Radiomic texture patterns of palatal donor tissue on CBCT images: an exploratory study

Introduction

Gingival recession, defined as the apical displacement of the gingival margin relative to the cemento-enamel junction, is a highly prevalent condition frequently associated with dentin hypersensitivity, esthetic concerns, and challenges in plaque control (1,2). Root coverage procedures using subepithelial connective tissue grafts harvested from the palatal mucosa are considered the gold standard for the treatment of these defects due to their predictable clinical outcomes (3,4). However, variability in the biological characteristics of the donor tissue may influence healing and clinical success, and objective methods for preoperative assessment of this tissue remain limited.

Cone-beam computed tomography (CBCT) has become an essential imaging modality in dentomaxillofacial radiology, particularly for the evaluation of mineralized structures, including periodontal bone loss and furcation involvement (5,6). Although CBCT is not primarily intended for soft tissue analysis, it enables visualization of soft tissue contours under standardized acquisition conditions, which has led to growing interest in extracting quantitative imaging information beyond conventional visual assessment (5). Alternative imaging modalities, especially ultrasound, have shown excellent performance in the evaluation of oral soft tissues and may allow for more accurate measurements of gingival and palatal thickness (7,8). However, ultrasound examinations are highly operator dependent and are not routinely incorporated into periodontal imaging workflows in many clinical settings. On the other hand, CBCT examinations are frequently available for treatment planning and may provide an opportunity to extract further quantitative information by radiomic analysis without the need for further imaging procedures.

Radiomics is an emerging field of medical image analysis that converts standard imaging data into quantitative features describing image intensity, texture, shape, and spatial relationships. By extracting information that may not be perceptible during routine visual interpretation, radiomics has been increasingly investigated as a tool for improving diagnosis, disease characterization, prognosis prediction, and treatment response assessment across multiple medical and dental applications (9-11). Texture analysis is one of the most widely used radiomic approaches and consists of a computer-based image processing method that evaluates the distribution and arrangement of gray shades within an image. By quantifying image patterns that may not be visible to the human eye, this approach can provide additional information about tissue characteristics and has emerged as a promising tool in medical and dental imaging (12,13). In dentomaxillofacial imaging, CBCT-based texture analysis has been applied to detect periodontal alterations (6), evaluate implant stability (14), and assess periapical healing (15). However, the interpretation of CBCT-derived texture features remains challenging, as these metrics may be influenced by acquisition parameters, image noise, and reconstruction processes rather than solely reflecting underlying tissue characteristics (9).

To date, no study has explored whether CBCT-derived texture features of palatal donor tissue are associated with clinical outcomes of root coverage therapy. Therefore, the aim of this study was to investigate the relationship between CBCT-based texture parameters and clinical outcomes following root coverage procedures. We present this article in accordance with the STROBE reporting checklist (available at https://fomm.amegroups.com/article/view/10.21037/fomm-2026-0014/rc).

Methods

Study design and patient population

This prospective cohort study was conducted at the Department of Diagnosis and Surgery, Institute of Science and Technology, São Paulo State University (UNESP), São José dos Campos, Brazil. Patients were recruited between January 2025 and June 2025 using a convenience sampling approach. Systemically healthy adult patients presenting recession type 1 (RT1) gingival recession defects in incisors, canines, or premolars and indicated for root coverage surgery using connective tissue grafts were included (1,3,16). All recession defects were classified as RT1 with a clinically identifiable cemento-enamel junction at the treated sites. Teeth presenting non-carious cervical lesions were not included in the study.

Inclusion criteria comprised: age ≥18 years; periodontal health (absence of periodontitis); good oral hygiene (plaque and gingival index <25%); absence of palatal pathology; and indication for mucogingival surgical treatment (17). Exclusion criteria were systemic diseases known to affect wound healing (uncontrolled diabetes mellitus, immunosuppressive disorders or current immunosuppressive therapy), smoking, pregnancy, use of drugs known to interfere with soft tissue repair, previous periodontal surgery in the donor site, and presence of image-degrading artifacts in the region of interest (ROI) (metallic restorations, orthodontic appliances, or motion artifacts) that could affect the CBCT texture analysis.

The a priori sample size calculation was based on the primary exploratory texture outcome, defined as the difference in the Sum average parameter between patients with and without postoperative improvement in dentin hypersensitivity. Sum average was selected because it was one of the gray-level co-occurrence matrix (GLCM) parameters previously used in CBCT-based texture analysis studies and because preliminary internal pilot observations suggested that this parameter could capture relevant differences in tissue gray-level distribution (18,19). A large effect size was assumed (Cohen’s d=1.5), consistent with the exploratory nature of the study and with the magnitude of effects reported in previous CBCT texture analysis investigations involving small clinical samples. Using R software (version 4.4.2), with a two-sided α of 0.05 and 80% power, the minimum estimated sample size was 8 participants per group.

Clinical evaluation

The clinical variables investigated were probing depth, dentin hypersensitivity, and esthetic perception. Clinical measurements were performed by a previously calibrated examiner. Clinical outcomes were assessed at the recipient sites undergoing root coverage therapy at baseline and three months postoperatively. Root coverage surgery was performed using a coronally advanced flap associated with a subepithelial connective tissue graft, following established periodontal plastic surgery principles for the treatment of gingival recession defects (3,16). The graft was harvested from the palatal donor region guided by the customized stent, positioned at the recipient site, and stabilized with sutures according to the surgical protocol.

Probing depth was recorded using a standardized periodontal probe and was included as a clinical indicator of periodontal healing and treatment response following surgery. Although root coverage procedures are traditionally evaluated using recession reduction and root coverage outcomes, probing depth was investigated as an additional indicator of postoperative periodontal stability and tissue integration at the recipient site. Patient-centered outcomes were assessed using Visual Analog Scale (VAS). Dentin hypersensitivity was evaluated using a standardized air stimulus and recorded on a 0–10 VAS scale, whereas esthetic perception was recorded using a 0–10 VAS scale. Esthetic perception was assessed by the patients themselves using a 0–10 VAS scale, where higher scores indicated greater satisfaction with the esthetic appearance of the treated site (20).

The calibration process consisted of repeated clinical measurements performed before the beginning of data collection to ensure consistency in probing depth assessment. The evaluated outcomes were selected to represent both clinical and patient-centered measures of treatment response, allowing exploration of potential associations between preoperative donor tissue texture characteristics and postoperative clinical performance.

CBCT acquisition and image standardization

All CBCT scans were acquired preoperatively using an i-CAT Next Generation unit (Imaging Science International, Hatfield, PA, USA) with a limited field of view (FOV) of 16 cm × 8 cm covering the palatal donor region. The acquisition parameters were as follows: voxel size of 0.20 mm (millimeter), 120 kVp (Kilovoltage Peak), 37.07 mAs (milliampere-seconds), and exposure time of 26.9 seconds. All examinations were acquired using the same CBCT unit, identical acquisition parameters, and the same image reconstruction protocol to minimize technical variability and improve the consistency of texture feature extraction across the study population. CBCT was used as an adjunct imaging tool in periodontal assessment and surgical planning when clinically justified (5,6). To ensure standardized and reproducible localization of the palatal donor site, a customized acetate stent was fabricated for each patient using a conventional impression and vacuum-forming technique. A rectangular window corresponding to the planned graft dimensions was created in the palatal region of the stent prior to CBCT acquisition. During image acquisition, the stent was positioned intraorally, allowing precise tomographic identification of the donor area. The stent therefore functioned both as a surgical guide for graft harvesting and as a tomographic reference to standardize slice selection and ROI delineation. The use of a customized stent, together with standardized acquisition and reconstruction parameters, was intended to reduce methodological variability known to influence CBCT-derived radiomic features and to improve the comparability of texture measurements among patients.

Image processing and ROI selection

All CBCT datasets were exported in Digital Imaging and Communications in Medicine (DICOM) format and anonymized prior to analysis. Image processing and slice selection were performed by one dentomaxillofacial radiologist with more than 5 years of experience in CBCT interpretation. In the multiplanar reconstruction (MPR) module of OnDemand 3D software (CyberMed, Seoul, South Korea), reference axes were aligned according to the long axis of the customized stent window, and three standardized coronal slices corresponding to the anterior, middle, and posterior portions of the palatal donor region were selected, following methodology adopted in previous CBCT texture analysis studies (14,18).

Texture analysis was subsequently performed by a second calibrated examiner, blinded to the clinical outcomes. The selected slices were exported in bitmap format and analyzed using MaZda software (version 3.20, Technical University of Lodz, Poland) (21). Standardized quadrangular ROI measuring 4 mm × 4 mm was manually delineated, encompassing the palatal mucosal tissue located directly beneath the stent window (Figure 1). The ROI size was selected to ensure standardized sampling of the palatal donor tissue while minimizing the inclusion of adjacent anatomical structures outside the planned graft area. This dimension was considered sufficient to provide a sufficient number of pixels for texture extraction while maintaining the correspondence with the dimensions of the stent window and the intended graft region. In dentomaxillofacial radiology, similar approaches for standardizing ROIs have been used in previous CBCT-based texture analysis studies to improve reproducibility and reduce sampling variability (6,14,15).

Figure 1 Example of a coronal CBCT slice analyzed in MaZda software. A customized acetate stent with a rectangular window is schematically indicated to delimit the palatal donor site, and the ROI corresponding to the palatal mucosal tissue beneath the stent window was manually outlined using the polygon tool for subsequent texture analysis. CBCT, cone-beam computed tomography; ROI, region of interest.

For each patient, texture parameters were extracted from the three slices, and the mean value was calculated to represent the donor tissue, as previously described (15,18). Intra-observer reproducibility was assessed by repeating the entire ROI delineation and texture extraction process after a 30-day interval using the same image dataset. Intraclass correlation coefficients (ICC) were calculated to evaluate measurement consistency, yielding an ICC value of 0.963, which indicates excellent reproducibility.

Texture analysis

Texture analysis was performed using GLCM features, as originally proposed by Haralick, to quantify spatial gray-level relationships not reliably perceived by visual inspection (9,12,13). Eleven second-order GLCM parameters were extracted [angular second moment (AngScMom), contrast (Contrast), correlation (Correlat), sum of squares (SumOfSqs), inverse difference moment (InvDfMom), entropy (Entropy), sum average (SumAverg), sum variance (SumVarnc), sum entropy (SumEntrp), difference variance (DifVarnc), and difference entropy (DifEntrp)].

Texture extraction was performed in MaZda software using the default normalization settings. Gray-level discretization was fixed at 6 bits per pixel (64 gray levels), according to the software configuration. All images were processed using identical texture extraction settings to ensure methodological consistency across the dataset.

GLCM features were calculated using predefined pixel offsets [S(x,y)], which represent the spatial relationship between pairs of pixels separated by a specific distance and direction. The notation S(x,y) indicates the horizontal (x) and vertical (y) displacement between pixel pairs used to construct the co-occurrence matrix. For example, S(0,3) represents a vertical displacement of three pixels, S(0,2) a vertical displacement of two pixels, and S(2,−2) a diagonal displacement of two pixels in opposite directions.

Texture parameters were computed across multiple spatial directions (0°, 45°, 90°, and 135°) and predefined pixel distances, consistent with prior CBCT texture analysis studies in dentomaxillofacial imaging (6,9,14,15,18).

Surgical procedure

Root coverage surgery was performed using the Coronally Advanced Flap (CAF) technique associated with a subepithelial Connective Tissue Graft (CTG), following established periodontal plastic surgery principles for the treatment of gingival recession defects (3,16). The connective tissue graft was harvested from the palatal donor region guided by the customized stent, transferred to the recipient site, and stabilized with sutures. Sutures were removed according to standard postoperative protocols.

Statistical analysis

Statistical analyses were performed using R software (version 4.4.2). Continuous variables were summarized as mean and standard deviation or median and interquartile range, according to data distribution. Categorical variables were presented as absolute frequencies and percentages. Because of the small sample size and the non-normal distribution of several variables, nonparametric statistical methods were adopted. Clinical variables were compared between baseline and postoperative periods using the Wilcoxon signed-rank test. Texture parameters were compared between subgroups stratified according to improvement in dentin hypersensitivity and esthetic perception using the Mann-Whitney U test. Associations between texture parameters and probing depth reduction were assessed using Spearman’s rank correlation coefficient.

Given the exploratory nature of the texture analysis and the large number of texture-related comparisons, P values obtained from the texture analyses were additionally adjusted using the Benjamini-Hochberg false discovery rate (FDR) procedure. Corrections were applied separately within each analytical family (probing depth reduction, dentin hypersensitivity improvement, and esthetic improvement). Unadjusted P values are reported to facilitate comparison with previous exploratory radiomics studies, whereas FDR-adjusted results were considered for interpretation. Statistical significance was set at 5%.

Ethical considerations

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of the São Paulo State University (UNESP) (protocol No. 70969723.1.0000.0077), and written informed consent was obtained from all participants before enrollment.

Results

The study consisted of a convenience cohort of patients recruited according to predefined eligibility criteria and that all pre-selected eligible patients were included in the final analysis, with no exclusions or losses after eligibility confirmation. Finally, a total of eleven patients were included in the study. The study population had a mean age of 41.1±13.3 years (range, 24–60 years) and was predominantly female (81.8%). All participants were non-smokers. Baseline demographic and clinical characteristics of the study population are presented in Table 1.

Clinical outcomes

The clinical variables investigated were probing depth, dentin hypersensitivity, and esthetic perception, evaluated at baseline and three months postoperatively. Significant clinical improvement was observed three months after surgery. Esthetic perception scores increased from a median of 5.0 at baseline to 10.0 postoperatively (P=0.02). Probing depth was significantly reduced from a median of 3.0 to 1.0 mm (P=0.006). Dentin hypersensitivity demonstrated a marked reduction, with median values decreasing from 7.0 at baseline to 0.0 postoperatively; however, statistical comparison was not feasible due to absence of variability in postoperative scores.

These findings are summarized in Table 2.

Correlation between probing depth reduction and texture parameters

Spearman correlation analysis identified several moderate-to-strong correlations between probing depth reduction and texture parameters. The strongest positive correlations were observed for AngScMom in direction S(0,3) (ρ=0.66) and Correlat in direction S(0,2) (ρ=0.62). Negative correlations were observed for Entropy in direction S(0,3) (ρ=−0.66), DifEntrp in direction S(2,−2) (ρ=−0.64), and DifEntrp in direction S(0,3) (ρ=−0.65). However, after Benjamini-Hochberg FDR correction, none of these associations remained statistically significant and should therefore be interpreted as exploratory findings. Figure 2 illustrates the distribution of the evaluated texture parameters according to dentin hypersensitivity outcome, together with the corresponding group means and standard deviations.

Figure 2 Distribution of texture parameters according to dentin hypersensitivity outcome. Dots represent individual values available in the dataset. Black circles indicate group means, and error bars represent ±1 SD. The evaluated parameters include angular second moment (AngScMom), contrast (Contrast), correlation (Correlat), sum of squares (SumOfSqs), inverse difference moment (InvDfMom), sum average (SumAverg), sum variance (SumVarnc), sum entropy (SumEntrp), entropy (Entropy), difference variance (DifVarnc), and difference entropy (DifEntrp). For SumOfSqs and SumAverg, only one value was available per group; therefore, SDs could not be estimated. SD, standard deviation.

Texture parameters and hypersensitivity improvement

Eight patients demonstrated improvement in dentin hypersensitivity, whereas three did not. Most texture parameters did not show statistically significant differences between groups (P>0.05). In the unadjusted analysis, the parameter SumAverg showed higher mean values in the group with hypersensitivity improvement (18.0±1.89) compared with the group without improvement (14.5±3.47; P=0.04). However, this association did not remain statistically significant after Benjamini-Hochberg FDR correction and should therefore be interpreted as exploratory. Figure 3 illustrates the strength and direction of the Spearman correlations between probing depth reduction and the evaluated texture parameters across different spatial offsets.

Figure 3 Heatmap illustrating the Spearman correlation coefficients between probing depth reduction and CBCT-derived texture parameters according to different spatial offsets [S(x,y)]. Color intensity reflects the magnitude and direction of the correlation coefficients, with blue indicating positive correlations and red indicating negative correlations. Darker shades represent stronger correlations. Texture parameters include angular second moment (AngScMom), contrast (Contrast), correlation (Correlat), inverse difference moment (InvDfMom), sum variance (SumVarnc), sum entropy (SumEntrp), entropy (Entropy), difference variance (DifVarnc), and difference entropy (DifEntrp). Asterisks (*) indicate statistically significant correlations at the 5% level based on unadjusted P values. CBCT, cone-beam computed tomography.

Texture parameters and esthetic improvement

Eight patients showed esthetic improvement, while three did not. In the unadjusted analysis, two texture parameters demonstrated differences between groups. The parameter S(1,−1)InvDfMom showed higher mean values in the group without esthetic improvement (0.49±0.09) compared with the group with improvement (0.32±0.08; P=0.03). Conversely, S(1,0)DifVarnc showed higher mean values in the group with esthetic improvement (5.90±3.38) compared with the group without improvement (1.57±0.63; P=0.03). However, neither association remained statistically significant after Benjamini-Hochberg FDR correction. Therefore, these findings should be interpreted as exploratory and hypothesis-generating.

Effect size analysis

Observed effect sizes varied considerably among the evaluated texture parameters. For the hypersensitivity outcome, SumAverg demonstrated a large effect size (Cohen’s d=1.523), whereas DifEntrp showed a small effect size (d=0.177).

For the esthetic outcome, SumAverg (d=1.001) and DifVarnc (d=0.981) also demonstrated large effect sizes. Based on the observed effect sizes, future confirmatory studies would require approximately 17 and 18 participants per group, respectively, to achieve 80% statistical power at a two-sided significance level of 5%.

Discussion

The present study investigated the potential of texture analysis applied to CBCT images as a predictive tool for palatal donor tissue quality and its association with the clinical success of root coverage therapy. The findings demonstrated significant postoperative clinical improvement in esthetic perception and probing depth, together with a marked reduction in dentin hypersensitivity. In the unadjusted analyses, selected texture parameters showed trends toward association with clinical outcomes. However, none of these associations remained statistically significant after Benjamini-Hochberg FDR correction and should therefore be interpreted as preliminary observations.

The significant improvement observed in esthetic perception is consistent with the literature, which recognizes esthetics as one of the main indications for root coverage procedures (3,16). Patient-centered outcomes, such as esthetic satisfaction and reduction of hypersensitivity, are critical determinants of treatment success in mucogingival surgery. The reduction in probing depth further indicates restoration of favorable periodontal conditions and adequate integration of the connective tissue graft, in agreement with previous reports on periodontal plastic surgery outcomes (22,23).

Regarding hypersensitivity, although statistical testing was limited by the absence of postoperative variability, the marked clinical reduction supports the established role of root coverage in managing exposed root surfaces (20). The present study suggests that imaging-derived quantitative descriptors of the palatal donor region may be associated with this clinical response.

The association of donor tissue texture characteristics with the improvement of dentin hypersensitivity may also be mediated by intermediate clinical factors such as the degree of root coverage achieved. Since recession depth reduction and root coverage percentage were not evaluated in the present study, this potential mechanism should be addressed in future research.

In the unadjusted analysis, patients who demonstrated improvement in dentin hypersensitivity tended to present higher values of the sum average parameter. According to Haralick’s definition, this parameter reflects the distribution of the sum of gray levels within the GLCM, representing overall intensity distribution and spatial organization (12,13). Although this trend did not remain statistically significant after correction for multiple comparisons, it may indicate differences in gray-level distribution patterns within the donor tissue that deserve further investigation in larger cohorts.

These findings align with previous dentomaxillofacial radiology studies demonstrating that texture analysis can detect subtle differences not perceptible to the human eye. Gonçalves et al. (6) showed that CBCT-based texture analysis assists in detecting furcation involvement. Costa et al. (14) demonstrated correlations between CBCT texture parameters and implant stability, suggesting that radiomic descriptors may reflect underlying structural conditions. More recently, Costa et al. (18) reported that texture analysis of CBCT images can differentiate odontogenic and non-odontogenic sinusitis, even in soft tissue-related conditions. Lopes et al. (15) further reinforced the concept of radiomics as a quantitative imaging approach for assessing periapical bone healing. Together, these studies support the potential of CBCT-derived texture features as imaging-based descriptors in oral tissues, while also highlighting the importance of validation studies before clinical implementation.

Although CBCT is traditionally employed for mineralized tissue assessment (24,25), the application of texture analysis extends its use by extracting quantitative information from grayscale distribution patterns. As previously emphasized in radiomics literature, texture analysis does not evaluate isolated voxel intensities, but rather spatial relationships between neighboring voxels, enabling the identification of image heterogeneity patterns (9,13). This methodological characteristic may explain the observed associations with clinical outcomes.

The trends observed for inverse difference moment and difference variance further suggest that image homogeneity and gray-level dispersion patterns may be related to clinical healing and esthetic outcomes. Patients with esthetic improvement tended to present lower inverse difference moment values and higher difference variance values than those without improvement. Since inverse difference moment reflects image homogeneity and difference variance reflects dispersion in gray-level distribution (12), these findings may indicate subtle differences in image organization within the donor tissue. However, because these trends did not remain statistically significant after FDR correction, their biological interpretation should be considered exploratory and hypothesis-generating rather than confirmatory.

From a clinical perspective, the possibility of obtaining quantitative information from preoperative imaging represents a meaningful advance. Current decision-making regarding donor site selection relies primarily on clinical examination and surgeon experience. The incorporation of quantitative imaging descriptors could contribute to a more objective and reproducible assessment, potentially improving treatment planning.

Recent evidence has highlighted that radiomic features derived from CBCT are highly sensitive to variations in acquisition parameters, reconstruction algorithms, and scanner characteristics, which may introduce variability unrelated to underlying tissue properties. Hatamikia et al. (26) demonstrated that differences in slice thickness, exposure, and reconstruction filters can significantly influence radiomic feature values, emphasizing the need for standardization and cautious interpretation. Similarly, previous studies have shown that radiomic features can be reproducibly extracted from CBCT images and may provide additional quantitative information beyond conventional image assessment, particularly in longitudinal and treatment response analyses (27).

The importance of radiomic feature stability has been emphasized in recent CBCT radiomics studies. Fave et al. (27) demonstrated that several texture features are highly sensitive to imaging protocol, scatter conditions, and motion, highlighting the need for standardized acquisition and processing workflows. More recently, Willam et al. (28) reported high feature stability in CBCT radiomics when image acquisition and reconstruction parameters were kept constant, reinforcing the importance of methodological standardization. In the present study, all examinations were acquired using the same CBCT unit and identical acquisition parameters, and texture extraction was performed using uniform software settings for all patients. Nevertheless, dedicated phantom-based robustness analyses were not performed and should be incorporated into future validation studies.

Despite this, these features are influenced by imaging conditions and technical factors and therefore should be interpreted as quantitative imaging descriptors rather than direct surrogates of tissue microstructure. In this context, CBCT radiomics may be better understood as an exploratory approach for identifying image-based patterns associated with clinical outcomes. To overcome these limitations, dedicated radiomic validation strategies should be included in future studies, such as phantom-based test-retest experiments, evaluation of feature stability to different acquisition and reconstruction settings, standardized gray-level discretization and normalization protocols, and external validation in independent multicenter cohorts. Such approaches would allow to identify robust and reproducible features and help to determine whether the preliminary associations observed in the present study are generalizable or specific to the imaging protocol used.

It is important to recognize that CBCT is not optimized specifically for soft tissue imaging. Recent studies have demonstrated that ultrasound has excellent accuracy in evaluating periodontal soft tissues, such as gingival and palatal thickness, avoids ionizing radiation, and can assess superficial structures with high resolution (7,8). This is why ultrasound might be more appropriate when the primary objective is direct characterization of soft tissues. However, the rationale of the present study was not to propose CBCT as a replacement for ultrasound, but to investigate if quantitative texture descriptors can be extracted from CBCT examinations that have already been obtained within the clinical workflow. In this context, texture analysis was investigated as an additional imaging modality that can yield further quantitative information from already existing datasets, without the need of additional examinations.

Nonetheless, the study has limitations. The relatively small sample size limits statistical power and may explain the absence of significant findings for most parameters. In addition, the 3-month follow-up period reflects early clinical healing and does not allow assessment of long-term graft maturation and stability. Because connective tissue remodeling and soft tissue integration may continue for several months after surgery, clinical outcomes evaluated after 6 months or longer could differ from those observed in the present study. Therefore, future investigations with extended follow-up periods are necessary to determine whether the observed radiomic patterns remain associated with long-term treatment outcomes.

Another limitation is that clinical outcomes directly related to modification of the soft tissue phenotype, such as gingival thickness and width of keratinized tissue, were not evaluated. As these parameters are more directly related to the biological characteristics of the donor tissue, future studies should investigate their relationship with texture features extracted from CBCT to better understand the clinical relevance of radiomic findings in periodontal plastic surgery.

Although intra-observer reproducibility was assessed, the use of a single evaluator may introduce potential bias. Additionally, dentin hypersensitivity and esthetic perception were assessed using patient-reported visual analog scales. Because patients were aware of the surgical treatment received, expectation bias cannot be completely excluded. Furthermore, the inherent limitations of CBCT imaging, including variability in gray-level values and susceptibility to noise and acquisition parameters, must be considered when interpreting the radiomic findings.

Ultimately, the study was conducted at a single center, and external validation is therefore required before clinical application.

Future research should include larger, multicenter cohorts with longer follow-up periods and standardized imaging protocols. The integration of radiomic features with clinical and biological data may further improve understanding of their potential role in treatment planning.

Taken together, the present findings support the feasibility of the quantitative imaging approach of texture analysis using CBCT for evaluation of palatal donor regions. Although some texture parameters showed trends toward association with clinical outcomes in the unadjusted analyses, none remained statistically significant after correction for multiple comparisons. However, the observed patterns suggest that radiomic features may provide additive quantitative information of properties of donor tissue and clinical response. These preliminary observations warrant further investigation in larger, adequately powered studies with external validation.

Conclusions

CBCT-based texture analysis showed trends toward association with clinical outcomes following root coverage therapy. Although none of the evaluated texture parameters remained statistically significant after correction for multiple comparisons, the observed patterns suggest that radiomic analysis may provide complementary quantitative information for the assessment of palatal donor tissue. These preliminary findings support further investigation of CBCT-derived texture features in larger and adequately powered studies before clinical application can be considered.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://fomm.amegroups.com/article/view/10.21037/fomm-2026-0014/rc

Data Sharing Statement: Available at https://fomm.amegroups.com/article/view/10.21037/fomm-2026-0014/dss

Peer Review File: Available at https://fomm.amegroups.com/article/view/10.21037/fomm-2026-0014/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://fomm.amegroups.com/article/view/10.21037/fomm-2026-0014/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of the São Paulo State University (UNESP) (protocol No. 70969723.1.0000.0077), and written informed consent was obtained from all participants before enrollment.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Cortellini P, Bissada NF. Mucogingival conditions in the natural dentition: Narrative review, case definitions, and diagnostic considerations. J Periodontol 2018;89 Suppl 1:S204-13. [Crossref] [PubMed]
2. Yadav VS, Gumber B, Makker K, et al. Global prevalence of gingival recession: A systematic review and meta-analysis. Oral Dis 2023;29:2993-3002. [Crossref] [PubMed]
3. Zucchelli G, Mounssif I. Periodontal plastic surgery. Periodontol 2000 2015;68:333-68. [Crossref] [PubMed]
4. Zucchelli G, De Sanctis M. Treatment of multiple recession-type defects in patients with esthetic demands. J Periodontol 2000;71:1506-14. [Crossref] [PubMed]
5. Acar B, Kamburoğlu K. Use of cone beam computed tomography in periodontology. World J Radiol 2014;6:139-47. [Crossref] [PubMed]
6. Gonçalves BC, de Araújo EC, Nussi AD, et al. Texture analysis of cone-beam computed tomography images assists the detection of furcal lesion. J Periodontol 2020;91:1159-66. [Crossref] [PubMed]
7. Qi B, Khazeinezhad R, Hariri A, et al. Three-dimensional mapping of the greater palatine artery location and physiology. Dentomaxillofac Radiol 2023;52:20230066. [Crossref] [PubMed]
8. Mancini L, Tavelli L, Barootchi S, et al. Echo-guided soft tissue harvesting: A novel approach identifying tissue thickness, density, vascularisation and a safe harvesting zone in the palatal region. Int J Oral Implantol (Berl) 2025;18:169-79. [PubMed]
9. Barioni ED, Lopes SLPC, Silvestre PR, et al. Texture Analysis in Volumetric Imaging for Dentomaxillofacial Radiology: Transforming Diagnostic Approaches and Future Directions. J Imaging 2024;10:263. [Crossref] [PubMed]
10. Flaiban E, Orhan K, Gonçalves BC, et al. Radiomics in Action: Multimodal Synergies for Imaging Biomarkers. Bioengineering (Basel) 2025;12:1139. [Crossref] [PubMed]
11. Mariotti F, Agostini A, Borgheresi A, et al. Insights into radiomics: a comprehensive review for beginners. Clin Transl Oncol 2025;27:4091-102. [Crossref] [PubMed]
12. Haralick RM, Shanmugam K, Dinstein I. Textural Features for Image Classification. IEEE Trans Syst Man Cybern 1973;SMC-3:610-21.
13. Lubner MG, Smith AD, Sandrasegaran K, et al. CT Texture Analysis: Definitions, Applications, Biologic Correlates, and Challenges. Radiographics 2017;37:1483-503. [Crossref] [PubMed]
14. Costa ALF, de Souza Carreira B, Fardim KAC, et al. Texture analysis of cone beam computed tomography images reveals dental implant stability. Int J Oral Maxillofac Surg 2021;50:1609-16. [Crossref] [PubMed]
15. Lopes DLG, Lopes SLPC, Ungaro DMT, et al. Radiomics-Driven CBCT Texture Analysis as a Novel Biosensor for Quantifying Periapical Bone Healing: A Comparative Study of Intracanal Medications. Biosensors (Basel) 2025;15:98. [Crossref] [PubMed]
16. Zucchelli G, Tavelli L, McGuire MK, et al. Autogenous soft tissue grafting for periodontal and peri-implant plastic surgical reconstruction. J Periodontol 2020;91:9-16. [Crossref] [PubMed]
17. Papapanou PN, Sanz M, Buduneli N, et al. Periodontitis: Consensus report of workgroup 2 of the 2017 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions. J Clin Periodontol 2018;45 Suppl 20:S162-70.
18. Costa ALF, Fardim KAC, Ribeiro IT, et al. Cone-beam computed tomography texture analysis can help differentiate odontogenic and non-odontogenic maxillary sinusitis. Imaging Sci Dent 2023;53:43-51. [Crossref] [PubMed]
19. Nussi AD, de Castro Lopes SLP, De Rosa CS, et al. In vivo study of cone beam computed tomography texture analysis of mandibular condyle and its correlation with gender and age. Oral Radiol 2023;39:191-7. [Crossref] [PubMed]
20. Cairo F, Pagliaro U, Buti J, et al. Root coverage procedures improve patient aesthetics. A systematic review and Bayesian network meta-analysis. J Clin Periodontol 2016;43:965-75. [Crossref] [PubMed]
21. Szczypiński PM, Strzelecki M, Materka A, et al. MaZda--a software package for image texture analysis. Comput Methods Programs Biomed 2009;94:66-76. [Crossref] [PubMed]
22. Santamaria MP, Neves FLDS, Silveira CA, et al. Connective tissue graft and tunnel or trapezoidal flap for the treatment of single maxillary gingival recessions: a randomized clinical trial. J Clin Periodontol 2017;44:540-7. [Crossref] [PubMed]
23. Werner N, Heck K, Walter E, et al. Probing pocket depth reduction after non-surgical periodontal therapy: Tooth-related factors. J Periodontol 2024;95:29-39. [Crossref] [PubMed]
24. Qiao J, Wang S, Duan J, et al. The accuracy of cone-beam computed tomography in assessing maxillary molar furcation involvement. J Clin Periodontol 2014;41:269-74. [Crossref] [PubMed]
25. Scarfe WC, Azevedo B, Pinheiro LR, et al. The emerging role of maxillofacial radiology in the diagnosis and management of patients with complex periodontitis. Periodontol 2000 2017;74:116-39. [Crossref] [PubMed]
26. Hatamikia S, Steiner E, Muniya EJ, et al. RadRepro CBCT: An Open-Access CBCT Phantom Dataset for Improved Standardization and Reproducibility of Radiomics Research. Sci Data 2026;13:454. [Crossref] [PubMed]
27. Fave X, Mackin D, Yang J, et al. Can radiomics features be reproducibly measured from CBCT images for patients with non-small cell lung cancer?. Med Phys 2015;42:6784-97. [Crossref] [PubMed]
28. Willam M, Eckl M, Oppitz H, et al. Radiomics feature stability and distinction power in organic low-contrast phantoms for novel CBCT imaging. Z Med Phys 2025;S0939-3889(25)00101-1. [Epub ahead of print]. [Crossref] [PubMed]