Regarding the TREs at the different mandibular anatomical landmarks according to the types of registration marker used in the surgical navigation process, the smallest errors were found when the Aurora EM tracker was used with invasive markers. Based exclusively on the marker methods, the invasive markers (screws) used as registration markers exhibited smaller TREs than the noninvasive markers (gutta-percha points).
For navigation surgery, two methods are available for registering images and the physical body, viz., marker-based and marker-free registration.6,7 The technique of placing screws as applied in the present study is an invasive method, which demonstrated fewer errors than using noninvasive markers with gutta-percha. Although it may be necessary to evaluate whether the gutta-percha used in the present study demonstrates sufficient functionality as a marker, noninvasive markers with gutta-percha can be an alternative marker method in mandibular surgery.
Marker-based registration uses reference markers that are clearly visible on CT images and easily identifiable for registration. In contrast, marker-free registration uses anatomical landmarks based on the anatomical structure of the patient or laser surface scanning for registration. To determine which method is more accurate, additional studies with varying surgical sites and methods are required. Using noninvasive markers in surgeries could be helpful for patients; therefore, further studies are necessary to assess marker methods optimized according to the patient and surgical characteristics and based on the additional studies on various noninvasive markers.
In the present study, TREs were derived and compared using optical and EM trackers, which are the tracking methods used in the surgical navigation system. The results demonstrated that using the Aurora EM tracker exhibited fewer errors than using the Polaris optical tracker. The Polaris optical tracker used in the present study was a Vicra version manufactured for use in maxillofacial surgery. According to the manufacturer’s specifications, the volumetric accuracy of the tracker is 0.25 mm RMS with a 95% confidence interval of 0.5 mm RMS. In the Aurora EM tracking system, planar field generator and 6-degree of freedom sensors were used, and the experiment was conducted at cube volume. The tracker had positional and orientation accuracies of 0.48 mm RMS and 0.30°, respectively, with 95% confidence intervals of 0.88 mm RMS and 0.48°, respectively.
The optical tracking system has an approximately equal accuracy to that of the EM tracking system.8 However, the biggest disadvantage of using an optical system is that all the trackers in all the cameras used require a direct line of sight for all targets. Hence, even a slight slope between the probe tip and the tracker could cause an error. Moreover, an optical tracking system may not be suitable for tracking needles or flexible objects.
For safe navigation surgery, tracking errors must be considered and reduced. Optical navigation systems could provide an accuracy of 0.1–0.4 mm for positional measurements, but the optical camera requires warming up for 15–30 min before the start of the surgery.9 The trackable field size is approximately 100 × 100 × 100 cm. Errors may occur when the vector angle between a stereo optical camera, probe, and surgical instrument is ≥ 60°.9 In the present study, the results may have been influenced by the position and angle of the probe and the size, angle, and length of the optical tracker. However, it was determined that such factors would have very little influence on the primary objective of the present study, which was to investigate the differences according to the type of registration markers and tracking methods.
Compared with optical navigation systems, EM navigation systems have a narrow measurement field, but they have the disadvantage of errors caused due to metallic instruments.10 In this study, the Aurora EM tracker exhibited fewer errors, especially for the mandibular condylar head that showed the largest error with the optical tracking system, where the mean error was 1.3 mm with invasive markers and 2.1 mm with noninvasive markers. In contrast, the Aurora EM tracker with invasive markers exhibited a significant difference of 0.9 mm. A previous study that investigated errors in a plastic skull model using a conventional EM navigation system reported that positioning of the condyle error consisted of a positional error of 0.65 mm and a rotation error of 0.38°.11
Because several surgical instruments are made of metal, there is a need for additional studies on the errors caused due to metallic instruments used during surgery when using the EM method on the mandible. In addition to the errors caused by metallic surgical instruments, errors could occur during the registration of information about the surgical equipment. Technical errors may occur in the process of the tracking camera actively (emission detection diode) or passively tracking the dynamic RF attached to the patient and surgical instruments. When RFs are directly attached to the surgical instruments or probes that have been registered in the system program with the navigation system already calibrated, the calibration process is required and the position is subsequently tracked by the camera. Navigation errors may occur if the probe is bent, the information about the instrument is incorrect, or if an error occurred in the calibration process.
Errors in navigation surgery systems may occur due to various factors, including errors during the process of using a computer to superimpose digital images.9 Technical measurement errors could occur in the proprietary software and hardware, such as a CT scan. There could also be imaging errors according to the software and imaging mode, such as the matrix size, slide thickness, and voxel size. Regarding registration errors, the process of registration after matching the coordinates of imaging data with the anatomical structures of a patient in the navigation system has the largest impact on errors in image-guided surgery. Accordingly, the present study aimed at reporting the registration errors with TREs according to the registration marker and the tracking methods.
Registration errors are those that are generated when the patient’s imaging data are linked to the actual physical anatomical parts of the patient.12 There are paired fiducial points, which are the points determined to represent the same area in the imaging data and the actual patient, and the error that occurs when finding such fiducial points is referred to as fiducial localization error. Furthermore, the error that occurs when the image and the physical body are superimposed during fiducial registration between the image and the patient is referred to as fiducial registration error, whereas the error representing the difference between the anatomical registration points and the actual coordinate values after image and body registration is referred to as TRE. This study investigated the TREs that could occur during navigation surgeries.
The application of the navigation system differs according to the surgery.4,13 It is necessary to select markers with fewer errors and set their positions, as well as set the RFs that do not interfere with the surgery. The mandible is connected to the cranium by the temporomandibular joint, which is an independent structure as it is flexible and performs significant movements. In mandibular navigation surgery, the mandible remains in the open state or there is traction during surgery, which may cause deviation; therefore, it is important to use reference points on the mandible itself.7 Accordingly, in the present study, markers on the mandible itself were used, and the RFs were located on the mandible. Furthermore, the RFs were placed on the mentum in the center of the mandible to prevent lateral error. Firm fixation and prevention of rotation are required for RFs. An optical tracker camera was used to fix the angle so that it could be recognized on the screen.
However, in actual surgeries, continued real-time monitoring may be difficult because the head of the surgeon attempting to view the surgical site could block the RF recognition path, unlike the experimental environment used in the present study. Using a navigation system may initially consume a significant amount of time due to a learning curve, but this would eventually decrease. It is necessary to establish a registration method with the least amount of registration errors for the position and number of markers when positioning the RF as suitable for the mandible.
In the present study, the number of markers for superimposing teeth images was set to five. To address errors occurring during registration, at least three nonplanar markers are generally required. The markers used in registration should be placed widely around the anatomical structure in the surgical site and close to the surgical site to ensure high accuracy.14 In actual surgeries, the use of markers for superimposition may be limited. The use of invasive markers and the positions of the markers should be decided by considering the surgical site and its anatomical structures. The head and neck areas have the characteristic of allowing the use of a bit splint, unlike registration in other surgical areas using mouth and teeth.15 Noninvasive marker-based registration could be performed by attaching the markers on a template used in navigation surgery. An external RF that uses fiducial markers in denture-fixed acrylic template for mandibular navigation has been reported to demonstrate a registration accuracy that is similar to that of invasive markers placed on the bone but does not entail the invasive process of placing screws as markers. However, it has the disadvantage of fabricating the marker template before imaging and biting down on the template during the CT scan.
The navigation software 3D Slicer supports multipurpose visualization, and it is based on a free open source software platform for medical image computing that provides several high-end functions for this purpose.16 In particular, it provides various applications for using medical images. The 3D Slicer is connected to IGTs through the Open IGT link, which uses the TCP/IP network communication protocol to enable real-time navigation function by sending and receiving data with various types of tracking systems in the operating room. Therefore, the Polaris optical and Aurora EM trackers could be used to realize the navigation surgery method. Regarding the experimental workflow, mandibular CT data in DICOM format could be imported into the Slicer program, and markers attached to the mandibular model could be used for registration after rendering to the 3D format.