Echo intensity and gray-level co-occurrence matrix analysis of soft tissue grafting biomaterials and dental implants: an in vitro ultrasonographic pilot study

Abstract

Objective:

To characterize different allogeneic and xenogeneic soft tissue graft substitutes and to assess their echo intensity and grayscale texture-related outcomes by using high-frequency ultrasonography (HFUS).

Methods:

Ten samples from each of the following biomaterials were scanned using HFUS: bilayered collagen matrix (CM), cross-linked collagen matrix (CCM), multilayered cross-linked collagen matrix (MCCM), human-derived acellular dermal matrix (HADM), porcine-derived acellular dermal matrix (PADM), collagen tape dressing (C) and dental implants (IMPs). The obtained images were then imported in a commercially available software for grayscale analysis. First-order grayscale outcomes included mean echo intensity (EI), standard deviation, skewness, and kurtosis, while second-order grayscale outcomes comprised entropy, contrast, correlation, energy and homogeneity derive from the gray-level co-occurrence matrix analysis. Descriptive statistics were performed for visualization of results, and one-way analysis of variance with Bonferroni post-hoc tests were performed to relative assessments of the biomaterials.

Results:

The statistical analysis revealed a statistically significant difference among the groups for EI (p < .001), with the group C showing the lowest EI, and the IMP group presenting with the greatest EI values. All groups showed significantly higher EI when compared with C (p < .001). No significant differences were observed for energy, and correlation, while a statistically significant difference among the groups was found in terms of entropy (p < 0.01), contrast (p < .001) and homogeneity (p < .001). IMP exhibited the highest contrast, that was significantly higher than C, HADM, PADM, CCM and CM.

Conclusions:

HFUS grayscale analysis can be applied to characterize the structure of different biomaterials and holds potential for translation to in-vivo assessment following soft tissue grafting-related procedures.

Introduction

Soft tissue graft substitutes have been introduced to augment the soft tissue phenotype around teeth and dental implants, without the need for harvesting autogenous tissue from the palate, that has been associated with substantial post-operative morbidity and overall patient anxiety toward the procedure. A large variety of these biomaterials is available on the market, including human acellular dermal matrix (HADM), porcine-derived acellular dermal matrix (PADM), bilayered xenogeneic collagen matrix (CM), cross-linked xenogeneic collagen matrix (CCM) and multilayered cross-linked xenogeneic collagen matrix (MCCM). Clinicians have been using these graft substitutes either alone or as scaffolds in combination with cells or biologic agents. Although there are limited clinical trials comparing head-to-head different biomaterials, pre-clinical studies have shown that the structural properties of grafts play a key role on cell migration and differentiation. ^1–3

Current methods for characterizing the structure of soft tissue graft substitutes include histological analysis and microscopy-based technologies. ^4,5 Histology and scanning electron microscopy involve biopsies and destructive processing procedures, while other microscopy-based techniques have limited penetration depth and cannot visualize graft materials in all their thickness. ^4,5 Aside from the fact that histological analyses are time-consuming and microscopy-based technologies are relative expensive, the main limitation of these approaches is their application in human clinical trials when assessing the wound healing process following the use of soft tissue graft substitutes at different time points.

High-frequency ultrasonography (HFUS) is a relatively novel tool in dentistry. This technology is based on acoustic waves with high frequency (>20,000 Hz) that are transmitted into the body using a transducer that is also able to record the pulses echo back of the waves encountering boundaries between tissues. These echoes are then translated to images that can be visualize real-time during the examination. HFUS allows to visualize periodontal and peri-implant structures, as well as the healing following grafting procedures, without exposing patients to ionizing radiation. ^6–9 Among the several applications of conventional and HUFS routinely performed in the medical field, echo intensity and gray texture analysis are commonly utilized to diagnose pathological conditions, ¹⁰ differentiate benign and malignant lesions, ^11,12 and to evaluate muscle morphology and quality. ^13,14 Ultrasonography allows not only to visualize anatomical structures, but also to quantify interactions between ultrasound waves and biological tissues. ^15–17 It is reasonable to assume that this technology can provide new information on the structural properties of the different soft tissue graft substitutes commonly utilized for augmentation at teeth and dental implants. However, this possible application of HFUS has not been investigated in previous studies in the dental field.

Therefore, the aim of the present study is to utilize HFUS to characterize soft tissue graft substitutes and to assess their echo intensity and grayscale texture-related outcomes.

Methods and materials

Experimental design

The present in-vitro study involved the evaluation of five different soft tissue grafting materials (bilayered collagen matrix [CM, Mucograft, Geistlich Pharma North America, Princeton, NJ], cross-linked collagen matrix [CCM, Fibro-Gide, Geistlich Pharma North America, Princeton, NJ], multilayered cross-linked collagen matrix [MCCM, Ossix Volumax, Dentsply Sirona, Charlotte, NC], porcine-derived acellular dermal matrix [PADM, NovoMatrix, BioHorizons, Birmingham, AL], and human-derived acellular dermal matrix [HADM, AlloDerm SELECT RTM, BioHorizons, Birmingham, AL]), as well as a bone-level dental implant (IMP, Zimmer Tapered Screw-Vent 3.7 × 10 mm, Zimmer Biomet, Warsaw, IN) and a collagen tape dressing (C, HeliTape, Integra LifeSciences, Princeton, NJ) serving as control groups. Two sets of HFUS scans were performed. First, 10 biomaterials for each group were scanned twice. Then, 10 additional Cs, CMs, CCMs, MCCMs, PADMs and HADMs were positioned over IMPs to assess the reflection and propagation of the ultrasound waves through the soft tissue biomaterials.

Sample preparation

The Cs, CMs, CCMs, MCCMs, PADMs and HADMs were trimmed with fresh 15c blades (Salvin Dental Specialties, Charlotte, NC) to obtain grafts of 7 mm in height and 10 mm in width. For the CCM, the commercially available product 3 mm in thickness was utilized. The graft materials were irrigated with sterile saline and then positioned over a metal graft card (Hu-Friedy, Chicago, IL) for the first set of HFUS scans. IMPs were inserted in gel blocks that were prepared from ultrasound gel pads (Aquaflex, Parker Laboratories, Inc., Fairfield, NJ) with the goal of leaving approximately 1 mm of gel only at the side of the implant that was targeted for the HFUS scans. For the second set of scans, this thin layer of gel block at one aspect of the implant was removed in order to position the biomaterials (Cs, CMs, CCMs, MCCMs, PADMs and HADMs) directly in contact with the implant surface. The scans were then taken using a transmission gel (Aquaflex, Parker Laboratories, Inc., Fairfield, NJ) that was applied over the graft material.

Ultrasound settings

A 24 MHz (64 µm axial image resolution) and miniature-sized (approximately 30 mm in width, 18 mm in height, and 12 mm in thickness) transducer (L30-8, Zonare/Mindray, Mountain View, CA) was coupled with a commercially available ultrasound imaging device (ZS3, Zonare/Mindray, Mountain View, CA). The scans were performed by a periodontist that has worked for more than 5 years with HFUS in the dental field (LT). For the first set of scans, the transducer was oriented parallel to the long axis of the grafts and the IMPs. For the second set of scans, where the grafts were positioned over the implant surface at its coronal and middle portion perpendicularly to the long axis of the IMPs to mimic clinical procedures, the transducer was maintained parallel to the long axis of the IMPs, and perpendicular to the graft materials (Figure 1). ^18,19 To optimize the ultrasound images, the following settings of the ultrasound device were adjusted prior to the first scan and maintained then constant throughout the duration of the study: (i) compound spatial harmonics (FCSH24), (ii) gain 70 dB, (iii) depth 1.0 cm, and (iv) frequency rate 18 Hz.

Figure 1.

Ultrasonographic representation of the imaged biomaterials, including: a xenogeneic CCM), a collagen tape dressing (C and a MCCM, alone and when on top of a dental IMP. (a) Clinical view of a CCM. (b) Ultrasound scan of a CCM. (c-d) Clinical and ultrasonographic image of a CCM positioned on top of an IMP. The IMP area underneath the graft cannot be visualized. (e-f) Clinical and ultrasonographic image of a C positioned on top of an IMP. It is possible to appreciate the IMP underneath the C. (g-h) Clinical and ultrasonographic image of a MCCM positioned on top of an IMP. It is possible to appreciate the IMP underneath the MCCM. CCM, cross-linked collagen matrix; IMP, implant; MCMM, multilayered cross-linked collagen matrix.

Two single image frames for scan of interested were saved in “B-mode” in the Digital Imaging and Communications in Medicine (DICOM) and in the JPEG format. “B-mode” generates two-dimensional grayscale images in which brightness is the result of the returned echo signal and its strength, which depends on the acoustical properties of the encountered structures.

Ultrasound image analysis

DICOM files were imported in a public-domain software package (Horos^TM, v. 3.3.6, Horos Project), where the thickness of graft biomaterials was assessed as previously described. ^6,8 Then, the scans previously saved in JPEG format were imported in another public-domain software package (ImageJ, National Institutes of Health, Bethesda, MD) where they were converted in 8-bit JPEG images. The regions of interested (ROIs) were defined for each scanned biomaterial (Cs, CMs, CCMs, MCCMs, PADMs, HADMs, and IMPs) by a pre-calibrated operator (LM) under the supervision of an operator with expertise in dental ultrasonography (LT). Operator calibration consisted in two sets of repeated measurements—that were executed one week apart—on 10 collagen matrices that were not utilized in the present investigation. The intraexaminer reproducibility for Mean pixel/echo intensity (EI) was 0.88.

First-order grayscale texture features

First-order grayscale texture outcomes depict the characteristics of the pixel values within the ROI and do not depend on the neighboring areas/pixels. The following outcomes were extracted from the histograms:

• mean pixel/EI, that was set as the primary outcome of the study;

• standard deviation of the mean pixel grayscale value;

• skewness, that measures the asymmetry of pixel distribution based on the mean. The skewness for a symmetrical distribution of pixels is zero, while negative values of skewness imply that the distribution of the pixel intensity is skewed left and positive values of skewness indicates data distribution skewed right;

• kurtosis, that is a measure of whether the data/pixels are heavy-tailed or light-tailed relative to a normal distribution. Positive kurtosis indicates a peaked distribution, while negative kurtosis implies a flat distribution.

Second-order grayscale texture features

Second-order grayscale texture outcomes derive from the gray-level co-occurrence matrix (GLCM) analysis and provide information on the spatial relationship of image pixels/texture. ^{15–17,20,21} The GLCM functions describe the texture of an image by calculating how often pairs of pixels with certain values and in a specific relationship occur. GLCMs were computed by the software, with the following outcomes of interest that were then extracted and reported ^15–17 :

• entropy, that measures the randomness of intensity distribution. Homogeneous ROIs display high entropy values;

• contrast, that measures the weighted mean differences in intensity of neighboring pixels;

• correlation, that measures the correlation between intensities of neighboring pixels;

• energy (also known as Angular second moment), that measures the variety of intensity found in the image;

• homogeneity (also known as Inverse difference moment), that measures the uniformity of the texture, the frequency with which near identical intensities are adjacent to each other.

For the second set of scans, where Cs, CMs, CCMs, MCCMs, PADMs, and HADMs, were positioned on top of IMPs, the appearance of the implant fixture underneath the graft material (dichotomous assessment) was the only investigated outcome.

Data analysis

Descriptive data on first- and second-order grayscale texture outcomes were computed and presented as means ± standard deviations (SDs) in gray levels (GLs), while descriptive data on mean graft thickness were assessed on DICOM files using Horos (Horos^TM, v. 4.0.0, Horos Project) and reported as means ± SDs in millimeters (mm). The appearance of the implant fixture underneath the graft material was reported as a percentage. One-way analysis of variance (ANOVA) was used to detect differences among Cs, CMs, CCMs, MCCMs, PADMs, HADMs and IMPs in terms of first- and second-order grayscale outcome measures. Bonferroni post-hoc pairwise comparisons were utilized to assess the source of the significant difference revealed by the ANOVA. A p-value of 0.05 was set for statistical significance. All analyses were performed using a specified software (RStudio, v. 1.3.959). Plots were generated to visualize the results and based on the analysis, using Prism (v. 9.4.1, GraphPad Software, San Diego, CA) and ImageJ (National Institutes of Health, Bethesda, MD).

Results

Ultrasonographic characterization of the biomaterials

The group C presented with a relatively uniform hypoechoic band compared to the graft substitutes. CM displayed a thin band-like superficial hyperechoic layer and an overall uniform echogenicity, while CCM exhibited a more heterogeneous echointensity. MCCM presented three layers characterized by slightly different echogenicity, while PADM exhibited one hypoechoic band delimitated by two hyperechoic layers. HADM appeared with a relatively homogenous echogenicity. It was possible to appreciate the threads of IMP, that appeared as regular thin hyperechoic bands running perpendicular to the long axis of the implant. The apex of the implant and the coronal portion of the fixture without threads (coronal microgrooves) could also be visualized (Figure 1 and Figure 2).

Figure 2.

Clinical, and ultrasonographic presentation of the tested biomaterials and their related interactive three-dimensional surface plots depicting their corresponding mean EI. C, collagen tape dressing; CCM, cross-linked collagen matrix; CM, collagen matrix; HADM, human-derived acellular matrix; IMP, dental implant; MCCM, multilayered cross-linked collagen matrix; PADM, porcine-derived acellular dermal matrix.

The mean thickness of CCM was 3.12 ± 0.49 mm, while for the lowest thickness was observed for C group (0.68 ± 0.09 mm). The mean thickness of CM, HADM, PADM and MCCM was 2.11 ± 0.29 mm, 1.45 ± 0.08 mm, 1.23 ± 0.18 mm and 0.95 ± 0.24 mm, respectively.

First-order grayscale texture outcomes

The mean EI of C was 69.96 GL, while IMP showed the highest EI (154.37 GL). Among the soft tissue grafting materials, CM revealed the greatest EI (140.37 GL), and CCM exhibited the lowest EI (124.15 GL). Results of ANOVA showed a statistically significant difference among the groups for EI (p < .001). The Bonferroni test revealed a highly statistically significant difference between all the groups vs C (p < .001). There were no statistically significant differences among the soft tissue graft substitutes (CM, CCM, MCCM, HADM, and PADM) in terms of mean EI, while IMP showed a significant superior mean EI than CCM, MCCM, HADM, and PADM (Table 1 and Figure 3). Figure 2 illustrates the interactive three-dimensional surface plots representing the mean EI of the biomaterials.

Table 1.

First-order grayscale texture outcomes

Outcome	C (N = 10)	CM (N = 10)	CCM (N = 10)	MCCM (N = 10)	HADM (N = 10)	PADM (N = 10)	IMP (N = 10)
Mean EI (mean ± SD) (Gray levels)	69.96 ± 8.14 d	140.37 ± 6.97 a	124.15 ± 19.74 a ,d	130.32 ± 17.78 a ,e	127.3 ± 17.23 a ,d	134.34 ± 12.56 a ,f	154.37 ± 7.10 a
SD (mean ± SD) (Gray levels)	20.77 ± 8.98	30.57 ± 5.86 b	28.24 ± 3.77	31.50 ± 4.26 b,f	21.46 ± 4.38 c	32.52 ± 6.75 a ,e	22.65 ± 2.78
Kurtosis (mean ± SD) (Gray levels)	0.48 ± 1.35	−0.66 ± 0.43 b	−0.38 ± 0.33	−0.48 ± 0.38 c	−0.43 ± 0.28 c	−0.34 ± 0.31	−0.17 ± 0.50
Skewness (mean ± SD) (Gray levels)	0.27 ± 0.45	−0.47 ± 0.09 a	−0.34 ± 0.33 b	−0.11 ± 0.47	−0.11 ± 0.24	0.26 ± 0.29	0.09 ± 0.20

C: collagen tape dressing. CCM: cross-linked collagen matrix. CM: collagen matrix. EI: echo intensity. HADM: human-derived acellular dermal matrix. IMP: dental implant. MCCM: multilayered cross-linked collagen matrix. PADM: porcine-derived acellular dermal matrix. SD: standard deviation.

^adenotes statistically significant difference compared to C (p < .001).

^bdenotes statistically significant difference compared to C (p < .01).

^cdenotes statistically significant difference compared to C (p < .05).

^ddenotes statistically significant difference compared to IMP (p < .001).

^edenotes statistically significant difference compared to IMP (p < .01).

^fdenotes statistically significant difference compared to IMP (p < .05).

Figure 3.

Results of the computed mean EI of the tested biomaterials, depicted in box plots (a), and presented in cross-tabulation format for relative assessment (b). Note that in figure a, asterisks (*) on top of a horizontal line connecting two groups, implies a statistically significant difference between the two, the magnitude of which can be read within the crosstab presented below (Figure b) along with its 95% confidence intervals and an asterisks (*) to show statistical significance. * denotes p < .05, ** denotes p < .01 and *** denotes p < .001. C; collagen tape dressing; CCM, cross-linked collagen matrix; CM, collagen matrix; EI, echo intensity; MCCM, multilayered cross-linked collagen matrix; HADM, human-derived acellular dermal matrix; IMP, dental implant; MCCM, multilayered cross-linked collagen matrix; PADM, porcine-derived acellular dermal matrix.

Results of ANOVA indicated a statistically significant differences among the groups in terms of SD (p < 0.001), kurtosis (p < 0.01) and skewness (p < 0.001). The SD of CM, MCCM and PADM were found to be significantly higher than the SD of HADM (p < .05, p < 0.01, and p < 0.001, respectively). PADM showed a significantly greater skewness than CCM and CM (p < 0.01 and p < 0.001, respectively). SD, kurtosis, and skewness are depicted in detail in Table 1 and Figure 4.

Figure 4.

Plots depicting the outcomes of kurtosis (a), skewness (b), contrast (c) and homogeneity (d) of the different groups. Asterisks (*) indicate statistically significant differences between the connected groups. C; collagen tape dressing; CCM, cross-linked collagen matrix; CM, collagen matrix; EI, echo intensity; MCCM, multilayered cross-linked collagen matrix; HADM, human-derived acellular dermal matrix; IMP, dental implant; MCCM, multilayered cross-linked collagen matrix; PADM, porcine-derived acellular dermal matrix.

Secondary grayscale analysis-derived outcomes

ANOVA showed that there were no significant differences among the groups for energy and correlation (Table 2). A statistically significant difference among the biomaterials was found in terms of entropy (p < 0.01), contrast (p < .001) and homogeneity (p < .001) (Figure 4). Bonferroni test demonstrated that the entropy of HADM was significantly lower than the entropy of C (p < .05) and IMP (p < .001). IMP showed the highest mean contrast (9.87 GL), that was statistically significantly higher than C (5.88 GL, p < .001), HADM (6.43 GL, p < .001), PADM (6.78 GL, p < .001), CCM (7.47 GL, p < .01), and CM (7.79 GL, p < 0.05). MCCM exhibited superior contrast than HADM (9.25 vs 6.43 GL, p < .001) and PADM (9.25 vs 6.78 GL, p < .01). The contrast in group C was significantly inferior to the contrast of the other groups (Table 2). In terms of homogeneity, the highest mean value was observed in group C, while the lowest mean homogeneity was obtained for IMP. Bonferroni test revealed that group C was associated with a significantly higher homogeneity than CM (0.43 vs 0.37 GL, p < .05) and IMP (0.43 vs 0.34 GL, p < .001). The homogeneity of IMP was significantly inferior to the one observed for HADM (0.42 vs 0.34 GL, p < .001), PADM (0.41 vs 0.34 GL, p < .01), and CCM (0.40 vs 0.34 GL, p < .05). HADM exhibited superior homogeneity compared to CM (0.42 vs 0.37 GL, p < .05) (Table 2 and Figure 4).

Table 2.

Second-order grayscale texture outcomes

Outcome	C (N = 10)	CM (N = 10)	CCM (N = 10)	MCCM (N = 10)	HADM (N = 10)	PADM (N = 10)	IMP (N = 10)
Entropy (mean ± SD) (Gray levels)	7.38 ± 0.12	7.35 ± 0.15	7.15 ± 0.34	7.24 ± 0.46	6.97 ± 0.23 b,c	7.33 ± 0.28	7.52 ± 0.22
Contrast (mean ± SD) (Gray levels)	5.88 ± 0.93 c	7.79 ± 0.93 b,e	7.47 ± 1.24 d	9.25 ± 1.64 a	6.43 ± 1.36 c	6.78 ± 1.02 c	9.87 ± 1.78 a
Correlation (mean ± SD) (Gray levels)	0.33 ± 0.08	0.48 ± 0.22	0.46 ± 0.27	0.41 ± 0.14	0.49 ± 0.25	0.32 ± 0.19	0.32 ± 0.24
Energy (mean ± SD) (Gray levels)	0.55 ± 0.39	0.50 ± 0.44	0.26 ± 0.42	0.30 ± 0.39	0.18 ± 0.38	0.28 ± 0.37	0.44 ± 0.39
Homogeneity (mean ± SD) (Gray levels)	0.43 ± 0.03 c	0.37 ± 0.01 b	0.40 ± 0.05 e	0.39 ± 0.06	0.42 ± 0.03 c	0.41 ± 0.04 d	0.34 ± 0.02 a

C: collagen tape dressing. CCM: cross-linked collagen matrix. CM: collagen matrix. EI: echo intensity. HADM: human-derived acellular dermal matrix. IMP: dental implant. MCCM: multilayered cross-linked collagen matrix. PADM: porcine-derived acellular dermal matrix. SD: standard deviation.

^adenotes statistically significant difference compared to C (p < .001).

^bdenotes statistically significant difference compared to C (p < .05).

^cdenotes statistically significant difference compared to IMP (p < .001).

^ddenotes statistically significant difference compared to IMP (p < .01).

^e denotes statistically significant difference compared to IMP (p < .05).

The effect of graft substitutes on the appearance of the implant fixture underneath

When C was positioned over IMP, the threads and the fixture underneath were visible in six cases (60 %). In nine cases (90%) of MCCM positioned over IMP, the implant body could be visualized below the graft, while in seven cases (70%) with HADM and in one case (10%) with CM the implant fixture underneath was visible. With the other graft substitutes, the area of IMP below the graft could not be seen (Figure 1).

Discussion

Soft tissue graft substitutes have become popular among clinicians to replace—when possible—autogenous grafts, providing patients overall a less invasive treatment. ^6,22–25 Although several soft tissue graft alternatives have been introduced in the past decades, few clinical studies comparing head-to-head different graft substitutes are available. ^26–29 Evidence from indirect comparisons revealed that soft tissue graft alternatives have heterogenous outcomes. ^26–29 It is reasonable to assume that the unique structure and mechanical characteristics of each graft can explain their different behaviors observed in the clinical setting. ^1–3,23 The present study described the application of HFUS to characterize the structural properties of different allogeneic and xenogeneic soft tissue graft substitutes. Among the main advantages of HFUS compared to traditional methods for assessing the structure of biomaterials, the non-invasiveness of the procedure and the possibility of repeating the examination in vivo at different time points following the surgical intervention should be highlighted. ^4,5

Overall, we observed a uniform ultrasonographic presentation of samples from the same group, while the ultrasonographic appearance of the biomaterials substantially varied among the different groups. CM presented a thin hyperechoic layer compared to the other region of the graft, MCCM showed multiple band-like layers of different echogenicity, and PADM exhibited two hyperechoic layers between a thicker hypoechoic region. C, CCM and HADM showed a more homogenous presentation in terms of echogenicity, while IMP displayed clear demarcations between the collar and the part of the fixture with the threads. These findings are in line with the physical features of these biomaterials described with conventional methods. ^23,30–34

An interesting observation from our study is related to the ultrasonographic images of IMPs when different biomaterials are positioned on top of them, as commonly performed in daily practice for soft tissue augmentation of dental implants with buccal bone dehiscence. The implant fixture could be observed in 90%, 70%, 60% and 10% of cases when MCCMs, HADMs, Cs and CMs were positioned over IMPs, respectively. The other graft substitutes did not allow to visualize underneath dental implants, suggesting already substantial differences in the physical and mechanical properties among the examined biomaterials. This observation is also relevant for clinical practice. As the reliability of ultrasound in visualizing buccal bone and implant components has been proven, ^8,9,35–37 scans obtained at early time points after soft tissue augmentation with CCM, CM, HADM, and PADM may not reveal the implant fixture underneath. Clinical studies from our group described the ultrasonographic features of the process of integration and remodeling of autogenous soft tissue grafts and substitutes over different stages of healing. The augmented soft tissue did not impede the visualization of dental implants and natural teeth at 3, 6 and 12 months. ^7,38,39

We utilized grayscale analysis to further investigate structural differences among commonly used graft substitutes. This methodology has been extensively performed for assessing muscle quality, damage, and strength, ^14,40,41 and for differentiating pathological conditions in several medical fields. ^42–50 The rationale of performing first- and second-order grayscale analysis is the possibility of quantifying the texture of an image beyond what is perceived by the human eye, which can detect approximately 80 levels of gray. ^{17,20,47,51,52} First-order grayscale outcome measures derived from histograms revealed a highly statistically significant difference among the groups in terms of EI, with IMP showing the highest mean EI (154.37 GL), and C the lowest mean EI (69.96 GL). No statistically significant differences were found among the soft tissue graft substitutes for mean EI, while IMP obtained a significantly higher mean EI than the other biomaterials. Differences among the biomaterials overall—and among the different soft tissue graft substitutes—were also observed for kurtosis, skewness, and standard deviation. The clinical translation of these findings is open to speculation. Quantitative ultrasonographic parameters have been utilized in tissue engineering to determine the mineral content of scaffolds, to monitor the osteoblastic differentiation in engineered tissues, and to detect differences in collagen fiber microstructure. ^4,5,53,54 We can assume that mean EI, kurtosis, skewness, and standard deviation also reflect the different composition of the graft substitutes (e.g. collagen types, cross-linking of the collagen, dermal matrix, etc.), their collagen fibers dimension, spatial organization, and density, and overall their physical and mechanical properties. ^4,5,53,54 Clinicians are now encouraged to use the described method and related outcomes to analyze the in vivo wound healing and integration process of allogeneic and xenogeneic soft tissue graft substitutes at different time points.

The present study also applied secondary grayscale texture outcomes to further investigate the structural properties of the biomaterials. These outcomes are based on the construction of GLCMs that are able to compute and report data on the spatial relationship of the pixels. ^{15–17,20,21} The strength of this method is the objective assessment of ultrasonographic textural features that are not dependent on the experience of the operator, the dimension and size of the region of interest, nor the utilized ultrasound system. ^16,55,56 We observed a highly statistically significant difference among the groups for contrast and homogeneity. IMP group exhibited the greatest contrast and the lowest homogeneity, while the group C showed the lowest contrast and the highest homogeneity. MCCM exhibited superior contrast than HADM and PADM, indicating higher local variations within the MCCM compared to the two dermal matrices. The highest homogeneity observed in the C group may be explained by the uniform composition of the dressing material, while the other graft substitutes are designed with specific structures and different layers, therefore their ultrasound appearance is less uniform. IMPs were associated with the lowest homogeneity, which is probably due to the morphology and macro design of the implant fixture.

When interpreting our findings, it should be highlighted that, in the present study, HFUS scans were taken in the most ideal scenario, while additional factors (e.g. limited patient opening, reduced vestibular depth, poor patient compliance, patient movement, etc.) may challenge the quality of the scanning of oral tissues in the clinical setting. Limitations of the present study include its design (in vitro) and the relative limited sample size that may have prevented finding further significant differences among the groups for the outcomes of interested. Nevertheless, to the best of our knowledge, this is the first study investigating structural properties of allogeneic and xenogeneic soft tissue grafts - commonly utilized around teeth and dental implants - using ultrasonographic echo intensity and GLCM analysis. Our findings may represent the first step to encourage this type of assessment in clinical trials.

Conclusions

Within its limitations, the present study demonstrated the application of HFUS grayscale analysis to characterize the structure of different biomaterials. By evaluating ultrasonographic first- and second-order gray texture outcomes, it is possible to distinguish the properties of different soft tissue graft substitutes. This method holds potential for translation to in-vivo evaluation of wound healing and dimensional changes following soft tissue augmentation.