Reproducibility of Occlusal Contacts in CAD/CAM Splint Fabrication: Analysis Using Hausdorff Distance and Clinical Validation: A Cross‑Sectional Study

Abstract

BACKGROUND: Despite the widespread adoption of CAD/CAM technologies for occlusal splint fabrication, the degree of agreement between virtually calculated occlusal contacts and the actual clinical contact pattern remains insufficiently investigated. The accuracy of occlusal transfer may be influenced not only by digital design algorithms but also by CAM manufacturing stages, post‑processing procedures, and clinical adaptation of the appliance. Quantitative criteria for comprehensive evaluation of occlusal contact reproducibility throughout all stages of the digital workflow are still lacking.

AIM: To evaluate the accuracy of occlusal contact transfer in digitally designed occlusal splints by sequentially comparing virtual data with the geometry of the fabricated devices and clinical contacts in the oral cavity.

METHODS: A cross‑sectional comparative clinical and laboratory study was conducted, including 20 clinical cases. For each case, two occlusal splints were fabricated (Avantis 3D and Exocad), totaling 40 appliances. Original STL models were compared with STL models of the splints after milling and scanning, and virtual contact maps were compared with clinical occlusal contacts recorded using 8‑ and 40‑μm articulating film. Geometric deviations between CAD and CAM models were assessed using the Hausdorff distance, with independent verification in GOM Inspect. Localization and intensity of occlusal contacts were additionally analyzed according to functional zones of the dental arch. Statistical analysis was performed using the Pearson's chi‑square test and Cohen's kappa coefficient.

RESULTS: Virtual contact maps of Avantis 3D and Exocad coincided in 16 of 20 paired comparisons (80%). During clinical validation, the agreement between virtual and actual contacts was 70% for Avantis 3D and 85% for Exocad; no statistically significant differences were found between the systems (χ² = 1.80; p = 0.180), and the kappa coefficient was κ = 0.41. The most pronounced discrepancies in contact localization were observed in the anterior segment of the dental arch, whereas the highest reproducibility was maintained in the molar segments. Analysis of contact intensity showed that the lowest agreement indices were recorded in the canine‑premolar segments. According to Hausdorff analysis, when virtual and clinical contacts agreed in ≥80% of cases, maximum deviations in the occlusal zone did not exceed 50–70 μm, whereas in cases of clinical discrepancies, the 95th percentile was 132 ± 24 μm, exceeding the threshold value of 100 μm.

CONCLUSION: The accuracy of occlusal contact transfer in digitally fabricated splints is determined by the combined influence of CAD design, CAM manufacturing, and clinical implementation stages. The proposed comprehensive approach, integrating clinical validation, zonal contact analysis, and Hausdorff‑based metric assessment, enables objective quality control of the digital protocol and identification of error accumulation stages. Study limitations include the relatively small sample size and single‑center design, warranting further validation in larger clinical cohorts.