Pitfalls of Intraoperative Image-Guided Spinal Navigation
Pitfalls of Intraoperative Image-Guided Spinal Navigation
Prior to starting a spinal fusion procedure, various operative considerations such as positioning, neuromonitoring, and equipment require appropriate selection. The surgical team is responsible for dissemination of information to OR personnel, who will in turn organize the OR.
Preoperative patient factors include the potential difficulty of performing adequate imaging on obese and morbidly obese patients. The increased soft tissue in patients with morbid obesity may create difficulty with positioning, beam penetration, and the ability to maneuver imaging devices around the patients. This results in poorer quality images that can make the registration process inaccurate, as well making the images difficult to use during surgery.
Surgeon learning curves when new technology emerges have been reported in robotic and laparoscopic surgery. For spine surgery, the learning curve for adopting image-guided technology has been reported by Bai et al. and Sasso et al. The components of the learning curve include the ability to direct instruments based on imaging visualized on a screen, the ability to replicate in-line maneuvers while placing instrumentation, as well as adopting and developing proper technique while using image-guided technology. The ability of surgeons to adapt to the use of this technology may depend on a generation of surgeons who grew up playing video games. Rosser et al. was able to correlate faster completion and reduced errors in laparoscopic surgeries when the surgeons had a background of more than 3 hours per week of video game play.
Appropriate selection of an OR surgical table for image guidance is important. Modern spine surgical ORs have a variety of equipment at the surgeon's disposal. Different operating tables have advantages and disadvantages for surgical positioning, exposure, and intraoperative imaging. The dimensions and designs of regular OR tables may hinder one's ability to perform imaging with modern cone-beam CT–based imaging systems. Two-dimensional fluoroscopy is compatible with many OR tables as long as the patient is positioned appropriately. The Jackson table enables greater deformity correction and correction of alignment along the entire neuraxis, making it ideal for complex spinal fusions. The design of the O-arm allows it to work ideally with the Jackson table, which does not have a base obstructing movement along the long axis of the patient and table. The Jackson table enables the O-arm to be positioned along any level of the spinal axis. The table is well designed for imaging purposes, with its core structure such that the table has minimal radiodense metal resulting in minimal radiographic artifact.
Sterile draping is a consideration. Current draping of the fluoroscope is simple and standardized such that the OR staff is able to perform this seamlessly. However, the sterile cover for O-arm use remains cumbersome, at times getting caught between the shields. This has occurred during our initial learning curve with spinal navigation using the O-arm. Once the drape is caught, the failsafe prevents it from opening in an automated manner and manual maneuvers have to be used to open out the O-arm (Fig. 3). In our experience, an alternate draping of the patient in a 360° circumferential manner is more efficient and avoids problems while the O-arm opens and closes around the patient and operating table (Fig. 4). This method of draping could also be adopted for the new intraoperative CT scanners as well.
(Enlarge Image)
Figure 3.
Photographs demonstrating manually opening the O-arm. A: A sequence of steps at 3 points, all of which are visible in A, is required. B: Inserting a key into the appropriate slot is the first step. C: Inserting a manual driver into slot 1 enables closing of the gantry. D: Inserting the driver into slot 2 enables manual opening of the gantry in case of emergency or nonfunctioning automated mechanisms.
(Enlarge Image)
Figure 4.
Intraoperative photographs demonstrating avoidance of draping problems. A: Initial sterile tube drape that is used with the O-arm. The drape is cumbersome and redundant due to its exaggerated dimensions to accommodate the O-arm, which means that excessive draping would at times get stuck while the gantry tries to close and open. B: Our sterile draping technique requires filling the surgical wound with antibiotic irrigation solution and using surgical split-sheet drapes to circumferentially enclose the patient, keeping the reference arc just above the drapes. C: With gantry closure there is no contact between the patient, reference arc, and the tube. Following imaging, these drapes are removed and procedure continued.
The registration process following patient positioning is crucial. Inaccurate registration with the use of image guidance and computer-assisted navigation potentially has multiple sources of error. This necessitates frequent validation and accuracy assessment on a continuous basis. Early computer-based navigation systems required registration of preoperative images with the surgical intraoperative space. The margin of error for insertion of a pedicle screw depends mainly on the size of the pedicle, size of the screw, and distance to the isthmus or narrowest point on the pedicle. Rampersaud et al. evaluated error margins and demonstrated very small tolerances, with less than 1 mm permissible for translation and less than 5° for rotational changes in the cervical, thoracic, and thoracolumbar junctions. Glossop et al. demonstrated a clinical utility error range of 2–3 mm of translation and 4°–7° of rotation using image-guided navigation.
Intraoperative image acquisition and registration should therefore be performed after surgical access is completed. This avoids angular and translational movements that may alter the image registration. With conventional open midline access, image acquisition and registration should be performed after completing the approach to eliminate motion-related inaccuracy. In contrast, percutaneous minimally invasive instrumentation may begin immediately after the intraoperative registration procedure without prior anatomical landmark dissection, as significant manipulation of the surgical field and landmarks do not occur in these procedures.
Errors during image registration can occur. Patients undergoing surgery have 2 body warmers to prevent hypothermia, 1 placed over the upper body and the other over the lower body. During image acquisition for registration with cone-beam CT scans, the upper body warmer has to be turned off to avoid increasing imaging artifacts and creating registration errors. Additionally, movements during respiration and image acquisition can cause significant changes and inaccuracies with registration. The anesthesiologist has to specifically hold the patient's respiration during this step, minimizing errors secondary to motion artifact. Image acquisition is also performed after the dissection, and afterward the deep retractors are left in situ, especially for mobile cervical spine segments. The surgeon should avoid changing the position of the table (Trendelenburg position or the reverse) and attempt to avoid any potential movements with instruments that may cause distortion of anatomy, such as drilling and tapping all holes prior to instrumentation.
There is potential inaccuracy with increasing distance of screw placement from the reference arc. Quiñones-Hinojosa et al. analyzed 3D fluoroscopy–identified errors of accuracy at different distances from the reference arc and different time points when used in lumbar spine fusion surgery. They identified increasing distance from the reference arc and increased duration of surgery as the main factors, with inaccuracy of 3 mm in 7% of the patients when surgery was 3 levels away from the reference arc, and inaccuracy of 3 mm in 17% about 1 hour into surgery. Scheufler et al. performed pedicle screw instrumentation using a single-registration sequence in 32 (91.4%) of their 35 patients, with as many as 12 vertebrae instrumented after a single registration sequence. They identified a statistically insignificant increase in misplaced screws by about 2 mm at distances 10 segments between the instrumented segment and reference arc. Interestingly, Holly et al. demonstrated that navigation error is an interaction between technology and human factors, remaining the same independent of registration techniques.
Preoperative Considerations
Prior to starting a spinal fusion procedure, various operative considerations such as positioning, neuromonitoring, and equipment require appropriate selection. The surgical team is responsible for dissemination of information to OR personnel, who will in turn organize the OR.
Obese Patients
Preoperative patient factors include the potential difficulty of performing adequate imaging on obese and morbidly obese patients. The increased soft tissue in patients with morbid obesity may create difficulty with positioning, beam penetration, and the ability to maneuver imaging devices around the patients. This results in poorer quality images that can make the registration process inaccurate, as well making the images difficult to use during surgery.
Surgical Learning Curves
Surgeon learning curves when new technology emerges have been reported in robotic and laparoscopic surgery. For spine surgery, the learning curve for adopting image-guided technology has been reported by Bai et al. and Sasso et al. The components of the learning curve include the ability to direct instruments based on imaging visualized on a screen, the ability to replicate in-line maneuvers while placing instrumentation, as well as adopting and developing proper technique while using image-guided technology. The ability of surgeons to adapt to the use of this technology may depend on a generation of surgeons who grew up playing video games. Rosser et al. was able to correlate faster completion and reduced errors in laparoscopic surgeries when the surgeons had a background of more than 3 hours per week of video game play.
Surgical Table Selection
Appropriate selection of an OR surgical table for image guidance is important. Modern spine surgical ORs have a variety of equipment at the surgeon's disposal. Different operating tables have advantages and disadvantages for surgical positioning, exposure, and intraoperative imaging. The dimensions and designs of regular OR tables may hinder one's ability to perform imaging with modern cone-beam CT–based imaging systems. Two-dimensional fluoroscopy is compatible with many OR tables as long as the patient is positioned appropriately. The Jackson table enables greater deformity correction and correction of alignment along the entire neuraxis, making it ideal for complex spinal fusions. The design of the O-arm allows it to work ideally with the Jackson table, which does not have a base obstructing movement along the long axis of the patient and table. The Jackson table enables the O-arm to be positioned along any level of the spinal axis. The table is well designed for imaging purposes, with its core structure such that the table has minimal radiodense metal resulting in minimal radiographic artifact.
Sterile Draping
Sterile draping is a consideration. Current draping of the fluoroscope is simple and standardized such that the OR staff is able to perform this seamlessly. However, the sterile cover for O-arm use remains cumbersome, at times getting caught between the shields. This has occurred during our initial learning curve with spinal navigation using the O-arm. Once the drape is caught, the failsafe prevents it from opening in an automated manner and manual maneuvers have to be used to open out the O-arm (Fig. 3). In our experience, an alternate draping of the patient in a 360° circumferential manner is more efficient and avoids problems while the O-arm opens and closes around the patient and operating table (Fig. 4). This method of draping could also be adopted for the new intraoperative CT scanners as well.
(Enlarge Image)
Figure 3.
Photographs demonstrating manually opening the O-arm. A: A sequence of steps at 3 points, all of which are visible in A, is required. B: Inserting a key into the appropriate slot is the first step. C: Inserting a manual driver into slot 1 enables closing of the gantry. D: Inserting the driver into slot 2 enables manual opening of the gantry in case of emergency or nonfunctioning automated mechanisms.
(Enlarge Image)
Figure 4.
Intraoperative photographs demonstrating avoidance of draping problems. A: Initial sterile tube drape that is used with the O-arm. The drape is cumbersome and redundant due to its exaggerated dimensions to accommodate the O-arm, which means that excessive draping would at times get stuck while the gantry tries to close and open. B: Our sterile draping technique requires filling the surgical wound with antibiotic irrigation solution and using surgical split-sheet drapes to circumferentially enclose the patient, keeping the reference arc just above the drapes. C: With gantry closure there is no contact between the patient, reference arc, and the tube. Following imaging, these drapes are removed and procedure continued.
Registration Process
The registration process following patient positioning is crucial. Inaccurate registration with the use of image guidance and computer-assisted navigation potentially has multiple sources of error. This necessitates frequent validation and accuracy assessment on a continuous basis. Early computer-based navigation systems required registration of preoperative images with the surgical intraoperative space. The margin of error for insertion of a pedicle screw depends mainly on the size of the pedicle, size of the screw, and distance to the isthmus or narrowest point on the pedicle. Rampersaud et al. evaluated error margins and demonstrated very small tolerances, with less than 1 mm permissible for translation and less than 5° for rotational changes in the cervical, thoracic, and thoracolumbar junctions. Glossop et al. demonstrated a clinical utility error range of 2–3 mm of translation and 4°–7° of rotation using image-guided navigation.
Intraoperative image acquisition and registration should therefore be performed after surgical access is completed. This avoids angular and translational movements that may alter the image registration. With conventional open midline access, image acquisition and registration should be performed after completing the approach to eliminate motion-related inaccuracy. In contrast, percutaneous minimally invasive instrumentation may begin immediately after the intraoperative registration procedure without prior anatomical landmark dissection, as significant manipulation of the surgical field and landmarks do not occur in these procedures.
Errors during image registration can occur. Patients undergoing surgery have 2 body warmers to prevent hypothermia, 1 placed over the upper body and the other over the lower body. During image acquisition for registration with cone-beam CT scans, the upper body warmer has to be turned off to avoid increasing imaging artifacts and creating registration errors. Additionally, movements during respiration and image acquisition can cause significant changes and inaccuracies with registration. The anesthesiologist has to specifically hold the patient's respiration during this step, minimizing errors secondary to motion artifact. Image acquisition is also performed after the dissection, and afterward the deep retractors are left in situ, especially for mobile cervical spine segments. The surgeon should avoid changing the position of the table (Trendelenburg position or the reverse) and attempt to avoid any potential movements with instruments that may cause distortion of anatomy, such as drilling and tapping all holes prior to instrumentation.
There is potential inaccuracy with increasing distance of screw placement from the reference arc. Quiñones-Hinojosa et al. analyzed 3D fluoroscopy–identified errors of accuracy at different distances from the reference arc and different time points when used in lumbar spine fusion surgery. They identified increasing distance from the reference arc and increased duration of surgery as the main factors, with inaccuracy of 3 mm in 7% of the patients when surgery was 3 levels away from the reference arc, and inaccuracy of 3 mm in 17% about 1 hour into surgery. Scheufler et al. performed pedicle screw instrumentation using a single-registration sequence in 32 (91.4%) of their 35 patients, with as many as 12 vertebrae instrumented after a single registration sequence. They identified a statistically insignificant increase in misplaced screws by about 2 mm at distances 10 segments between the instrumented segment and reference arc. Interestingly, Holly et al. demonstrated that navigation error is an interaction between technology and human factors, remaining the same independent of registration techniques.
