INTRODUCTION
Alveolar cleft is a discontinuity of the maxillary alveolar process commonly associated with cleft lip and palate, leading to significant functional morbidity, including inadequate bony support for teeth, malocclusion, and oronasal fistulae.1–4 Secondary alveolar bone grafting (SABG) is the standard procedure to restore alveolar arch continuity, provide a bony matrix for tooth eruption and orthodontic movement, and eliminate oronasal communication.5–9 The success of SABG directly influences dental rehabilitation, orthodontic stability, and long-term oral function.
Historically, graft outcome assessment relied on 2D radiographic criteria such as the Bergland classification, which is subjective and lacks volumetric data.10 The advent of 3D computed tomography (CT) and cone-beam computed tomography (CBCT) has enabled more objective, quantitative evaluation.11–16 Contemporary protocols often use image segmentation, 3D model reconstruction, and Boolean operations to calculate newly formed bone ratio (NFBR), defined as the volume of newly formed bone (NFB) divided by the initial defect (ID) volume.12,17–21 Studies by Di et al22 and Teramoto et al23 have reported high reliability, establishing 3D volumetric analysis as a modern standard.
Despite excellent reproducibility, conventional 3D techniques have critical limitations. First, they focus exclusively on bone quantity (volume) and neglect bone quality as reflected by mineralization density. A graft with substantial volume but poor mineralization may not provide adequate functional support.24 Second, a systematic error arises from the mismatch in volumetric definition: the preoperative cleft is considered a solid defect without voids, whereas NFB has a porous, cancellous architecture.
Standard threshold-based segmentation captures only mineralized trabecular bone, while ignoring marrow and vascular spaces. This “solid-void discrepancy” systematically underestimates true NFBR and may misrepresent surgical success. To address these gaps, we developed an improved 3D evaluation protocol integrating NFB density assessment with a volumetric correction for graft porosity, enabling simultaneous evaluation of bone quantity and quality.
METHODS
Patients and Imaging
This retrospective study analyzed CT data from 37 patients who underwent alveolar bone grafting at our institution between January 2023 and December 2024. The study protocol was approved by the ethics committee of the Plastic Surgery Hospital, Chinese Academy of Medical Sciences (approval ID: 2022(186)). Written informed consent was obtained from the guardians of all patients.
All identifying information was removed from the 3D models to ensure patient anonymity. Appropriate authorization for scientific publication of the imaging data was obtained. Patients aged 7–12 years (mean 9.92 ± 1.51 y; 21 males, 16 females) with unilateral alveolar clefts (with or without cleft palate) underwent SABG using autologous iliac cancellous bone granules.
Craniofacial CT scans were acquired preoperatively and at 6 months postoperatively using a Brilliance 64 scanner (Philips, the Netherlands) with the following parameters: 120 kV, 200 mA, 1-mm slice thickness, and 0.5-mm interval. Preoperative defect volume and postoperative bone volume were independently measured by 2 observers (Deng Menghao and Wei Siqi), each performing duplicate measurements 2 weeks apart; the mean of 4 measurements was used for analysis. Both observers had more than 2 years of experience in craniofacial 3D image analysis using Mimics software.
All evaluators were blinded to patient identifiers. Scans followed the ALARA (as low as reasonably achievable) principle.
Conventional Volumetric Assessment Protocol
The conventional method (control) followed protocols by Chen et al,25 Li et al,26 and Di et al22:
Video 1.This video demonstrates the conventional Boolean subtraction method for volumetric assessment of alveolar bone grafts. It shows preoperative 3D model reconstruction, defect delineation using multiple slice editing, registration of postoperative CT to the preoperative model, and isolation of newly formed bone (NFB). The final step calculates the newly formed bone ratio (NFBR) as (NFB volume / initial defect volume) × 100%.
This method follows established protocols but does not correct for trabecular porosity.GOX.0000000000008191video1.mp4Kaltura
Bone Density Measurement of the Graft Region
The technique of measuring mean HU within a defined 3D region of interest to assess bone mineral density is well established in orthopedic and dental implant literature.27,28 This approach has also been previously applied in alveolar cleft repair to evaluate bone density in the recipient site adjacent to the defect.29 In the current study, we adapted this method to measure the density within the graft region itself, rather than that of the adjacent native bone. The following steps were performed:
Video 2.This video illustrates the method for measuring bone density (Hounsfield units, HU) of the newly formed bone after alveolar bone grafting. It shows co registration of the preoperative skull model with the postoperative CT dataset, erasure of the graft region within the boundaries defined by the preoperative model using the “Remove” function, and isolation of the graft mask. The mean HU value is then obtained from this mask, providing a quantitative measure of bone quality.GOX.0000000000008191video2.mp4Kaltura
Modified Volumetry Using Smart Fill
To correct for underestimation due to porosity, the Smart Fill function in Mimics was applied to the postoperative dataset. The procedure was as follows:
Video 3.This video presents the Smart Fill enhanced volumetric method that corrects for porosity underestimation in alveolar bone graft assessment. It demonstrates how the Smart Fill function in Mimics software fills internal voids within the postoperative bone mask, creating a contiguous geometry that represents the total spatial extent of regenerated tissue (mineralized plus unmineralized matrix). The modified newly formed bone ratio (NFBR SF) is then calculated using the same registration and Boolean subtraction steps as the conventional method.GOX.0000000000008191video3.mp4Kaltura
Data Analysis
Descriptive statistics (mean ± SD) were calculated. Paired comparisons between Smart Fill and conventional methods used a paired t test (P < 0.05). SPSS 26.0 (IBM Corp.) was used.
Intra- and interobserver reliability were assessed using the intraclass correlation coefficient (ICC).
RESULTS
Patient and Defect Characteristics
All 37 patients had successful graft placement. Two minor events (wound oozing and postoperative fever) resolved spontaneously without intervention. No infections, graft loss, wound dehiscence, or donor-site morbidity occurred.
All patients completed 6-month CT follow-up (100%). Preoperative defect volume averaged 1.04 ± 0.33 cm3 (range 0.43–1.81 cm3), with no sex difference. Bony bridge formation was observed in all patients.
Conventional Versus Smart Fill–enhanced NFBR
Conventional 3D volumetry yielded a mean NFBR of 44.87% ± 14.48%, consistent with previous reports (eg, Di et al22: 39.81% at 1 y). With Smart Fill–enhanced measurement, the mean NFBR was 55.85% ± 12.99% (absolute increase 10.98%, relative increase 24.47%; P < 0.001, paired t test) (Table 1). Improvement occurred in 33 of 37 (89.2%) patients; the remaining 4 had minimal voids and high graft density.
Figure 2 shows a representative case; Figure 3 illustrates the effect. The relative increase averaged 22.19%. Gains were largest in grafts with porous or undermineralized zones (eg, conventional 40.44% → Smart Fill 60.04%) and smaller in dense grafts (72.73% → 76.42%).
Bone Density Outcomes
Mean graft density at 6 months was 483.01 ± 78.11 HU (range 358.77–614.64 HU), substantially lower than cortical bone (~1000 HU) and consistent with trabecular bone. No linear correlation was found between density and NFBR (Pearson r = −0.16, P = 0.33). Volume and density are complementary dimensions: for example, 1 patient had NFBR 75.43% with a density of 362.64 HU, where another had an NFBR of 34.02% with a density of 611.59 HU.
Reliability of the Smart Fill–enhanced Volume Measurement
Intraobserver ICC was 0.985 (95% confidence interval: 0.970–0.992); interobserver ICC was 0.965 (95% confidence interval: 0.940–0.981) (both P < 0.001) (Table 2), indicating excellent reproducibility.
DISCUSSION
This study introduces a dual-parameter postoperative assessment method for alveolar bone grafting, combining bone density evaluation with porosity-corrected volumetry. The results demonstrate a more comprehensive characterization of graft outcomes.
Dual-parameter 3D Assessment (Quality and Quantity)
Including bone density (HU) addresses a critical gap.30 Volumetric fill alone may overestimate success; a 75% filled but poorly mineralized graft may fail clinically. Mean graft density (483.01 HU) confirms predominantly cancellous bone at 6 months. The wide range highlights individual variability (graft size, blood supply, and patient factors).
Our method can facilitate future comparisons of graft materials or techniques.31–33 For example, Bianchi et al34 compared bone substitutes with autograft by volume (~69% versus 67%); adding HU analysis might reveal density differences not apparent from volume alone. In the present study, we extended this approach by measuring density specifically within the graft region, isolated via Boolean subtraction of coregistered preoperative and postoperative models, thereby avoiding contamination from adjacent native bone (see Bone Density Measurement of the Graft Region).
Smart Fill Method and True Regeneration Estimation
Smart Fill–enhanced volumetry increased NFBR by approximately 10 percentage points, compensating for the effect of incomplete mineralization and porosity.22,23 Smart Fill does not invent bone where none exists; it fills gaps within the demarcated graft site. If the cleft were empty, no volume would be added. However, partial volume effects in CT imaging can cause voxels containing mixed tissue (bone and marrow) to fall below the standard HU threshold, leading to systematic underestimation of trabecular bone volume.
Smart Fill partially corrects for this by filling small enclosed voids, but it cannot completely eliminate partial-volume artifacts. At 6 months, the graft site contains mineralized spicules, fibrous tissue, and marrow; Smart Fill approximates the total scaffold (mineralized + unmineralized matrix), which may correlate better with eventual ossification in the longer term than conventional threshold-based segmentation. Nevertheless, absolute NFBR values obtained with Smart Fill should be interpreted as estimates, and future validation against micro-CT or histology is warranted.
Biological Limitations of CT-based Assessment
CT measures radiodensity (HU), which reflects the degree of mineralization but does not directly assess cell viability, vascular perfusion, or the capacity for stress-induced remodeling. In trabecular bone, graft material in the center of the defect may be far from vascular support, and visible mineralization does not guarantee living osteocytes. Consequently, our dual-parameter method should be interpreted as an anatomical–radiographic evaluation rather than a direct measure of graft biological functionality.
Future studies incorporating histology, micro-CT, or perfusion imaging are needed to validate the biological correlates of our CT-based metrics.
Clinical Significance
Traditional success criteria (eg, Bergland) focus on bone height. Our 3D metrics add depth: a moderate-volume fill (eg, 50%) may suffice if the crestal region is filled, whereas a high–volume graft with low density may fail to support tooth eruption. Bone density can guide orthodontic timing: if HU is very low, delaying heavy forces or augmenting density before implant placement may be prudent.35,36 We propose that graft regions with mean HU less than 400 may warrant delayed orthodontic loading or additional augmentation before implant placement, whereas values greater than 500 suggest sufficient mineralization for routine orthodontic forces.
Conversely, a graft with moderate volume (eg, NFBR 40%–50%) but acceptable density may still provide adequate functional support if subsequent tooth eruption and orthodontic forces stimulate adaptive remodeling. Therefore, our proposed HU thresholds (400 and 500 HU) should be interpreted as guides rather than absolute cutoffs. Clinical decisions should integrate both imaging metrics and individual patient factors, including the expected timing of canine eruption and orthodontic loading.
Comparison With Other Methods
Erkert et al20 reported approximately 58% restoration at 6 months using mirroring; our Smart Fill result (55.85%) converges with that, suggesting similar conceptual capture of the “fraction of defect truly restored by bone.” Liu et al19 reported approximately 47.7% NFBR at 1 year, comparable to our conventional result (44.87%). Our method streamlines earlier protocols and adds the Smart Fill step. Zhang et al37 used machine-learning auto-filling for maxillary growth analysis, reflecting growing recognition of virtual defect filling.
Our method adds a novel density component and a simple filling step using existing software. Although we used Mimics software, the principles of porosity-corrected volumetry and density measurement can be implemented in other segmentation platforms that offer morphological closing functions and region-of-interest density sampling.
Limitations
Limitations include the retrospective design, modest sample size (n = 37), and lack of a prospective power calculation. Only autologous iliac cancellous grafts were used; comparisons with bone substitutes or bone morphogenetic protein–augmented grafts are lacking. Confounders (cleft width, soft-tissue closure quality, orthodontic status, and nutrition) were not controlled.
Furthermore, our postoperative CT scans were obtained at 6 months, which reflects early graft consolidation rather than definitive functional outcome; long-term studies correlating our imaging metrics with canine eruption and orthodontic tooth movement are needed. In particular, we cannot determine whether our proposed HU thresholds predict clinical success, as we did not correlate these metrics with functional outcomes. Moreover, the Smart Fill algorithm has not been validated against a gold standard (eg, micro-CT or histology).
Consequently, the absolute NFBR values reported should be considered estimates, and the accuracy of the method requires independent validation. Additionally, HU values serve as a surrogate for bone mineral density and do not directly quantify cellular or molecular properties of the graft (eg, stem cell concentration or angiogenic factors). Notably, the proposed HU thresholds are exploratory and have not been validated against functional outcomes; they should be regarded as reference values only, not as definitive cutoffs.
Therefore, the proposed biological mechanisms remain speculative and require histological or molecular validation. Postoperative CT for densitometry raises radiation and cost concerns; therefore, this protocol is not suitable yet for routine clinical use but rather for research or complex assessments. HU values from medical CT are not directly comparable to CBCT-based studies due to the lack of standardized calibration across CBCT devices.38 Relative comparisons within the same modality are preferable.
CONCLUSIONS
We developed an integrated 3D CT-based evaluation method that overcomes key limitations of conventional techniques by incorporating bone density and modified volumetry. Bone quality was quantified via mean HU, and the Mimics Smart Fill algorithm corrected for internal voids, increasing measured NFBR by approximately 10% compared with traditional volumetry. This dual-parameter framework provides a more accurate and clinically relevant assessment of alveolar bone grafts, supporting improved treatment planning and outcome evaluation in cleft care.
DISCLOSURES
The authors have no financial interest to declare in relation to the content of this article. This study was supported by the Beijing Municipal Science & Technology Commission (No. Z221100007422039), CAMS Innovation Fund for Medical Sciences (No.
2025-I2M-C&T-A-009), and China University Industry-University-Research Innovation Fund (New Generation Information Technology Innovation Project; no. 2024IT015).
PATIENT CONSENT
The ethics committee of the Plastic Surgery Hospital, Chinese Academy of Medical Sciences, has waived the requirement for informed consent because this was a retrospective study.
ETHICAL APPROVAL
This study was approved by the ethics committee of the Plastic Surgery Hospital, Chinese Academy of Medical Sciences (approval ID: 2022(186)).