Optimal fiducial marker distribution for navigated tilted implants in edentulous maxillae: an in vitro study Authors: Jinyan Chen, Guanyu Su, Ziang Wu, Feng Wang Published: 2025 ABSTRACT BackgroundTo determine the most appropriate fiducial markers distribution strategy in navigation surgery for three different tilted implant scenarios by comparing the target registration error (TRE) of tilted implants in fully edentulous maxillae.MethodsThe cone beam computed tomography scanned a maxillary edentulous model with 16 bone-anchored mini-screws (8 for registration and 8 for target registration error measurement), which was further imported into the navigation system for registration and target registration measurement. Each of the 5 strategies was performed six times, and the measurements were repeated twice around 4 target points each time. A mixed linear model was utilized for data analysis.ResultsRegistration distribution, measurement site, and their interaction had a statistically significant influence on TREs (P = 0.010, 0.000, and 0.001, respectively). Two markers on the anterior alveolar crest, 2 in the buccal of the anterior alveolar ridge, and 2 on both sides of the maxillary tuberosity can realize the minimal TREs for tuberosity-pterygoid-palatine implant (TPP) at both coronal and apical points. No significant TRE differences were found for apical points of mesial tilted implant in ‘all on four’ and zygomatic implants among the 5 registration strategies.ConclusionsFor TPPs, incorporating fiducial markers on maxillary tuberosity for registration can help reduce the TRE in dynamic navigation.Trial registrationNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12903-025-07113-9. Background Atrophic maxilla presents one of the most prevalent clinical scenarios where implant placement can prove challenging. Based on the quantity of residual bone in the edentulous maxilla, various classifications of implant treatment, with or without immediate loading protocols, have been suggested. Several tilted implant techniques have been introduced to achieve dense cortical bone engagement, including the ‘all-on-four’ concept,1,2 TPP,3,4 and zygomatic implant (ZI) [5, 6]. These techniques can shorten treatment duration, facilitate immediate loading, and eliminate the need for grafting procedures [1–7]. The aforementioned tilted implant techniques share several common characteristics: semi-blind surgery, a narrow surgical area, proximity to critical anatomical structures, and a longer drilling path than conventional implant protocols [8, 9]. However, with computer-assisted implant surgery (CAIS), surgeons can position these implants more accurately in three dimensions [10–15]. This precise and individualized placement minimizes the risk of intraoperative surgical complications, while effective utilization of residual bone enhances the possibility of immediate loading [10, 16]. Dynamic computer-assisted implant surgery (dCAIS) represents a clinically dependable approach for implant placement. Recent research has demonstrated that dCAIS can significantly surpass the accuracy of freehand surgery, yielding outcomes comparable to those achieved with static computer-assisted implant surgery (sCAIS) [17, 18]. The accuracy of dCAIS is influenced by numerous factors. The inherent system accuracy is determined by the tracking system’s precision and the software’s algorithm [19]. Clinically relevant variable parameters include the image quality of cone beam computed tomography (CBCT) and the registration process [20–23]. Since enhanced image quality leads to a higher radiation dose for the patient, optimizing the registration process—establishing the spatial relationship between the virtual coordinate system and the intraoperative patient coordinate system [24]—is the most effective way to improve the accuracy of dCAIS. For a fully edentulous maxilla, invasive fiducial markers (bone-anchored mini-screws) are typically fixed in the jawbone to aid in registration during the CBCT scan [22, 25]. However, in severely atrophied maxillae, factors such as maxillary sinus pneumatization and osteoporosis can hinder the placement of these fiducial markers. Additionally, the surgeon must adhere to the West principle while ensuring that the registration screws do not obstruct the surgical path [26]. Consequently, careful consideration is essential when determining the location, distribution, and number of fiducial markers before performing dCAIS. When planning a tilted approach for implant placement in edentulous maxillae, the increasing length and varying angle of the implant cause the target point to move farther from the registration center. As a result, the distribution of fiducial markers suitable for conventional axial implants may not be appropriate in this scenario, potentially leading to significant registration errors. Therefore, proposing a registration strategy that considers the amount of bone available in atrophied edentulous maxillae is crucial, ensuring both clinical logic and feasibility. While previous research has explored registration strategies for ZIs in fully edentulous maxillae through both in vitro and clinical studies, other tilted implant protocols have yet to be comprehensively investigated [27–29]. In this model study, three tilted implant approaches—mesial tilted implant in ‘all on four’, TPP, and ZI—were virtually planned. Additionally, several clinically feasible registration strategies were proposed to identify a scheme that achieves clinically acceptable registration accuracy. Methods The model utilized in the study was not based on a clinical case, and therefore, no ethical review was required. Sample size calculation The sample size for a multifactorial repeated-measures analysis was determined using G*Power software (Version 3.1.9.4; Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany). The study design involved 20 combinations of five registration strategies and four measurement sites, each repeated twice (once on each side of the maxilla). Based on an effect size (f) of 0.25, a significance level (α) of 0.05, and a desired power (1-β) of 0.9, the calculation indicated that a total of 120 samples were required to achieve a power of 0.905. CBCT selection A maxillary edentulous model (Shanghai Jade Digital Dental Laboratory Co., Ltd,) which featured adequate bone volume in the anterior maxilla and bilaterally in the maxillary tuberosity-pterygoid area but limited bone volume in the bilateral premolar and molar regions, was selected for CBCT scanning. The scanning was performed using a Planmeca ProMax machine (Planmeca Oy, Helsinki, Finland) with the following settings: 96 kV, 8 mA, a voxel size of 0.2 mm, and a scanning duration of 12 s. Subsequently, the CBCT images were exported in digital imaging and communications in medicine (DICOM) format for further digital planning. Implant and fiducial markers planning The DICOM data were initially imported into the Nobel Clinician software (Nobel Biocare, Göteborg, Sweden) for virtual implant planning and to pinpoint the locations of fiducial markers. The two anterior implants (φ3.5 × 11.5 mm, Nobel Active, Nobel Biocare) were positioned axially at the mid-crestal entry point of the lateral incisor region. The two posterior implants (φ3.5 × 13 mm, Nobel Active, Nobel Biocare), entering at the mid-crest, were placed at the second premolar level with a 30° apical mesial tilt along the anterior maxillary sinus wall (Fig. 1a) [2]. Fig. 1Schematic of the planned implants (a) All-on-four; (b) TPP; (c) ZI. The figures were created with biorender.com Schematic of the planned implants (a) All-on-four; (b) TPP; (c) ZI. The figures were created with biorender.com The TPPs (φ4.3 × 15 mm, Nobel Active, Nobel Biocare) were placed at a 45° inclination relative to the Frankfort plane. Their entry points were situated 10 mm mesial to the pterygo-maxillary junctions, at the midpoints of the alveolar crests. The trajectories were oriented posteriorly toward the pterygoid processes, with the exit points positioned at the deepest concavities of the cortical bone within the pterygoid fossae (Fig. 1b) [30]. The ZIs (φ4.3 × 50 mm, Nobel Biocare) were positioned at the mid-crestal points of the second premolar regions, with cortical exit points determined based on the thickness and width of the zygomatic bone, following the zygomatic anatomy-guided approach (ZAGA) principles for implant trajectories (Fig. 1c) [31]. The software clearly marked the platform and apex of the virtually planned tilted implants. These markings would later serve as target points. Phantom preparation and fiducial markers insertion Eight carbon steel mini-screws (φ1.7 mm × 10 mm, Jianwei Co., Ltd, Shenzhen, Guangdong, China) were inserted into the maxillary model as target points based on the implant plan described before. Specifically, the apical regions of mesial-tilted implants in the ‘all-on-4’ configuration (Points e and f) and ZIs (Points g and h), along with both coronal (Points b and c) and apical regions (Points a and d) of TPPs, were chosen for TRE evaluation. These locations were selected due to their relatively distant positions from the registration center. Another 8 carbon steel mini-screws (φ1.7 mm × 10 mm, Jianwei Co., Ltd, Shenzhen, Guangdong, China) were inserted directly into the model as fiducial markers, ensuring that these screws would not obstruct the implant trajectories. Two screws were placed bilaterally at the maxillary tuberosity regions (Points 1 and 6), two on the anterior alveolar crest (Points 2 and 5), two on the buccal aspect of the anterior alveolar ridge (Points 3 and 4), and two along the midline palatine suture (Points 7 and 8). All mini-screws were anchored into the model with sufficient depth to ensure stability (Fig. 2). Fig. 2Maxillary edentulous model with the tracking plate (a) Occlusal view; (b) Right side view; (c) Left side view. Numbers 1–8 represent fiducial markers; Letters a-g represent the measurement points for target registration error Maxillary edentulous model with the tracking plate (a) Occlusal view; (b) Right side view; (c) Left side view. Numbers 1–8 represent fiducial markers; Letters a-g represent the measurement points for target registration error Registration The model’s image data were acquired using a CBCT scanner (Planmeca ProMax, Planmeca Oy, Helsinki, Finland) with the following settings: 96 kV, 8 mA, a voxel size of 0.2 mm, and a scanning duration of 12 s. Subsequently, these images were exported to the navigation planning software (iPlan Navigator, Brainlab) for comprehensive three-dimensional reconstruction. All 16 screws used as either fiducial markers or target points were selected as available points for registration in the software. Afterwards, a bony-anchored tracking plate was securely fixed to the model’s midline of the maxillary alveolar ridge. This allowed for the clear visualization and identification of all fiducial markers and target points within the software across the coronal, sagittal, and horizontal planes. During the registration process, a navigation probe with the tracking device was used to sequentially click on the fiducial markers inserted in the model according to different registration strategies. Four or more markers were enough to complete the registration procedure. Five groups of registration strategies were conducted, and each group was replicated 6 times. Configurations of fiducial markers in these groups are illustrated in Fig. 3. Fig. 3Configurations of fiducial markers in different groups (a) Occlusal view of the model; (b) Fiducial markers on anterior alveolar crest, buccal aspect of anterior alveolar ridge, and maxillary tuberosity (Group’123456’); (c) Fiducial markers on buccal aspect of anterior alveolar ridge, maxillary tuberosity and midline palatine suture (Group’13468’); (d) Fiducial markers on the crest and buccal aspect of anterior alveolar ridge (Group’2345’); (e) Fiducial markers on anterior alveolar crest, maxillary tuberosity and midline palatine suture (Group’12568’); (f) Fiducial markers on anterior alveolar crest, buccal aspect of anterior alveolar ridge, and midline palatine suture (Group’23458’) Configurations of fiducial markers in different groups (a) Occlusal view of the model; (b) Fiducial markers on anterior alveolar crest, buccal aspect of anterior alveolar ridge, and maxillary tuberosity (Group’123456’); (c) Fiducial markers on buccal aspect of anterior alveolar ridge, maxillary tuberosity and midline palatine suture (Group’13468’); (d) Fiducial markers on the crest and buccal aspect of anterior alveolar ridge (Group’2345’); (e) Fiducial markers on anterior alveolar crest, maxillary tuberosity and midline palatine suture (Group’12568’); (f) Fiducial markers on anterior alveolar crest, buccal aspect of anterior alveolar ridge, and midline palatine suture (Group’23458’) Target registration error measurement TRE represents the discrepancy between a selected point outside the fiducial marker the virtual coordinate system and its actual corresponding position after the registration process [32]. Following each registration procedure, the navigation probe was utilized to detect the 8 target points (Point a-h). Since the 8 screws designated as target points were excluded from the registration process, whenever the probe was aligned with the screw, the value displayed on the screen reflected the positional discrepancy between the virtual coordinate system and the actual physical location, thereby quantifying the TRE at that specific target point (Fig. 4). Given that 30 registrations were conducted, with 8 points measured per registration, a total of 240 measurements of TRE were recorded. Fig. 4Measurement of TRE (a) Measuring the TRE at the coronal point of pterygoid implant (Point b); (b) Measuring the TRE at the apical point of conventional tilted implant (Point e); (c) Measuring the TRE at the apical point of zygomatic implant (ZI) (Point g); (d) Navigation screen displaying the TRE value at Point b; (e) Navigation screen displaying the TRE value at Point e; (f) Navigation screen displaying the TRE value at Point g Measurement of TRE (a) Measuring the TRE at the coronal point of pterygoid implant (Point b); (b) Measuring the TRE at the apical point of conventional tilted implant (Point e); (c) Measuring the TRE at the apical point of zygomatic implant (ZI) (Point g); (d) Navigation screen displaying the TRE value at Point b; (e) Navigation screen displaying the TRE value at Point e; (f) Navigation screen displaying the TRE value at Point g Statistical analysis All statistical analyses were conducted using IBM SPSS Statistics (Version 27.0, IBM Corp., Armonk, NY). This study utilized 5 registration distribution strategies, each implemented 6 times. TREs were measured at 8 symmetrical left-right sites during each implementation, which were considered repeated measurements. And a total of 240 measurements were obtained. The mean and standard deviation (SD) of TREs for each site under each distribution strategy were presented for descriptive statistical parameters. A mixed effects model was used for data analysis, with TRE as the dependent variable. The model included distribution, site, and measurement occasion as covariates, with distribution, site, measurement occasion, and the interaction between distribution and site set as fixed effects. Additionally, the experiment group number was incorporated as a random effect. If the interaction term was statistically significant, Bonferroni’s method was used to conduct pairwise comparisons between and within groups. A significant difference was defined as P < 0.05. Results Table 1 summarizes the average TREs obtained at four specific points under five distribution strategies. Analysis using a linear mixed-effects model revealed that registration distribution, measurement site, and their interaction had a statistically significant impact on TREs, with P-values of 0.010, 0.000, and 0.001, respectively, as detailed in Table 2. Conversely, no statistically significant correlation was observed between TRE measurements from left-right symmetric sites within the same experimental set (P = 0.182). Supplementary Tables S1 and S2 present the results of inter- and intra-group comparisons. At the apical point of the pterygoid implant (Point a), the Group ‘123456’ exhibited a significantly reduced TRE compared to the Groups ‘2345’ and ‘23458’ (mean difference = 0.400, 0.408; P = 0.039, 0.033, respectively). For the coronal point of the TPP (Point b), the Groups ‘123456’, ‘12568’, and ‘13,468’ significantly decreased TRE compared to the Group ‘23,458’ (mean difference = 0.542, 0.608, 0.600; P < 0.001 for all). No significant differences in TRE were observed for the apical points of conventional and ZIs (Points e and g) among the five registration strategies. Across all five distribution strategy groups, the TREs at Point a were generally higher than those at Points b and e, although some of these differences were not statistically significant. Similarly, the TREs at Point e were consistently lower than those at Point g, with statistical significance only in the Groups ‘2345’ and ‘23458’. Only in Group ‘23,458’ was a significant difference in TREs between Point b and Point e (mean difference = 0.433, P = 0.001). The TREs at Point b differed significantly from those at Point g only in Group ‘12,568’ (mean difference = −0.375, P = 0.002). No significant differences in TREs were observed between Points a and g across all groups. Table 1TREs at four points under five distribution strategies (mean ± SDa)Registration strategyabeg2345Bilateral crests and buccal aspects of the anterior alveolar ridges1.00 ± 0.450.63 ± 0.400.44 ± 0.170.84 ± 0.47123,456Bilateral anterior alveolar crests and buccal aspects, and bilateral maxillary tuberosities0.60 ± 0.190.43 ± 0.160.53 ± 0.310.65 ± 0.2612,568Bilateral anterior alveolar crests and maxillary tuberosities, and midline palatine suture0.71 ± 0.250.37 ± 0.220.58 ± 0.240.74 ± 0.3113,468Bilateral buccal aspects of the anterior alveolar ridges and maxillary tuberosities, and midline palatine suture0.68 ± 0.140.38 ± 0.210.32 ± 0.190.50 ± 0.2723,458Bilateral crests and buccal aspects of the anterior alveolar ridges, and midline palatine suture1.01 ± 0.220.98 ± 0.270.54 ± 0.290.82 ± 0.46aSD: standard deviation Table 1TREs at four points under five distribution strategies (mean ± SDa)Registration strategyabeg2345Bilateral crests and buccal aspects of the anterior alveolar ridges1.00 ± 0.450.63 ± 0.400.44 ± 0.170.84 ± 0.47123,456Bilateral anterior alveolar crests and buccal aspects, and bilateral maxillary tuberosities0.60 ± 0.190.43 ± 0.160.53 ± 0.310.65 ± 0.2612,568Bilateral anterior alveolar crests and maxillary tuberosities, and midline palatine suture0.71 ± 0.250.37 ± 0.220.58 ± 0.240.74 ± 0.3113,468Bilateral buccal aspects of the anterior alveolar ridges and maxillary tuberosities, and midline palatine suture0.68 ± 0.140.38 ± 0.210.32 ± 0.190.50 ± 0.2723,458Bilateral crests and buccal aspects of the anterior alveolar ridges, and midline palatine suture1.01 ± 0.220.98 ± 0.270.54 ± 0.290.82 ± 0.46aSD: standard deviation TREs at four points under five distribution strategies (mean ± SDa) aSD: standard deviation Table 2Type III tests of fixed effectsSourceF P Intercept398.536< 0.001***Distribution4.1480.010*Site19.957< 0.001***Left and right1.7920.182Distribution * Site3.0840.001***P < 0.05; **P < 0.01; ***P < 0.001 Table 2Type III tests of fixed effectsSourceF P Intercept398.536< 0.001***Distribution4.1480.010*Site19.957< 0.001***Left and right1.7920.182Distribution * Site3.0840.001***P < 0.05; **P < 0.01; ***P < 0.001 Type III tests of fixed effects *P < 0.05; **P < 0.01; ***P < 0.001 Discussion It was observed that when Points 1 and 6 were included in the registration strategy, the TREs at Points a and b decreased significantly. This finding is particularly insightful for marker-based registration in dCAIS guided pterygoid implant surgery. Incorporating registration points near the maxillary tuberosity as much as possible in the registration strategy can help keep the registration point closer to the target implant. This approach can reduce the TRE at both the coronal and apical points of the pterygoid implant, thereby enhancing the accuracy of dCAIS-guided implant placement. Based on the results of the present study, we conclude that the ‘123,456’ strategy is the most suitable registration strategy for dCAIS-guided pterygoid implant placement. Clinically, cases requiring pterygoid implants often present with maxillary sinus pneumatization and severe bone deficiency in the posterior maxillary region. Therefore, identifying suitable sites for placing registration points 1 and 6 in the edentulous maxilla is challenging. Additionally, the placement of these registration points must not interfere with the drilling path for the pterygoid implants. Preoperative software planning and a registration template during surgery can be employed to address these issues. This approach allows optimal utilization of the remaining maxillary bone to secure the registration pins while avoiding perforation of the maxillary sinus and interference with implant drilling. However, there was no significant difference in the TRE at Points e and g among the five distribution strategies proposed in this study. Combined with intra-group analysis, it is evident that the TRE at Point e is relatively minimal, likely within 0.6 mm. The apex of the mesial implant in the ‘all-on-four’ is located near the center of the maxillary arch, which also corresponds to the center of the registration points. Therefore, the TRE of the tilted implant in the ‘all-on-4’ concept can achieve a relatively ideal range when the registration points are maintained in a symmetrical and dispersed distribution in the fully edentulous maxilla. Although there was no statistically significant difference in the TRE at Point e between the groups, the results suggest that the ‘13468’ distribution strategy achieves the lowest TRE and is close to significant difference, and this distribution may be the most suitable for dCAIS-guided implant placement in the ‘all-on-4’ procedure. Clinically, when performing ‘all-on-4’ navigation-guided implant surgery, even if the surgical area is limited to the anterior maxilla and premolar regions, adding registration points in the maxillary tuberosity and palatal areas can enhance the accuracy of registration, and thereby further improve the precision of the navigation surgery. Regarding Point g, its larger TRE is due to the extended path of the ZI and the apical point’s location within the zygomatic bone, causing the target point to be distant from the distribution area of the registration points. Several distribution strategies proposed in this study exhibited insignificant differences in the distance between the registration center and the target point of the zygomatic bone, indicating that merely altering the distribution of registration points within the arch is insufficient for improving the TRE of the apical point for ZI. Therefore, future studies can explore methods to position the registration points closer to the zygomatic bone to reduce the TRE of ZI and enhance the accuracy of dCAIS-guided ZI surgery. Based on this experiment, we can tentatively conclude that the ‘13,468’ distribution strategy is most suitable for dCAIS-guided ZI placement. This conclusion is consistent with the findings of a previous study, which claimed that a scattered distribution with a polygon span of five regions (both sides of the anterior nasal spine, bilateral maxillary tuberosity, midline palatine suture) with at least five fiducial markers in the edentulous maxilla for registration seems to achieve an acceptable TRE value for dCAIS-guided ZI surgeries [27]. However, it is important to note that our model experiment is conceptual and cannot fully replicate the complex and variable clinical scenarios. In cases of severe bone deficiency, optimal placement of fiducial markers can be difficult. To prevent injury to critical structures such as the maxillary sinus or mandibular canal, several practical strategies may be employed: using shorter screws, angulating the screw to avoid interference with planned implant trajectories, or utilizing alternative sites such as the palate or zygomatic buttress [29]. Preoperative virtual planning and the fabrication of a custom surgical guide for fiducial marker placement are also recommended in these situations. The inherently poor cortical bone quality of the maxillary tuberosity often limits the ability to achieve rigid and stable screw fixation. Nevertheless, within safe anatomical boundaries, several techniques can be used to improve stability, such as selecting longer and wider screws with a coarser pitch and deeper threads, or applying surgical protocols like under-drilling—or even omitting drilling entirely in certain cases—to enhance stability through bone condensation. The bone fixation screws used in this study are considered the current gold standard for fiducial marker stabilization due to their high rigidity and reliability, despite their invasive nature and potential for additional trauma and patient discomfort. A variety of non-invasive or minimally invasive registration alternatives are available, including mucosal-adhesive markers [33], surgical templates [34], and doubel-scan protocol [35]. The spatial distribution principles for registration points evaluated in this study are also relevant to these point-based methods. Consequently, screws were adopted in this experiment. Additionally, for patients with severe maxillary defects requiring ZI placement, when intraoral registration cannot be accomplished due to the inability to achieve intraoral anchorage of fiducial markers, extraoral markers can be considered [36]. It should also be noted that marker-free registration techniques—such as surface registration—are emerging [37], though in edentulous cases, their performance may be influenced by the lack of recognizable anatomical landmarks. Therefore, the application of surface matching in edentulous cases requires further optimization and validation. In actual clinical practice, physicians should consider the patient’s bone quality and bone volume in conjunction with our recommendations to determine the optimal distribution of fiducial markers. Conclusions Incorporating fiducial markers on the maxillary tuberosity for registration can be beneficial in reducing the TRE at both the coronal and apical points of pterygoid implants during dCAIS. Maintaining a symmetrical and dispersed distribution of registration points for dCAIS-guided ‘all-on-4’ procedure is advisable. Future research can explore methods to position the registration points closer to the zygomatic bone to minimize the TRE of ZIs in dCAIS. Supplementary Information Supplementary Material 1. Supplementary Material 1.