Quantification of Supracrestal Tissue Attachment Using High-Frequency Ultrasound Authors: Suhel Khan, Pie Yun Tsai, Brett Weatherington, Jesse V Jokerst Published: 2026 ABSTRACT AbstractObjectives:To develop and validate a noninvasive ultrasound (US)-based framework for quantifying supracrestal tissue attachment (STA) and evaluating tooth-type variation in STA dimensions, to assess the accuracy of conventional periodontal probing against photographic ground truth (GT), and US images.Methods:A high-frequency, miniaturized side-facing US transducer was used to simultaneously image gingival soft tissue, alveolar bone, and periodontal probes. Ex vivo swine jaws were imaged to compare US-derived probing depth and STA measurements with ground-truth photographic measurements and conventional manual probing. Measurement bias and agreement were evaluated using statistical analyses. Observed anatomical trends were further examined using full-mouth human cadaver imaging and validated in vivo in human subjects. Power analysis was performed to determine sensitivity for detecting biologically relevant differences between tooth types.Results:US-derived probing depth measurements showed lower bias (0.23 ± 0.42 mm) relative to GT than manual probing (0.54 ± 0.25 mm) and enabled direct visualization of STA with a periodontal probe. Agreement analyses showed improved accuracy of US measurements compared with conventional probing. Systematic differences in STA were identified between molar (1.86 ± 0.88 mm) and premolar (2.59 ± 0.53 mm) teeth in ex vivo specimens. These trends were confirmed in full-mouth cadaver imaging and in vivo human validation. Power analysis indicated 80% statistical power to detect clinically meaningful differences in STA dimensions across tooth types.Conclusions:STA is not uniform across the dentition and varies depending on the tooth type. High-frequency US provides a robust, noninvasive modality for direct anatomical measurement and may improve the quantitative assessment of periodontal structures in clinical practice.Advances in knowledge:This study challenges the long-standing assumption of uniform STA and introduces a validated US-based method for direct anatomical measurement. The findings highlight clinically relevant tooth-type variation and support the integration of high-frequency US into periodontal diagnostics and treatment planning.Keywords: Supracrestal tissue attachment (STA), biological width (BW), periodontal probing, ultrasound imaging, tooth-type variation, statistical validation Introduction The dentogingival junction (DGJ) is an essential interface between the tooth and periodontal tissues, forming a biologic seal that protects the periodontium from damage and infection [1, 2]. It includes the junctional epithelium (JE) and the supracrestal connective tissue (CT) attachment, which together define the supracrestal tissue attachment (STA), also known as biologic width (BW) [3, 4]. Violation of this zone through caries, restorations, or surgery can cause inflammation, attachment loss, and bone resorption [5]. The STA is rooted in classic histologic studies, notably Gargiulo et al.’s work measuring the dentogingival junction during passive eruption [6], which reported average values of 0.97 mm for JE and 1.07 mm for CT attachment. It is clinically assumed that the periodontium maintains a predictable attachment dimension since the epithelium component varies the most, while the CT stays constant [6]. This histologic average has been widely adopted as a universal constant STA across the dentition [7, 8]. Vacek et al. is another study on STA with measurements of JE 1.14 mm and CT 0.77 mm [9]. These mean values from histologic studies continue to guide restorative margins, crown lengthening, and periodontal planning despite the lack of STA data regarding tooth type [7]. In clinical practice, gold-standard periodontal diagnosis is typically based on periodontal probing and dental radiographs. Probing provides clinical measurements like periodontal probing depth (PPD), clinical attachment level (CAL), and gingival height (GH). In contrast, radiographs assess hard-tissue information but cannot monitor soft-tissue changes [10]. Unlike ultrasound, radiography exposes patients to ionizing radiation. [11, 12]. Currently, STA estimated from periodontal probing, transgingival bone sounding, and radiographic [13] methods do not provide direct visualization of STA and rely on tactile feedback [14]. Notably, probing can penetrate healthy tissues, overestimating sulcus depth and misjudging attachment levels. Studies show clinical measurements often differ from histologic standards [15]. Furthermore, probing of STA between tooth types and surfaces were performed using unpaired two-tailed t-tests, with statistical significance defined as p < 0.05. The suitability of sample size for determining measurement differences between teeth was estimated via power analysis for a two-tailed significance test with 95% significance (α=0.05), 80% power (β= 0.20), variance = 0.3 mm2, and minimum differences of 0.4, 0.5, or 1.0 mm. Box-and-whisker plots were combined with an unpaired, two-tailed t-test to compare different measurement methods. All statistical analyses were performed using GraphPad Prism 10 (San Diego, CA) and Microsoft Excel (Redmond, Washington). We assessed how consistently measurements within each group agreed using the intraclass correlation coefficient (ICC) analysis with a 95% confidence interval (CI), applying an average-measure, absolute agreement, two-way mixed model [33]. ICC values were interpreted as follows: <0.5 = poor, 0.5–0.75 = moderate, 0.75–0.90 = good, and >0.9 = excellent reliability [34]. The statistical analysis was conducted using IBM SPSS 30.0.0.0 (IBM, NY, USA). The authors utilized artificial intelligence tools to assist with proofreading and language refinement. All final content, interpretations, and conclusions were critically reviewed and approved by the authors. Results US imaging of clinical periodontal probe to assess probing depth Figure 2a shows the PL57x_1-128 transducer, which enables simultaneous US imaging during periodontal probing. US images of swine teeth before and during the clinical probing are shown in Figures 2 (b–e). In both molar and premolar teeth, the gingival soft-tissue boundary and anatomical landmarks were annotated. The periodontal probe appeared as a highly echogenic structure with a characteristic metallic tail artifact [30] during insertion, highlighted with a white circle (Figure 2c). This direct visualization enabled localization of the probe tip relative to the gingival margin, sulcus base, and bone crest, providing the basis for image-based measurements, where 1 = iPD, 2 = iSTA, and 3 = iGH. In Figure 2f, the measurements from the US images of the clinical probe show a clear shift from molars to premolars, mean iPD decreases from 3.22±1.16 mm to 2.2±0.64 mm, while mean iSTA increases from 1.93±1.15 mm to 2.5±0.71 mm, respectively. Simultaneous ex vivo ultrasound imaging of pocket depth with clinical probing. a) The schematic shows a side-facing miniaturized transducer imaging a tooth along with the Williams probe. b, c) US image of a molar tooth before and during probing. The circled area shows a metallic echoic tail from the Williams probe in the US image (c). d, e) Representative US images of a premolar tooth before and during the probing. The US can image the clinical probe while probing and provide an exact dimension of the supracrestal tissue attachment (iSTA). f) The results show a clear difference from molars to premolars, mean iPD decreases from 3.22±1.16 mm to 2.2±0.64 mm, while mean iSTA increases from 1.93±1.15 mm to 2.5±0.71 mm. 1: image based-pocket depth (iPD), 2: image-based STA (iSTA), 3: image-based gingival height (iGH) measured as the distance between the gingival margin (GM) and the alveolar bone crest (ABC), and CEJ: cementoenamel junction. In this ex vivo study, we imaged 82 swine teeth (34 molars and 48 premolars) and evaluated agreement in pocket depth measurements obtained as iPD, PPD, and GT using US, a conventional probing, and photographic ground-truth. Figure 3a shows the box and whisker plots of pocket depth measurements on the buccal surface of molar and premolar groups. Figure 3b shows the pocket depth measurements on the lingual surfaces. The mean pocket depths for buccal molars were 2.97 ± 1.15 mm, 2.53 ± 1.02 mm, and 3.07 ± 0.93 mm for iPD, PPD, and GT, respectively. The mean pocket depths of buccal-premolar teeth were 2.07 ± 0.57 mm, 1.80 ± 0.63 mm, and 2.30 ± 0.60 mm, respectively. In Figure 3c, Bland–Altman summary of pairwise method comparisons showed better agreement of US with GT on the buccal surface (molar 0.23 ± 0.42 mm, premolar 0.42 ± 0.30 mm) while the lingual surface showed larger SD values (molar 0.05 ± 0.69 mm, premolar 0.35 ± 0.57 mm) in the US comparisons. This higher variability is likely due to the greater curvature of the lingual tooth surface, which can introduce more US imaging artifacts and make landmark identification less consistent. Overall, the data show that US provides good agreement with GT, with variability depending on tooth surface geometry. These probing depth measurements were used to extract the STA of the different tooth types. Comparison of pocket depth measured by ultrasound, clinical probing, and photographic GT in ex vivo swine teeth. (a, b) Box-and-whisker plots show image-based probing depth (iPD), clinical probing depth (PPD), and photographic ground truth (GT) for molar and premolar teeth on the buccal and lingual surfaces. US-derived measurements were closer to GT than clinical probing, especially on the buccal surface. On the buccal surface, mean pocket depths in molars were 2.97 ± 1.15 mm, 2.53 ± 1.02 mm, and 3.07 ± 0.93 mm for iPD, PPD, and GT, respectively, whereas in premolars they were 2.07 ± 0.57 mm, 1.80 ± 0.63 mm, and 2.30 ± 0.60 mm. On the lingual surface, the corresponding values were 2.48 ± 0.98 mm, 2.00 ± 0.93 mm, and 2.59 ± 0.70 mm for molars, and 1.94 ± 0.59 mm, 1.67 ± 0.53 mm, and 2.22 ± 0.67 mm for premolars. (c) Bland–Altman summary of pairwise method comparisons by tooth type and surface. US showed better agreement with GT on the buccal surface, while the lingual surface showed larger SD values in the US comparisons. This higher variability is likely due to the greater curvature of the lingual tooth surface, which can introduce more US imaging artifacts and make landmark identification difficult. Quantitative evaluation of supracrestal tissue attachment (STA) measurement Figure 4 illustrates image-based measurements of iGH, iPD, and iSTA alongside the clinical periodontal probe (PPD and STAc) and the photographic (GT, STAGT) method used for validation (Figure 4 a–c). Bland–Altman analyses were performed to compare the differences between iGH and probing depth across the US, clinical probing, and photographs. The observed differences arise from the CT and JE between the ABC and the gingival sulcus, collectively referred as the supracrestal tissue attachment [3]. In the group of molar teeth, the mean STA (bias) values were 2.30 ± 0.97 mm, 1.86 ± 0.88 mm, and 1.75 ± 0.97 mm for clinical probing, US, and photographs, respectively (Figures 4 d–f). This demonstrated that the STA estimated from clinical probing (2.30 mm) was larger than that derived from US and the GT photographs. STA for premolar teeth were 2.86 ± 0.62 mm, 2.59 ± 0.53 mm, and 2.35 ± 0.57 mm for STAc, iSTA, and STAGT, respectively (Figures 4 g–i). Here again, clinical probing underperformed relative to US and photographs, and US showed higher agreement with the GT. Another observation is that the premolar tooth group showed an STA value more than 30% higher than the molar tooth group in all three modalities. These results support the US’s ability to measure the anatomical components contributing to STA with better agreement. Estimating supracrestal tissue using ultrasound, clinical probing, and photographic ground-truth (GT) of ex vivo swine teeth. (a) Conventional periodontal chart showing probing depth. (b) US image of a swine tooth with image-based iGH, iPD, and iSTA annotated. (c) GT photograph of the swine tooth showing the periodontal probe where (A - B) corresponds to the actual probing depth. A is the known probe marking length, and B is the distance from the GM to the marking terminus. (d-i) Bland-Altman analysis between the measurement methods shows a bias comparing differences between iGH and probing depth. The observed differences arise from the connective tissue and junctional epithelium between the ABC and the gingival sulcus, collectively referred to as the supracrestal tissue attachment (STA) [3]. Mean bias (STA) for molar teeth 1.86 ± 0.88 mm shows better agreement with GT (1.75 ± 0.97 mm) compared to conventional probing (2.30 ± 0.97 mm). While comparing STA from different tooth type, premolars show >30% higher STA (2.59 ± 0.53 mm) compared to the molar teeth (1.86 ± 0.88 mm). All values shown in the figure are in mm. The main advantage of ultrasound-guided periodontal probing was direct visualization of iSTA instead of calculating iSTA from iGH-iPD. iSTA was determined using two complementary definitions: (1) indirectly as iSTA = iGH-iPD, and (2) directly as the distance from the tip of the periodontal probe to ABC as visualized on US during probing (Figure 2d). The two approaches showed very strong agreement, with a highly significant correlation (p < 0.0001, r = 0.99). Bland–Altman analysis further demonstrated minimal bias between the two measurements, with mean differences of 0.02 mm for molars and 0.03 mm for premolars (Figure A.1). These differences are approximately an order of magnitude smaller than the discrepancies observed between different measurement modalities, indicating that both US-based definitions provide highly consistent estimates of iSTA. Further, extracted STA values from tooth-groups were compared to assess systematic variability across tooth type (Figure 5). Figures 5(a,b) show box-and-whisker plots of iSTA, STAc, and STAGT for molar and premolar teeth on buccal and lingual surfaces. In both buccal and lingual cases, we noted statistically significant differences between clinical measurements and GT measurements (unpaired, two-tailed t-test, p < 0.0001). On the lingual surface, the difference between US and GT was not statistically significant (p = 0.081, Figure 5b). The mean iSTA for premolar teeth was 2.59 ± 0.53 mm, [1.53, 3.64](mean,[LoA]) significantly higher than the mean iSTA 1.86 ± 0.88 mm, [−0.16, 3.88] for molar teeth. In Figures 5 (c–e), Bland–Altman analysis was used to quantify method-dependent bias. In molars (n = 34), US and clinical probing showed a bias of −0.44 mm (iSTA lower than STAc), with limits of agreement of [−1.938, 1.053] mm and 83.8% of measurements within the clinically acceptable ±1 mm range(Table 1). In premolars (n = 48), US similarly underestimated clinical probing (bias = −0.26 mm) with LoA of [−1.580, 1.041] mm and 80.5% within ±1 mm (Figure 5 (f–h), Table 1). Agreement with GT was higher for both tooth types: in molars, US vs GT and clinical vs GT showed 93.6% and 93.5% agreement within ±1 mm, respectively, and in premolars the corresponding agreements were 85.4% and 90.2%. The tooth-type dependence and the consistent offset between US and probe-based estimates suggest STA variability across molars and premolars and highlight systematic differences between imaging-based and probing-based STA measurements. Comparison between measured-STA using US, periodontal probe, and photographic GT using ex vivo swine teeth. (a, b) Box-and-whisker plots compare molar and premolar teeth on US (iSTA), periodontal probing (STAc), and GT photographs (STAGT). Pairwise comparison values are p-values (unpaired, two-tailed t-test) (ns = 0.081, **** = <0.0001). The molar group exhibits greater probing depth (iPD) (2.53 ± 1.02 mm) and reduced iSTA (1.86 ± 0.88 mm) compared to premolars (2.59 ± 0.53 mm). (c-e) Bland–Altman analysis in molars (n = 34), US (iSTA) shows a bias of 0.44 ± 0.76 mm relative to STAc with 83.8% of agreement within ± 1 mm, whereas agreement with GT is higher (iSTA vs STAGT : 93.6%, STAc vs STAGT : 93.5%). (f-h) In premolars (n = 48), US again slightly underestimates clinical values (0.36 ± 0.66 mm) with 80.5% within ±1 mm, while comparisons against GT show improved agreement (US vs GT: 85.4%; clinical vs GT: 90.2%). Overall, the biases are modest (within about ±0.5 mm), and the highest agreement is observed for comparisons involving GT, particularly in molars. Table 1Bland–Altman summary for molar and premolar STA measurements. Bias is the mean difference (Method 1 – Method 2). n = sample size. B = Bias. LoA = Bias ± 1.96 SD. % Agr. = % Agreement within ±1 mm.Tooth typeComparison n B (mm), [LoA] (mm)Agr. ±1 mmMolarUS vs Clinical34−0.44 [−1.938, 1.053]83.8%MolarUS vs GT34+0.11 [−1.290, 1.500]93.6%MolarClinical vs GT34+0.54 [−0.211, 1.306]93.5%PremolarUS vs Clinical48−0.26 [−1.580, 1.041]80.5%PremolarUS vs GT48+0.23 [−1.038, 1.502]85.4%PremolarClinical vs GT48+0.50 [−0.243, 1.246]90.2% Bland–Altman summary for molar and premolar STA measurements. Bias is the mean difference (Method 1 – Method 2). n = sample size. B = Bias. LoA = Bias ± 1.96 SD. % Agr. = % Agreement within ±1 mm. Human cadaver imaging and in vivo feasibility demonstration The iSTA method was first optimized and validated using an ex vivo swine model and then applied to human cadaver imaging. Full-mouth human cadaver US imaging was performed across anterior, premolar, depths are subjective and have been shown to vary across different providers, or even the same provider with repeat measurements [16]. These discrepancies have important clinical implications. Relying on a fixed STA can lead to improper subgingival margin placement. It may also cause excessive bone removal during crown lengthening or persistent inflammation if the STA is encroached upon [17, 18]. These risks are increased by the lack of quantitative data on whether dentogingival dimensions vary systematically by tooth type. Although there are well-known differences in crown morphology, root anatomy, gingival thickness, and functional loading across the dentition, their impact on dentogingival dimensions remains unclear. In the context of dental implants, studies consistently show that the peri-implant STA is frequently larger (4.0–4.5 mm) than that around natural teeth [19]. The peri-implant mucosa forms a soft tissue seal similar to the natural STA. Factors such as implant surface properties, surgical technique, and implant position can influence the dimensions and stability of this peri-implant mucosa, with STA violations increasing the risk of peri-implantitis, mucosal recession, and bone loss [20]. Ivanovski et al. emphasize that respecting the biological space during implant placement and restoration is essential for long-term tissue health and implant success [21]. Ultrasound (US) imaging has emerged as a valuable diagnostic tool in dental research and clinical practice, offering unique advantages for assessing oral and periodontal structures [22, 23, 24]. Dental imaging has typically relied on radiographic techniques, which are useful for seeing hard tissues and exposing patients to ionizing radiation [10, 25]. In contrast, US imaging uses high-frequency sound waves to generate real-time images, enabling the visualization of both soft and hard tissues non-invasively without using ionizing radiation [26, 27]. Recent advances in miniaturized, high-frequency transducers have enabled intraoral imaging of periodontal anatomy with superior resolution to identify the gingival margin, cementoenamel junction, and alveolar bone crest [23, 26]. However, US imaging of these anatomical features was unable to accurately quantify the STA attachment and probing depth [24]. In this study, we directly image the STA beneath the soft tissue using a high-frequency US transducer and a periodontal probe. Using an ex vivo swine tooth model, we imaged teeth with the periodontal probe before and during probing. US-imaged probing depth and STA measurements were compared with conventional periodontal chart, and validated both against ground-truth photographic images. This approach enables a rigorous, quantitative evaluation of US image precision and reliability for STA assessment. Through a combination of ex vivo swine model, human cadaver, and in vivo human measurements, we show that supracrestal tissue attachment systematically varies among different tooth groups. Methods This study aims to determine whether STA is consistent across different tooth types. Understanding tooth-specific STA variation is important for accurate interpretation and for distinguishing normal anatomy from disease-related changes. Figure 1a shows the anatomical components contributing to STA, including the JE and CT attachment. US images clearly identified important landmarks such as the crown, gingival margin (GM), cementoenamel junction (CEJ), alveolar bone, and alveolar bone crest (ABC). Figure 1b shows examples of high-crest, low-crest, and very-low-crest anatomies, demonstrating variation in the position of the ABC relative to sulcus depth. Figures 1(c–e) illustrate the three methods used to measure pocket depth: conventional periodontal probing, US imaging, and photographic ground-truth to assess the STA. Supracrestal tissue attachment anatomy, classification, and measurement. (a) Schematic illustration of the dentogingival junction showing the components of supracrestal tissue attachment (STA), including the junctional epithelium and supracrestal connective tissue (CT) attachment, measured from the base of the gingival sulcus to the crest of the alveolar bone. (b) Representative categories of biologic width based on the vertical relationship between the gingival margin and alveolar crest (high crest, low crest, and very low crest). (c) Three measurement methods for assessing pocket depth (PD) are illustrated: conventional periodontal probing depth (PPD), US imaging-based pocket depth (iPD), and photograph-based ground truth (GT) measurements of pocket. US imaging enables noninvasive visualization of soft tissue and alveolar bone, and allows direct observation of the periodontal probe within the clinical probe (Williams), facilitating comparison between clinical probe-based estimation of STA and true anatomic dimensions. Ultrasound imaging system and clinical probe visualization US imaging was performed using a miniaturized side-facing high-frequency intraoral transducer (PL57x_1-128; VisualSonics Corp.) (Figure 2a). PL57x_1-128 has a 128-element linear array operating at 40 MHz, with a bandwidth of 25 to 57 MHz, resolution of 49 μm, and integrates with a data acquisition (Vevo-F2, VisualSonics, Inc., USA) system (see more details [26]). The miniaturized PL57x_1-128 allows access to all lingual and buccal surfaces of human dentition, which was previously restricted to the buccal surfaces [24]. The side-facing configuration was used to capture the tooth surface, gingival soft tissue, and alveolar bone within a single imaging plane while permitting simultaneous periodontal probing. A Williams periodontal probe (PCP-UNC 15, Hu-Friedy) was used for all clinical measurements (transgingival probing; gold standard method) [28, 29]. US images were acquired both before and during active probing to the pocket depth and to visualize the probe position in real time. During probe insertion, the metallic probe shaft produced a characteristic high-intensity echogenic signature [30] that enabled direct identification of probe tip position relative to the GM, CEJ, and ABC. Ex vivo swine model and ground-truth validation Six ex vivo porcine mandibles containing intact molar and premolar teeth were obtained and used to enable controlled measurement validation under direct visualization of the periodontal probe. The selected swine teeth were 34 molar scans (18 buccal side and 16 lingual side) and 48 premolar scans (18 buccal side and 30 lingual side). Swine jaws were selected based on anatomical similarity to human dentition in tooth root structure and periodontal tissue architecture [31]. PL57x_1-128 transducer head was encased in disposable couplant sleeve and sterile adhesive film (Tegaderm) [32]. The high-frequency transducer visualizes anatomical landmarks such as GM, CEJ, and ABC while simultaneous periodontal probing was performed using a Williams probe. US images of the swine teeth were acquired before and during the periodontal probing. Image-based gingival height (iGH) was defined as the distance from the GM to the ABC [24]. Image-based probing depth (iPD) was defined as the distance from the GM to the sulcus base as identified by the Williams probe in US images. The STA was defined as the linear distance from the base of the gingival sulcus to ABC, encompassing both the JE and supracrestal CT attachment. The image-based STA (iSTA) was defined as the linear distance from the tip of the Williams probe to ABC in the US image. Although the iSTA imaging method can only be performed with a periodontal probe in the mouth, it is a fast and simple approach to make a measurement that is quite difficult to obtain otherwise. For each tooth, pocket depth was measured on buccal and lingual surfaces using three independent methods: (i) conventional clinical periodontal probing depth (PPD) using a Williams probe, (ii) US image-based probing depth (iPD), and (iii) photograph-based GT (GT). Ground-truth probing depth (GT) was determined from calibrated probe markings visible in high-resolution photographs by a trained dentist (B.W.). The distance GT from the GM to the visible probe reference mark was measured using digital image analysis software (ImageJ, version 1.48v) used without any additional processing or modifications that could alter their integrity. Only a standard line measurement tool within ImageJ was used to measure the distances between the visible reference marking line of the clinical probe and GM to extract the probing depth. Minimum measuring accuracy was 5 μm for each pixel size of the US image. This value was converted to absolute insertion depth based on known probe calibration. US-derived parameters (iPD, iGH, and iSTA) were measured directly from US images using the VevoLAB software. Clinical STA (STAc) was measured from the difference between GH and PPD. Ground truth STA (STAGT) was measured from the difference between GH and GT. All clinical probing measurements were performed by a single examiner blinded to US measurement values. To ensure robustness, a blinded test was performed to assess the reliability of the results. In vivo human imaging Full-mouth US imaging was performed on a fresh human cadaver jaw obtained from Unite Tissue Network. US imaging was conducted across anterior teeth (incisors and canines), premolars, and molars using the same anatomical definitions and imaging protocol established in the swine model. The cadaver specimen was frozen and thawed before imaging. Additionally, in vivo US imaging was performed on healthy human volunteers with simultaneous clinical probing to demonstrate real-time probe visualization under clinical conditions. Imaging focused on anterior (incisor) and premolar teeth to validate the technique in living tissue with active blood flow and tissue turgor. The study protocol involving human subjects was approved by the UCSD Institutional Review Board (project 170912) and was conducted in accordance with ethical standards set by the IRB and the Helsinki Declaration of 1975. Three healthy adult volunteers with good oral hygiene were enrolled after providing written informed consent. Exclusions were made for individuals with bleeding disorders, bloodborne pathogen infections, or acute oral infections. The subject was seated supine and imaged with the PL57x 1–128 transducer, which was wrapped in a disposable couplant sleeve and covered with sterile adhesive film (Tegaderm). The sterilization protocol was followed throughout the study. The hydrogel couplant was single-use and was changed for each subject. Statistical analysis and method comparison Agreement between measurement methods was evaluated using Bland–Altman analysis, with photograph-based GT serving as the reference standard. Bias is reported as (mean ± SD) with 95 % limits of agreement (LoA). Analyses were stratified by tooth type (molar vs. premolar) and surface (buccal vs. lingual) to evaluate potential sources of systematic variation. Group-level comparisons and molar teeth to obtain quantitative measurements of probing depth, gingival height, and supracrestal tissue attachment (Figure 6a). The mean probing depth measured by US was 1.93 ± 0.65 mm for anterior teeth, 1.43 ± 0.52 mm for molars, and 1.06 ± 0.23 mm for premolars (Figure 6b). Figure 6c shows the iGH measured from US images of all teeth. Figure 6d shows the direct iSTA values measured from the US images. Statistical analysis revealed that premolars (1.07 ± 0.40 mm) exhibited significantly greater iSTA than anterior teeth (0.44 ± 0.37 mm) and molars (0.6 ± 0.23 mm) (Figure 6d). The cadaver measurements showed the same qualitative pattern observed in the ex vivo swine model: anterior teeth exhibited the smallest STA, while premolars exhibited the largest. Full-mouth human cadaver quantitative imaging and in vivo human feasibility demonstration. (a) Tooth type-based classification of a human cadaver imaging. Blue: Molars, Pink: Premolars, and Gray: Anterior. (b) US image-based probing depth (iPD) measurements across anterior, molar, and premolar teeth in the cadaver specimen. (c) Image-based gingival height (iGH) comparison among anterior, molar, and premolar teeth. (d) Image-based supracrestal tissue attachment (iSTA) comparison across tooth groups, demonstrating the highest iSTA (1.07 ± 0.40 mm) in premolars and the lowest in anterior teeth (0.44 ± 0.37 mm) in the cadaver dataset. (e, f) Representative in vivo US image and corresponding schematic of an incisor acquired during simultaneous clinical probing in a healthy volunteer. (g, h) Representative in vivo US image and corresponding schematic of a premolar acquired during simultaneous clinical probing in a healthy volunteer. Panels (e–h) are included to demonstrate clinical feasibility and real-time visualization of the periodontal probe in living tissue, whereas panels (a–d) present the quantitative cadaver analysis. In addition, representative in vivo images of healthy human volunteers are presented in Figures 6e–6h. We then confirmed that the simultaneous US imaging with the probing method is applicable to healthy human subjects. iSTA measured from the incisor tooth was 0.96 mm, whereas the iSTA for the premolar tooth was 1.61 mm (Figures 6e, 6g). The volunteer images were not intended as a statistically powered validation cohort, but rather to demonstrate the translational applicability of the imaging approach. Tables A.2 & A.3 summarize intra-rater and inter-rater reliability for US-derived iGH, iPD, and iSTA measurements across molar and premolar teeth. Two blinded raters (Rater1, Rater2) participated in this study. Each rater performed three repeated measurements. Intra-rater repeatability was consistently excellent for all metrics, with ICCs ≥ 0.958 across surfaces and tooth types, and near-perfect agreement for iSTA (ICC ≈ 0.996 – 1.00). Inter-rater reliability between Rater1 and Rater2 was also high overall, with all-surface ICCs 0.919 – 0.988 for molars and 0.911 – 0.966 for premolars. Reliability remained strong on buccal and lingual surfaces for iPD and iSTA (typically ICC ≥ 0.903), while iGH showed comparatively lower agreement on the buccal surface for molars (ICC = 0.650), indicating greater variability for iGH between raters. Since iGH corresponds to the longest measurement distance, it is more prone to cumulative measurement error, which likely contributed to the reduced inter-rater reliability (Table A.3). Discussion Conventional periodontal probing provides an indirect estimate of pocket depth based on tactile feedback and probe markings, without visualization of underlying supracrestal tissue anatomy [6]. Radiography can image alveolar bone but less sensitive to soft tissue [10]. In contrast, US imaging enables direct visualization of gingival soft tissues and alveolar bone [17]. This allows assessment of probe position relative to true anatomic boundaries and reveals potential discrepancies between probe-based estimates and US imaging-derived pocket depth. This study demonstrates that high-frequency US imaging provides a robust, noninvasive method for quantifying supracrestal tissue attachment and assessing periodontal probing accuracy beneath the soft tissue, while simultaneously imaging the periodontal probe within the sulcus. By directly visualizing the anatomical landmarks of the dentogingival junction, including the GM, CEJ, and ABC, US imaging overcomes the limitations of conventional probing and radiographic techniques, which lack direct visualization and are prone to measurement bias. US-derived measurements of pocket depth and STA exhibit significantly lower bias than conventional periodontal probing with respect to GT photographic measurements. Specifically, Bland–Altman analysis revealed that US imaging of probing depth had a mean bias 0.23 ± 0.42 mm for molars and 0.42 ± 0.30 mm for premolars on the buccal surface, which shows better agreement with GT compared to higher biases observed with manual probing (0.54 ± 0.25 mm and 0.75 ± 0.71 mm) (Figure 3c). The apparent combination of lower bias but higher SD for iPD relative to GT reflects the distinction between accuracy and precision. A smaller mean difference indicates reduced systematic error, whereas a larger SD indicates greater variability among individual measurements. In our study, US-derived iPD was closer to GT on average than manual probing, but exhibited a wider spread, especially in premolars. Importantly, repeatability of US-derived measurements remained high, as supported by excellent intra-rater and high inter-rater ICC values for iPD, suggesting that this variability does not primarily arise from poor measurement reproducibility. Rather, it likely reflects sample-level heterogeneity and the greater sensitivity of US to anatomical variation around the sulcus and probe tip. Tactile probing appears to overestimate pocket depth due to tissue penetration and variability in probing pressure [14]. The stronger agreement (93.6%) between US and ground-truth measurements supports imaging-based methods for more accurate periodontal evaluation. STA represents the soft tissue complex located between ABC and the gingival sulcus, consisting primarily of connective tissue and junctional epithelium [9]. In healthy dentition, the dimensions of this attachment are commonly reported to be approximately 2–3 mm, although variability across individuals and tooth types has been reported [9, 6]. Accurate assessment of STA is clinically important because it reflects the structural relationship between periodontal soft tissues and supporting bone. However, most clinical measurements rely on tactile periodontal probing, which is inherently limited by tissue penetration, probe angulation, and operator-dependent variability [16, 11]. In the present study, US-based measurements of STA were compared with clinical probing and photographic ground-truth measurements (Figure 4). For molar teeth, the mean STA values were 2.30 ± 0.97 mm for clinical probing (STAc), 1.86 ± 0.88 mm for US (iSTA), and 1.75 ± 0.97 mm for ground-truth photographs (STAGT). Thus, probing produced values approximately 24% higher than US and 31% higher than the photographic reference, indicating a consistent overestimation of STA by tactile probing. Similar trends were observed in premolar teeth, where STA values were 2.86 ± 0.62 mm (STAc), 2.59 ± 0.53 mm (iSTA), and 2.35 ± 0.57 mm (STAGT) (Figures 4 g–i). The closer correspondence between US and photographic measurements suggests that US more accurately resolves the anatomical boundaries contributing to STA. Method-dependent differences were further quantified using Bland–Altman analysis (Figures 5 c–h, Table A.1). In molars (n = 34), US measurements were slightly lower than clinical probing with a bias of −0.44 mm and LoA of [−1.938, 1.053] mm, with 83.8% of measurements falling within the clinically acceptable ±1 mm range. In premolars (n = 48), the bias between US and probing was −0.26 mm, with LoA of [−1.580, 1.041] mm, and 80.5% of measurements within ±1 mm. Importantly, agreement with the GT was higher: 93.6% of US measurements in molars and 85.4% in premolars fell within ±1 mm of the photographic reference, compared with 93.5% and 90.2% for clinical probing, respectively. Another notable observation is the tooth-type dependence of STA measurements. As illustrated in Figures 5 a–b, premolars exhibited larger STA values than molars across all measurement modalities. The mean US-derived STA for premolars (2.59 ± 0.53 mm) was approximately 39% higher than that for molars (1.86 ± 0.88 mm). This difference suggests potential anatomical variation in the dentogingival complex across tooth groups, which has been previously reported in histological studies of STA and dentogingival junction dimensions [6]. Such variability highlights the importance of considering tooth-specific anatomical differences when evaluating periodontal measurements. Further, both ex vivo swine models and human cadaver imaging revealed systematic differences in STA between molars, premolars, and anterior teeth, with premolars exhibiting the largest STA and anterior teeth the smallest (Figure 6). These findings align with emerging literature suggesting that anatomical and functional variations among tooth groups influence the dimensions of the supracrestal tissue attachment. Recognizing this variability is crucial for clinicians, as reliance on a fixed STA value may lead to inappropriate restorative margin placement or excessive bone removal during crown lengthening procedures [17]. This study has several limitations. First, the reference standard was based on photographic measurements in the ex vivo model rather than using histology. Photography enables direct visualization of probe position and is quantitative, but it does not offer microscopic confirmation of the true histological boundaries of the junctional epithelium and connective tissue attachment. Second, the primary validation was performed in ex vivo swine mandibles, which cannot fully replicate the biological and mechanical properties of living human periodontal tissues. Third, the in vivo human component involved a limited number of healthy volunteers. While the human study showed the feasibility of the method, it did not offer a statistically powered clinical validation. Future studies will include larger and more diverse in vivo cohorts. The suitability of the sample size for determining measurement differences between teeth was estimated using power analysis for a two-tailed significance test at 95% (α=0.05), 80% power (β=0.20), variance = 0.3 mm2, and minimum differences of 0.4 mm, 0.5 mm, or 1.0 mm (Figure A.2). The sample size of the swine teeth was larger than the minimum required (n = 32) number of samples to achieve 80% power. Conclusion This study establishes high-frequency US imaging as a precise, noninvasive method for quantifying supracrestal tissue attachment and assessing periodontal probing accuracy beneath the soft tissue, demonstrating greater agreement with ground-truth measurements. By revealing systematic differences in STA among tooth types, the findings challenge the traditional assumption of uniformity and underscore the importance of individualized assessment in clinical practice. Future research should focus on expanding in vivo studies to larger, more diverse populations, refining US protocols for routine clinical use, and exploring the integration of this technology to monitor periodontal therapy and guide restorative procedures. Supplementary Material Acknowledgments This work was supported by the National Institutes of Health under grant numbers R01 DE031307 and S10 OD032268. Footnotes Conflicts of interest The authors declare that they have no competing interests. Data availability The data underlying this article are available upon request to the corresponding author. References Associated Data Supplementary Materials Data Availability Statement The data underlying this article are available upon request to the corresponding author.