This study explored the precision of a cone-beam computed tomography (CBCT) image intelligent segmentation model based on a feature fusion mechanism in the quantitative assessment of oral hard tissue repair outcomes, as well as the bioactivity and antibacterial properties of zinc-based nanocomposite coatings. Clinical CBCT imaging data were collected to construct a deep learning segmentation model based on a feature fusion mechanism. The segmentation accuracy was evaluated using the Dice similarity coefficient, precision, and other indicators, and compared with traditional segmentation methods. Hydroxyapatite (HA), nano-fluoridated HA (FHA), and their zinc composite coatings Zn/HA, Zn/FHA were prepared. The microstructure and composition of the coatings were characterized using scanning electron microscopy (SEM), transmission electron microscopy (TEM), and X-ray diffraction (XRD). The antibacterial properties were assessed through inhibition experiments (Staphylococcus aureus, Escherichia coli, Porphyromonas gingivalis). The proliferation and osteogenic differentiation induction potential (key osteogenic genes Runx2, BMP2, ALP, OCN, and OPN) were evaluated using bone marrow mesenchymal stem cells (BMSCs) with the CCK-8 and real-time quantitative PCR methods. The degree of enamel surface mineralization was assessed using a simulated enamel remineralization model, and the antibacterial performance of the composite coating materials was evaluated using Staphylococcus aureus, Escherichia coli, and Porphyromonas gingivalis as model strains. The in vivo bone repair effect was assessed through the rabbit bone defect model (postoperative 8 weeks, 12 weeks), combined with the CBCT intelligent segmentation model to quantify the proportion of newly formed bone area. The Dice coefficients for the segmentation of enamel, alveolar bone, pulp, and dentin by the feature fusion model reached 91.83±0.04, 93.50±0.03, 93.90±0.05, and 95.27±0.03, respectively, which were significantly higher than those of traditional models (P < 0.05). The Zn/FHA composite coating exhibited a porous surface structure with uniformly distributed fluorine elements, the highest hardness (325±18 VHN), and higher BMSCs cell viability (vs. FHA) (P < 0.05); Runx2, BMP2, ALP, OCN, and OPN were noticeably elevated (vs. the control and HA) (P < 0.05). The antibacterial rates against Staphylococcus aureus, Escherichia coli, and Porphyromonas gingivalis were 92.5±2.8%, 78.6±3.5%, and 85.8±3.0%, respectively, which were significantly higher than those of the Zn/HA and HA groups (P < 0.05). In the rabbit bone defect model, the proportion of newly formed bone area by Zn/FHA noticeably increased from 8 weeks to 12 weeks postoperatively, and was significantly higher than that of other groups (P < 0.05). The CBCT intelligent segmentation model based on a feature fusion mechanism achieves an accuracy comparable to that of manual segmentation performed by experienced clinicians, while offering the advantages of automation, high consistency, and reproducibility. This provides an efficient and objective quantitative tool for assessing oral anatomical structures and the effectiveness of bone repair. The Zn/FHA nanocomposite coating exhibits excellent antibacterial properties, osteogenic activity, and bone defect repair capability. By integrating the Zn/FHA nanocomposite coating with this CBCT intelligent segmentation model, a synergistic “material-imaging-evaluation” system can be established, enabling precise, dynamic monitoring and quantitative assessment of the oral hard tissue repair process. This integration provides a novel approach for the systematic evaluation of clinical repair outcomes.
Fang et al. (Thu,) studied this question.