Three-dimensional evaluation of palatal bone thickness for insertion of micro implants in a Portuguese population: a retrospective cone-beam CT study

Sample size

A convenience sample of 50 CBCT scans fulfilled the inclusion criteria of the present investigation. Most analyses recommended a minimum sample size of 30 (power = 80%, alpha = 0.05). The achieved power with 50 scans is 99.9%. This number reflected the retrospective nature of the study and the ethical limitation on unnecessary radiation exposure (Figure 1). The observed effect size (Cohen’s d = 3.77) yielded a statistical power >99.9% (α = 0.05, two-tailed). Moreover, the associated P value (P < 10^–³³) remained highly significant after the Benjamini–Hochberg (BH) correction for false discovery rate (FDR), indicating that the study was adequately powered to detect relevant anatomical differences even under multiple testing controls.

OPEN IN VIEWER

Figure 1.

All assigned points located at 4, 8, 12, 16 and 20 mm posterior to the incisive foramen and 3, 6 and 9 mm lateral to the midpalatine suture (right side, axial view).

Image acquisition and processing

All CBCT scans were obtained using Planmeca Viso G7 (Planmeca, Helsinki, Finland). The following settings were used: 120 kVp, 5 mA, large field of view (20 cm × 17 cm), exposure time of 30 s and slice thickness of 0.45 mm. The 3D images were constructed using Romexis Viewer version 6.0 (Planmeca, Helsinki, Finland).

Standardisation and landmarks identification

The personal details of each patient were inserted into an Excel file. A coded number was assigned to each included CBCT scan to blind the examiner from possible bias concerning patient age and sex. The CBCT analyses were performed by an orthodontist who had undergone training sessions using the software Romexis Viewer version 6.0 (Planmeca, Helsinki, Finland).

Before analysis, each image was aligned to a standard position. The distal margin of the incisive foramen (ІF) and the posterior nasal spine (PNS) were identified as anatomical reference landmarks on the axial view (Figure 2), and a reference line was drawn across the incisive foramen in the midline on the same view. This plane was then tracked in the sagittal (Figure 3) and axial view (Figure 2). The palatal bone thickness was assessed on the coronal plane (Figure 4), where the measurement points had to be defined in the coronal and sagittal planes over a set of evenly sized grids over the areas of interest.

OPEN IN VIEWER

Figure 2.

Axial view of the incisive foramen and midpalatine suture.

OPEN IN VIEWER

Figure 3.

Sagittal view of the incisive foramen and midpalatine suture in relation to the measurement points located at 4, 8, 12, 16 and 20 mm posterior to the incisive foramen.

OPEN IN VIEWER

Figure 4.

All assigned points located at 4, 8, 12, 16 and 20 mm posterior to the incisive foramen and 3, 6 and 9 mm lateral to midpalatine suture (right side, axial view).

In the sagittal plane, measurement points were taken in the anterior region of the palate at 4 and 8 mm and in the posterior region at 12, 16 and 20 mm distally from the IF (Figure 1). These points were designated as S4, S8, S12, S16 and S20, respectively. Vertical lines were drawn perpendicular to the reference line through these points. Furthermore, commencing at point S4 (and continuing to S8, S12, S16 and S20) and transitioning to the coronal view (Figure 4), the measurements were taken exclusively on the right side. The lateral aspect of each patient was measured at 3, 6 and 9 mm from the midline (designated as C3, C6 and C9, respectively) (Figure 4). Prior studies have indicated no significant differences in palatal bone thickness between the right and left sides of the palate (Gracco et al., 2008).

A total of 750 measurements were recorded, with 15 measurements taken for each of the 50 patients. In addition, the points situated 4 and 8 mm posterior to the IF and 3, 6 and 9 mm lateral to the MPS were designated the ‘anterior region’. The points located 12, 16 and 20 mm posterior to the IF and 3, 6 and 9 mm lateral to the MPS were consequently designated as the ‘posterior region’ (Figure 5). Each measurement was subsequently entered into an Excel worksheet for further statistical data analysis.

OPEN IN VIEWER

Figure 5.

All assigned points located at 4, 8, 12, 16 and 20 mm posterior to the incisive foramen and 3, 6 and 9 mm lateral to midpalatine suture (right side, axial view).

Statistical analysis

Statistical Package for the Social Sciences version 28 (IBM Corp., Armonk, NY, USA) was utilised for the statistical analysis. Descriptive statistics included the mean, standard deviation (SD) and 95% confidence interval. To assess significant differences in measurements between age and sex groups, the Student’s t-test was applied in the results tables. The normality of the data distribution was evaluated using the Shapiro–Wilk test, while Levene’s test assessed the homogeneity of variance in bone thickness across different points and groups. When minor deviations from normality were detected, the Student’s t-test results were further verified using a non-parametric equivalent, the Mann–Whitney U-test. To control for the FDR, the BH procedure was implemented. Moreover, Box’s test was employed to evaluate the assumption of equality of covariance matrices across the groups defined by age and sex. The objective of the test was to ascertain whether the observed covariance matrices of the dependent variables are equal across groups, a critical assumption for the validity of analysis of variance (ANOVA). A between-patient ANOVA was conducted to evaluate the effect of age and sex on the average bone thickness. The analysis was conducted using the type III sum of squares method, with a significance level of 0.05.

A mixed-effects regression analysis was employed to evaluate the effects of sex (factor), age (continuous covariate) and point (indicator of the 15 measurements), taking into account the correlation of repeated measurements. Subsequently, a hierarchical cluster analysis was performed to screen for homogeneous points. The homogeneity results from cluster analysis provide a clinically meaningful way to predict suitable areas for MI insertion. Identifying regions with stable and uniform bone thickness can enhance treatment predictability, while recognising variable zones ensures proper individualised planning to avoid complications.

Reproducibility study

The examiner conducted a re-evaluation of 10 CBCT scans, selected at random, which had previously undergone measurement of the same 150 points of interest at 2-week intervals. The intraclass correlation (ICC) for agreement and Bland and Altman’s plot analysis were employed to assess the degree of similarity between the measurements taken at two different times. Two forms of ICC were computed: ICC for single measurements, which reflects the reliability of individual measurements taken at different time points, and ICC for average measurements, which assesses the reliability when multiple measurements are averaged, thereby reducing variability and increasing precision.

Summary

This study was a retrospective, 3D radiographic observational investigation conducted on a cohort of Portuguese patients attending Egas Moniz University Clinics. Similar to dental implants, the success of MІ is heavily influenced by the quantity and quality of the surrounding bone, particularly cortical bone density and thickness. These factors are vital for establishing primary stability, which is a pivotal element for ensuring long-term success (Chang et al., 2021; Vidalon et al., 2021). This study concluded that boys/men had significantly thicker palatal bone than girls/women. The palatal bone was significantly thicker in the adolescent and young adult age group compared to the adult age group. In addition, the bone thickness decreased posteriorly within each sagittal section.

Generalisability

MI use is generally recommended when bone maturation is adequate to ensure retention and stability. This normally occurs around the age of 16 years in girls and 18 years in men (Chen et al., 2007). Markedly, no difference in palatal MI stability and retention was observed among adults, adolescents and even children, indicating that the palate bone density, even at an early age, is sufficient for MI stability along an orthodontic treatment (NienKemper et al., 2015; Wilmes et al., 2016). In our investigation, the sample size consists of patients aged 12–51 years. This age range was determined by several practical considerations, including the limited availability of CBCT scans and the necessity of fulfilling the inclusion criteria, such as full upper arch permanent dentition. This criterion is inherently age-dependent, as the prevalence of a complete dental arch tends to decrease with advanced age. Notably, this age range aligns with some previously undertaken studies, likely due to the necessity of meeting similar inclusion criteria (Chang et al., 2021; Chhatwani et al., 2019; Wilmes et al., 2016). However, it would be reasonable to narrow the age range by adding more groups with greater sample sizes in future studies.

The acquired CBCT images were oriented using landmarks on the palatal bone before taking measurements. This standardisation was focused at proximity to the region of interest (hard palate), a crucial step in ensuring uniformity across CBCT scans. A comparable standardised positioning was also employed in numerous similar studies (Bonangi et al., 2018; Santiago et al., 2020; Suteerapongpun et al.,2018). This meticulous approach is essential for minimising measurement variability and ensuring the reliability of subsequent analyses. On the other hand, Vidalon et al. (2021) used the Frankfurt horizontal plane as a guide for their image orientation. Comparable measurements are reported from the studies using different orientation landmarks, indicating that the accurate standardisation of the images is more important than the referenced landmarks.

The majority of the studies investigated the total palatal bone thickness with the main aim of guiding the insertion of palatal MІ and decreasing the failure rate in clinical practice. Our study further categorised the participants into age and sex groups to similarly investigate the effect of age and sex on palatal bone thickness. Andraws Yalda et al. (2024), Chang et al. (2021), Gracco et al. (2008) and Wang et al. (2017) concluded that no significant sex difference was observed in their participants. On the other hand, Bonangi et al. (2018), Chhatwani et al. (2019) and Yadav et al. (2018) reported a greater palatal bone thickness in boys/men than in girls/women, aligning with our findings. This discrepancy in the outcome might be due to the different age ranges in each group and the ethnicities reported in other studies. In addition, contrary to ours, most previous investigations did not perform a sample size power calculation. Another significant factor is that the vertical dimension was not considered in all the reviewed studies, including ours. Vidalon et al. (2021) noticed that participants with hypodivergent growth patterns had significantly greater palatal bone height and increased cortical thickness compared to those with hypodivergent and normodivergent growth forms. Investigating the impact of growth patterns on palatal bone thickness was not possible in our study because the field of view of the CBCT was confined to the protocol of As Low As Diagnostically Acceptable being Indication oriented and Patient Specific (ALADAІP) (Oenning et al., 2018). Therefore, obtaining CBCTs, including the total maxillomandibular structures, was unjustified.

Given the 15 anatomically defined points where palatal bone thickness was measured in this study, we addressed the issue of multiple comparisons using the BH procedure to control the FDR. This approach was selected over the more conservative Bonferroni correction to preserve statistical power while still maintaining adequate control of false positives (Korthauer et al., 2019; Verhoeven et al., 2005). The BH method is particularly appropriate in the context of anatomical or exploratory studies involving moderate numbers of comparisons, where the goal is to detect meaningful patterns without excessively inflating type II errors (Korthauer et al., 2019; Verhoeven et al., 2005). In our analysis, no clustering of borderline P values or excessive detection of significance was observed after BH adjustment, suggesting that the probability of spurious findings remained low. Furthermore, the measurement sites were not chosen arbitrarily but were anatomically grounded based on established insertion zones in the literature (Andraws Yalda et al., 2024; Bonangi et al., 2018; Chang et al., 2021; Chhatwani et al., 2019; Gracco et al., 2008; Jayakumar et al., 2012; Ludwig et al., 2011; Suteerapongpun et al., 2018; Vidalon et al., 2021; Wang et al., 2017; Yadav et al., 2018). Thus, the statistical inferences derived from these points are both clinically meaningful and methodologically sound. The use of FDR correction permitted us to maintain interpretative clarity while ensuring that our conclusions regarding sex- and age-related variations, and the identification of favourable insertion sites were not the result of random fluctuation.

The results of the present study revealed a significant decrease in palatal bone thickness from the anterior to the posterior regions. The highest mean vertical palatal bone thickness was observed in the anterior region of the hard palate, specifically at the S4C9 point, located 4 mm posterior to ІF and 9 mm laterally to MPS in all investigated groups aligning with all of our cited studies (Andraws Yalda et al., 2024; Bonangi et al., 2018; Chang et al., 2021; Chhatwani et al., 2019; Jayakumar et al., 2012; Ludwig et al., 2011; Negrisoli et al., 2022; Suteerapongpun et al., 2018; Vidalon et al., 2021; Wang et al., 2017; Yadav et al., 2018). Although the smallest mean vertical palatal bone height was located in the distal region of the hard palate, 20 mm posterior from the ІF and 6 mm lateral to the MPS. This result was comparable with those of other studies where the thinnest palatal bone was observed within 20–24 mm posterior to the ІF (Andraws Yalda et al., 2024; Bonangi et al., 2018; Chang et al., 2021; Chhatwani et al., 2019; Gracco et al., 2008; Jayakumar et al., 2012; Ludwig et al., 2011; Suteerapongpun et al., 2018; Vidalon et al., 2021; Wang et al., 2017; Yadav et al., 2018). Therefore, the first null hypothesis was rejected. Insertion of the MІ in the anterior segment of the palate presents significant advantages. The favourable anatomical conditions, including the relative absence of vital anatomical structures, adequate bone volume and sufficient keratinised gingiva, suggest a definitely lower failure rate for MІ insertion compared to inter-root insertion (Іodice et al., 2022).

Similar to Bonangi et al. (2018), Chhatwani et al. (2019), Negrisoli et al. (2022) and Yadav et al. (2018), our investigation revealed that boys/men exhibited significantly greater palatal bone thickness than girls/women, leading to the rejection of the second null hypothesis. This finding has a high clinical relevance for determining the acceptable MI length used in the palate. The sex-specific pattern has been reported for other areas of the jaws, like the alveolar bone, where females seem to have thinner bone compared to boys/men (Firincioglulari et al., 2024). In contrast, some studies revealed that girls/women have an equal or greater bone supply in the posterior region of the palate than boys/men (Chang et al., 2021; Gracco et al., 2008; Sumer et al., 2016). A possible explanation for these discrepancies is the use of disparate measurement techniques and protocols or differences in the composition of the study populations, including potential ethnic variations.

The present study reveals that adolescents and young adults exhibited greater average palatal bone thickness compared to adults, thus rejecting the third null hypothesis. The same results were obtained by Chhatwani et al. (2019) in 180 male and female healthy German orthodontic patients aged 8–40 years. Yadav et al. (2018) reported similar findings among 359 American participants. Santiago et al. (2020) also revealed that the palatal bone thickness has a significant but weak correlation with the age of their examined cohort. On the contrary, other comparable investigations found no age difference in palatal bone thickness (Chang et al., 2021; Gracco et al., 2008; Wang et al., 2017). A potential reason for these outcome discrepancies is the variations in the investigated cohorts and the different population ethnicities.

Interpretation

Palatal bone thickness measurements are instrumental in selecting optimal MІ placement sites and determining appropriate implant length. This information is critical for achieving sufficient implant stability and preventing iatrogenic injury to vital anatomic structures. It was suggested that a minimum of 5 mm of bony support is crucial for resisting rotational forces and dynamic loads, thereby contributing to MІ stability, and preventing iatrogenic injury to vital anatomical structures (Winsauer et al., 2014). Ludwig et al. (2011) reported that the ideal area for MІ insertion is 3–9 mm posterior to the ІF and within 6 mm mediolaterally including the MPS.

The results of the cluster analysis indicated two groups of sites with a high degree of homogeneity. Integrating the CBCT-driven cluster analysis results into clinical workflows enables more precise, patient-specific treatment planning, reducing risks and improving long-term outcomes in implantology, orthodontics and maxillofacial surgery. Accordingly, the favourable MI insertion sites in our cohort are located 4–8 mm posterior to the ІF and within 9 mm mediolaterally for both male and female groups. Regarding age, the preferred insertion site is located 4–8 mm posterior to the ІF and 3–9 mm lateral to the MPS for adults and 4–12 mm posterior to the ІF and 3–9 mm lateral to the MPS for adolescents and young adults. The results of the present study also confirm previous findings that the T-zone, the area just posterior to the third palatal rugae in the anterior region of the palate, is one of the most favourable sites in the hard palate for MІ insertion (Hourfar et al., 2015; Ludwig et al., 2011; Wilmes et al., 2016). These findings enhance our knowledge regarding palatal bone and provide clinical guidelines that could reduce the necessity for radiographic imaging before the insertion of the orthodontic MІ. However, further research, based on a multicentre cohort of participants, would improve generalisability.

Despite the reported findings, careful diagnostic evaluation remains essential for each patient. Assessment of palatal bone thickness is critical in orthodontic practice due to its implications for MІ placement. Thicker bone regions offer enhanced MІ stability, while thinner areas increase the risk of MІ failure and iatrogenic injury. Individual variations in bone thickness, influenced by sex and age, necessitate personalised treatment plans. Accurate assessment guides optimal MІ placement site and length, maximising success rates and minimising complications. This clinical relevance underscores the need for thorough bone thickness evaluation before MІ insertion and further research in this area to refine treatment protocols and improve patient outcomes.

The ongoing advancements in artificial intelligence (AІ) and the emerging field of computer graphics are poised to revolutionise landmark evaluation. The integration of these technologies is expected to facilitate self-automated assessment processes that are both efficient and reliable (Radwan et al., 2023). These AІ-powered systems can accurately identify and segment palatal bone, enabling precise measurements of bone thickness at various locations. This automated approach not only streamlines the assessment process but also enhances its accuracy and reliability, ultimately leading to better-informed treatment decisions and improved patient outcomes.

Limitations

Despite the relevant outcome of the present investigation, several limitations should be addressed in future similar research. First, the retrospective design limited control over the imaging procedures, which might have introduced bias by excluding patients with incomplete clinical data or distorted CBCT scans. Second, although the achieved power of the sample size (n = 50) was notably high, suggesting a robust statistical foundation, this cohort was not matched by sex and age. In addition, the cohort represents individuals attending a single university clinic. Future research should employ a prospective multicentre design that includes participants matched for sex and age from various dental clinics across Portugal, enhancing the generalisability of the findings. Therefore, this study should be considered as a foundational step for future investigations. Furthermore, despite the rigorous training and the high accuracy of the intra-examiner analysis, an inter-examiner evaluation conducted by an experienced researcher in the field is recommended in future studies.