Novel method of identifying unknown dental implants via 3D scan data matching: Proof-of-concept study

To evaluate whether 3D intaglio-surface comparisons of elastomeric impressions can discriminate matching from non-matching implant connection families using root mean square error (RMSE) as an objective metric.

A dataset of 25 unique implants was impressed with a scannable light body polyvinylsiloxane and digitized using a lab scanner (IdenticaT500, Medit, Seoul, Korea). Impressions captured implant platforms and internal anatomic details, and cross-comparisons using Geomagic Control X software generated 325 RMSE values. Bootstrap validation (SAS/STAT) tested whether values differed significantly for like and unlike implant groups. To validate the model, an impression was made of an implant in situ, and RMSE values were calculated using the original dataset.

RMSE values ranged from 0 to 0.2049. Across 325 pairwise comparisons, intra‑group RMSE values were substantially lower than inter‑group values (means ∼0.037 vs. 0.129). ROC analysis supported an RMSE threshold of 0.0866 for discriminating between matches and non‑matches. The area under the ROC curve was 0.9761, indicating near-perfect discrimination between like and unlike implant groups. An RMSE of 0.0374 in the matched in situ case further validated the model.

RMSE values ≤ 0.0866 were statistically significant in predicting accurate implant identification. RMSE analysis offers an objective method to distinguish implant connection types and may complement radiographic assessment when prior records or optimal radiographic views are unavailable. Further work is needed to expand reference libraries for fixture‑level identification. This commonly encountered scenario has many clinical, professional, and ethical implications for routine maintenance and treatment of complications in implant dentistry.