS, CAS improves surgical accuracy and reduces the risk of major intracranial or intraorbital complications for residents. In additional, our data suggest that this technology may enhance surgical efficiency and improve the learning curve by reducing operative time (below ones normal baseline) while maintaining a greater than 90% accuracy in identifying critical anatomical landmarks. Key Words: Computed tomogra-

phy, endoscopic sinus surgery, computed-assisted surgery. Laryngoscope, 110:12771282, 2000

INTRODUCTION
Since first described in 1993, intraoperative computerassisted surgery (CAS) has been advocated to improve the efficacy of endoscopic sinus surgery (ESS) and reduce complications.110 Assuming appropriate calibration and utilization, this technology has the theoretical potential to enhance the surgeons surgical skills. There may also be an educational benefit for residents, fellows, and practicing otolaryngologists. However, although it should be intuitive that CAS complements current endoscopic instrumentation, there is little scientific evidence confirming or refuting its efficacy in improving surgical accuracy and clinical outcomes or reducing complications.11 To date, the vast majority of studies have dealt predominantly with the accuracy of a number of commercially available CAS devices and accessories. Image-guided surgerys efficacy may be assessed by a variety of methods depending on which end point one measures. It may also be measured by the ability of the instrumentation to enhance the surgeons skills by assisting him or her to more precisely remove tissue and complete the proposed surgical procedure safely. It may also be measured by the advantage it offers in reducing operative time, improving clinical outcomes, reducing the number of revisions performed, or minimizing complications. It might have an additional educational benefit. However, the latter is one of the most difficult things to measure objectively. The efficacy of CAS may also differ depending on the degree of experience (i.e., resident in training, community practice, or academic practice) and/or the degree or type of disease (i.e., primary cases, revision cases, inflammatory or neoplastic disease). This study was undertaken to evaluate the efficacy of CAS when used by a group of inexperienced surgeons. To ensure comparable levels of experience with ethmoid surgery, second-year residents were used as a study group. For the scope of this study, efficacy was defined as a reduction in major complications, improved accuracy in identifying key anatomical landmarks, and improved operative time. Casiano and Numa: Computed Tomography

Presented at the Meeting of the Southern Section of the American Laryngological, Rhinological and Otological Society, Inc., St. Pete Beach, Florida, January 1315, 2000. From the Department of Otolaryngology, University of Miami School of Medicine, Miami, Florida. Editors Note: This Manuscript was accepted for publication April 11, 2000. Send Correspondence to Roy R. Casiano, MD, University of Miami Hospitals and Clinics, 1475 Northwest 12th Avenue, Suite 4025, Miami, FL 33136. U.S.A.

Laryngoscope 110: August 2000

1277

MATERIALS AND METHODS

Four residents with no prior experience performing an ethmoidectomy (external or endoscopic) performed ESS on eight formalin-fixed human cadaver heads (16 sides). All of the residents had at least 1 year of experience performing more limited endoscopic procedures such as endoscopic debridement, septoplasty, biopsy, turbinoplasty, and/or middle meatal antrostomy. All participated in a formal ESS dissection course (given annually) within 2 weeks before commencing the study. Each resident received instructions and completed the entire surgical procedure at least once before the study. They also received an in-service training on the use of the CAS device, because none of them had prior experience with CAS. The computed tomography (CT) scan on every specimen was reviewed with each resident before proceeding with the dissection. Each resident performed the surgery once with and without the aid of CT imaging (Xomed Landsmarx System, Jacksonville, FL). A total of four sides were performed, two with and two without CT imaging. The order was randomized with respect to the side operated on and whether CT imaging was used. Each side was timed independently from the start of the procedure to the successful completion of the assigned dissection as determined by the resident surgeon. The resident surgeon was instructed to perform a complete ethmoidectomy (removal of all ethmoid cells), middle meatal antrostomy, and a wide sphenoid and frontal sinusotomy, and to identify critical skull base and orbital landmarks. A checklist was used to grade the surgeon on his or her ability to achieve visualization of these structures in the assigned order (Table I). None of the resident surgeons was coached or supervised. An experienced ESS surgeon (R.R.C.) was present only as a passive observer to verify and record the successful completion of each portion of the dissection with identification of the key anatomical landmarks. No feedback was provided to the residents at any time during the course of the study. At the completion of the surgical procedure the residents were dismissed, the specimen was carefully examined on each side for the presence of major orbital or skull base penetrations, and the findings were recorded. Major penetrations consist of any violation of dura or periorbita with or without visualization of orbital fat or brain parenchyma. The location, size, and number of penetrations for each side were tabulated. At the conclusion of the study the resident surgeons were asked to fill out a confidential survey rating their experience with the CT imaging device (Table II).

RESULTS
The items identified with and asterisk in Table I represent the final list of anatomical landmarks that is the basis of the accuracy rating. These landmarks were incorrectly identified on at least one occasion either with or without the use of CAS. The remaining items were correctly identified 100% of the time by all of the residents with or without the use of CAS. The mean accuracy rate for the 16 operated sides was 87% (range, 58%100%). The mean accuracy rate was 76.8% (range, 58%97%) for the group operating without CAS and 97% (range, 88%100%) for the group operating with CAS (P .001). The entire group had a mean operative time of 74 minutes per side (range, 42128 min). The mean operative times were 67 minutes (range, 42 89 min) for the group without CAS and 80 minutes (range, 59 128 min) for the group dissecting with CAS (P .187). Although not statistically significant, CAS appeared to accelerate the learning curve, affording improved surgical efficiency (reduced operative time with improved accuracy) as the surLaryngoscope 110: August 2000

geon proceeded with subsequent surgical dissections (Figs. 1 and 2). Certain anatomical landmarks were more accurately and consistently identified with CAS (Table I). These landmarks were the maxillary ostium (P .026), the lateral wall of the infundibulum (P .041), and the lamina papyracea at the level of inferior posterior ethmoid cavity (P .038). Although falling short of reaching statistical significance, several other landmarks showed a trend toward being more accurately identified with CAS. These landmarks were the agger nasi (P .141), anterior ethmoid lamina papyracea (P .233), anterior ethmoid artery (P .285), posterior ethmoid artery (P .338), anterior ethmoid roof (P .233), the ability to accurately localize the sphenoid sinus inferomedially (P .2), and the ability to perform a wide-open sphenoidotomy (P .165) by removing its common wall with the posterior ethmoid sinus cavity. There were three major complications committed by resident surgeons on separate specimens. They were all inadvertent intracranial penetrations. The point of penetration was at the posterior ethmoid roof in one, the frontal recess in one, and the cribriform plate in another specimen (two left-sided and one right-sided specimen). All of these major complications occurred in the group not using CAS. Two specimens had posterior septal perforations. Both of these occurred in the group using CAS. In these two cases the residents failed to activate the device to verify their location, straying medially after identifying the skull base and orbit with CAS, because they were not working in real-time mode. There were no major orbital complications. On the survey rating the usefulness of the device, residents uniformly agreed that CAS is a useful adjunct to ESS. They all agreed that the device enhanced their educational experience at least slightly. All but one believed that the device improved their endoscopic surgical skills, the precision of tissue removal, and their ability to identify critical anatomical landmarks. Only two residents (50%) believed that the use of CAS had only a slightly protective effect on the incidence of serious complications. None of the residents believed the device worsened any of the survey parameters.

DISCUSSION
The ideal CAS device should correlate well to the anatomy being operated on. To accomplish this, it must be accurate within at least 2 mm to minimize the risk of complications from the device itself. In addition, the CAS device should not be affected by head movement or require a second CT or magnetic resonance imaging (MRI) scan; it should require minimum training for its use or calibration, should be cost-effective (current price, approximately $150,000), and should be easily operated by the surgeon. The latter eliminates the need for additional operating room personnel whereby cost may be increased. Ideally, intraoperative image guidance should also provide a realtime image to minimize the chance of intraoperative distortion of tissue by the surgeon. This is especially important for surgeons performing more advanced procedures of the orbit, skull base, or brain. Casiano and Numa: Computed Tomography

1278

TABLE I. Anatomical Landmark and Percent Accuracy With and Without Computer-Assisted Surgery (CAS).
Structure CAS No CAS

Nasal septum Inferior turbinate Middle turbinate Superior turbinate Supreme turbinate Sphenoethmoidal recess structures Sphenoid recess and ostium Onodi recess Posterior ethmoidal recess Nasopharyngeal and choanal structures Rosenmullers fossa Torus tubaris Posterior septum Reduce inferior turbinate and find nasolacrimal system Sac area Osseus canal Membranous canal and Hasners valve Reduce middle turbinate head and find osteomeatal complex structures Ethmoid bulla Retrobullar space Sinus lateralis Hiatus semilunaris Infundibulum Uncinate process Superior uncinate attachment Posterior fontanelle Accessory ostium Infundibulotomy and uncinectomy Maxillary ostium* Lateral wall of infundibulum* Ethmoid bulla Anterior wall* Suprabullar space* Anterior ethmoid lamina papyracea* Inferior posterior ethmoidectomy at junction of horizontal and vertical crest/antrostomy* Posterior ethmoid lamina papyracea* Sphenoid sinusotomy at floor of orbit, 7 cm, adjacent to septum* Sphenoid widely opened* Lateral sphenoid structures identified Area of carotid artery* Area of optic nerve (canalicular portion)* Sphenoid roof* Posterior ethmoid roof* Posterior ethmoid artery* Anterior ethmoid roof* Anterior ethmoid artery* Most superior infundibular and suprabullar cell* Agger nasi cell* Frontal recess/ostium widely opened and sinus clearly transilluminated*

100 100 100 100 100 100

100 100 100 100 100 100

100

100

100

100

100

100

100 100 100 100 87.5 100 87.5 100 100 100 100 100 100 75 100 100 75 100 87.5 87.5 87.5 100

100 100 100 62.5 25 87.5 75 75 50 50 62.5 50 75 75 87.5 75 62.5 75 62.5 62.5 50 75

*Items identified with an asterisk represent the final list of anatomical landmarks that is the basis of the accuracy rating.

Laryngoscope 110: August 2000

Casiano and Numa: Computed Tomography

1279

TABLE II. Questionnaire: Personal Ratings of Efficacy. Compared with doing surgery without an imaging device, how has CT imaging made a difference for each of the following? Incidence of serious complications (i.e., inadvertent penetration of orbit or skull base) Endoscopic surgical skills Precision of tissue removal Ability to identify assigned landmarks Intraoperative time Educational benefit
Responses were on a scale of 1 to 5: 1 significantly improved; 2 slightly improved; 3 no change; 4 slightly worsened; 5 significantly worsened.

Currently, two types of tracking technologies have been commercially available in the United States: optical and electromagnetic. With appropriate calibration the mean error for most of these products is approximately 2 mm. Electromagnetic digitizers (i.e., Instatrak, Visualization Technology, Inc., Boston, MA) superimpose a magnetic field around the area of interest. The position can be determined by using a probe that can actually detect gradients in the magnetic field. Ferromagnetic substances, aluminum, and electromagnetic radiation all distort the magnetic fields and impair the accuracy of localization. However, many of these initial problems have been solved. One advantage with the newer electromagnetic systems is that the CT scans can be performed at any time before surgery rather than on the previous or same day. A preoperative plastic headset serves as a reference and is positioned before the preoperative CT scan. Reregistration for head movement is generally not necessary with the newer systems. A potential disadvantage is the weight of the probe attachment. This may vary the weight and balance of the surgical instrumentation. Optical digitizers (i.e., Flashpoint 5000 3D localizer, Boulder Software Foundry, Boulder, CO, and Northern Digital OPTOTRAK System, Waterloo, Ontario, Canada) rely on infraredemitting diodes placed on the operating probe or instrument in a known geometric pattern. A channel array near the operating table detects the position of the infraredemitting diodes. This obviates the need to immobilize the head during surgery or re-registration for intraoperative head movement. Disadvantages of this system include the need for a direct line of sight between the infraredemitting diodes on the surgical instrumentation and the optical camera. Care must be taken to ensure that the diodes are in full view of the camera at all times. There have been many proponents for the use of CAS.37 They have noted that the duration of surgery is not significantly increased once the nursing staff or physician gain experience in the setup and calibration of the instrumentation. Intraoperative safety is improved with these procedures, especially in surgery of the anterior skull base with the endonasal approach. There have also been no reported complications while using the device. Suggested indications for the use of CAS include revision surgery, massive disease, disease within the sphenoid or Laryngoscope 110: August 2000

frontal recess, and the presence of Onodi cells or other anomalies that can lead to intraoperative complications. In additional, proponents of CAS claim that the device allows the surgeon to improve the thoroughness of the surgical procedure by detecting and removing disease from areas that are difficult to visualize (i.e., hidden or unopened ethmoid cells). However, despite these rather anecdotal comments, the proponents of CAS fail to show any data supporting their statements. The only study attempting to assess the efficacy of image-guidance technology was performed by Fried et al.8 in 1997. They reviewed their experience with the Instatrak System on 55 patients. In this multicenter study, the surgeons using the device were asked to rate their opinions on a five-point scale about how the device affected the surgical procedure (ranging from 2 [decreased very much] to 2 [increased very much]). The average response was 1.78 for revision cases and 1.53 for primary cases. There were no negative responses indicated by any surgeon in this study. According to the authors, this indicates that the surgeons believed the device was somewhat more useful for revision cases. The authors concluded that it is hoped that the use of image-guidance will reduce complications and the number of revisions required, resulting in lower total costs to the patient and society. However, the authors stated that this technology should not be relied on as the sole means of localization. Knowledge of anatomy is the basis of functional ESS and must be obtained by cadaver dissection and experience. One study was equivocal with predominantly negative comments in its evaluation of image-guidance technology. Roth et al.9 in 1995 evaluated the advantages and disadvantages of three-dimensional CT intraoperative localization for functional ESS. They performed a retrospec-

Fig. 1. Mean accuracy (solid line) and dissection times (dashed line) for all subjects under both conditions. Notice the steadily increasing identification accuracy rate and the simultaneously decreasing dissection time, presumably accounting for the expected learning curve.

Casiano and Numa: Computed Tomography

1280

Fig. 2. Top. Mean accuracy (solid line) and dissection times (dashed line) for all subjects dissecting without the aid of computed tomography (CT) guidance. Although identification rates and dissection times do improve, a suboptimal final identification accuracy rate was obtained when compared with the cohort dissecting with CT guidance. The mean dissection time was attained after two dissections. Bottom. Mean accuracy (solid line) and dissection times (dashed line) for all subjects dissecting with the aid of CT guidance. Notice the steep improvement with time in the identification rates and in time of dissection. An optimal identification accuracy rate was obtained in the final dissections compared with the cohort dissecting without CT guidance. The mean dissection time was attained after one dissection.

intraoperative localizations but were identified as potential good candidates during the course of surgery. These patients were found to have significant scarring of the skull base or posterior ethmoid at the time of surgery. In that study preoperative nasal endoscopy and CT scan failed to predict which patients were potential candidates for intraoperative localization. The authors noted that the extent of disease is not a good indication for intraoperative localization in experienced hands. The authors concluded that a thorough understanding of paranasal sinus anatomy and complexities, along with surgical experience gained through laboratory dissection, remains the most important factor for decreasing operative morbidity. Other than cost, there were few disadvantages to having image guidance in the operating room. Only one study cited a significant disadvantage. Drake et al.10 reviewed their experience with the ISG Technologies (Mississauga, Ontario, Canada) viewing wand in 1994. They used this technology predominantly for neurosurgical procedures and noted a fundamental limitation having to do with frame distortion or movement. They noted that the brain is deformable during surgery and localization becomes less accurate as the tissues are manipulated and surgically altered. Because the images are preoperative, the accuracy of the data are progressively degraded. They concluded that intraoperative localization based on preoperative CT scan are largely to be used for planning and initial trajectory. Differences in opinion regarding the efficacy of image-guidance technology may reflect this fundamental lack of scientific evidence. They may also reflect differences in the degree of expertise with endoscopic surgical instruments or with the complicated anatomy of the paranasal sinuses, orbit, and skull base. What value imageguidance may have in the hands of well-trained surgeons trained at residency or fellowship programs having a structured curriculum in anatomically based ESS without the use of image-guidance technology is unknown. Except in select cases, surgeons with more experience may find the instrumentation cumbersome and not worth the expense with little added value. On the other hand, surgeons with less experience may find the instrumentation useful. It may even give them a greater sense of confidence to deal with more complicated cases. However, primary reliance on intraoperative image guidance rather than knowledge of surgical landmarks could also lead to complications. Without solid scientific evidence confirming or refuting its efficacy, increasing pressures to reduce health care costs may make it more difficult to justify the cost of intraoperative image guidance to hospitals, insurance plans, and patients.

tive review of 220 patients over a 1-year period who underwent ESS and identified 12 patients (5%) who underwent ESS in conjunction with intraoperative localization. The authors noted that the average additional time for registration was 20 minutes with an average additional patient cost of $1900, which reflected the need for an additional CT, technician time, and threedimensional reconstruction. The cost did not vary based on the type or indication of surgery. The authors noted that eight additional patients (4%) were not selected for Laryngoscope 110: August 2000

CONCLUSION
This study suggests that CAS improves surgical accuracy. Despite a limited sample size, our results suggest that CAS may actually enhance surgical efficiency by reducing operative time (below ones normal baseline) while maintaining greater than 90% accuracy of identification. The overall operative time remains slightly higher with CAS, although it fails to reach significance when one takes Casiano and Numa: Computed Tomography

1281

into account the improved efficiency. However, the use of CAS assumes that the surgeon can interpret the CT anatomy of the paranasal sinuses well. In additional, the operating surgeon needs to use the device during the course of surgery, not only when he or she thinks it is needed. This is particularly important for the inexperienced surgeon who may not be totally familiar with the endoscopic anatomy of the paranasal sinuses. Waiting to activate the device when one feels lost may be too late. The two septal perforations are an example of complications from not activating the CAS device and verifying ones location. For both cases the residents strayed medially after identifying the skull base and orbit, failing to stay in real-time mode. There is a learning curve associated with the familiarization with the device itself that may initially increase the operative time. However, this event appears to be short-lived.

BIBLIOGRAPHY
1. Zinreich SJ, Tebo SA, Long DM, et al. Frameless stereotaxic integration of CT imaging data: accuracy and initial applications. Radiology 1993;188:735742. 2. Anon JB, Klimek L, Mosges R, Zinreich J. Computer-assisted endoscopic sinus surgery an international review. Otolaryngol Clin North Am 1997;30(3):389 401.

3. Mosges R, Klimek L. Computer-assisted surgery of the paranasal sinuses. J Otolaryngol 1993;22:69 71. 4. Anon JB, Lipman SP, Oppenheim D, Halt RA. Computerassisted endoscopic sinus surgery. Laryngoscope 1994;104: 901905. 5. Klimek L, Mosges G, Laborde G, Korves B. Computerassisted image-guided surgery in pediatric skull-base procedures. J Pediatr Surg 1995;30(12):16731676. 6. Fried MP, Kleefield J, Jolesz FA, et al. Intraoperative image guidance during endoscopic sinus surgery. Am J Rhinol 1996;10(6):337342. 7. Gunkel AR, Freysinger W, Thumfart WF. Computer-assisted surgery in the frontal and maxillary sinus. Laryngoscope 1997;107:631 633. 8. Fried MP, Kleefield J, Gopal H, Reardon E, Ho BT, Kuhn FA. Image-guided endoscopic surgery: results of accuracy and performance in a multicenter clinical study using an electromagnetic tracking system. Laryngoscope 1997;107: 594 601. 9. Roth L, Lanza DC, Zinreich J, Yousem D, Scanlan KA, Kennedy DW. Advantages and disadvantages of threedimensional computed tomography intraoperative localization for functional endoscopic sinus surgery. Laryngoscope 1995;105:1279 1286. 10. Drake JM, Rutka JT, Hoffman HJ. ISG viewing wand system. Neurosurgery 1994;34(6):1094 1097. 11. Casiano RR. Intraoperative image-guidance technology. Arch Otolaryngol Head Neck Surg 1999;125:12751278.

Laryngoscope 110: August 2000

Casiano and Numa: Computed Tomography

1282