IADR Abstract Archives

Morphological and Functional Evaluation of AI-Generated and CAD-Designed Crowns

Objectives: This study aimed to compare occlusal morphology and fracture resistance of single crown restorations generated by knowledge-based artificial intelligence (AI) system and those designed by an experienced dental technician using Computer-Aided Design (CAD) software.
Methods: 12 human dental casts were digitized and then 3D printed into resin models where teeth #45 were prepared, and the tooth preparations were scanned. Crown designs were generated using the CEREC biogeneric individual function (group BI) and accomplished by an experienced technician using the CAD software (Zfx Manager 2.0) (group TD), respectively. The original tooth morphology and crown designs were superimposed (Geomagic Control 14.0). Occlusal morphological parameters, including average positive and negative profile discrepancy, standard deviations (SD), estimated root mean square (RMSestimate), volume discrepancy, profile discrepancy using volume/area and z-difference method, and cusp angle, were analyzed. Monolithic lithium disilicate crowns (IPS e.max CAD) were milled, sintered, and adhesively luted to 3D-printed dies. Then load-to-fracture test was applied (Instron universal testing machine, crosshead speed: 0.5mm/min) to determine the fracture resistance. The failure mode of the specimens was recorded and examined under microscopy and SEM. Paired t-test, repeated measurements of ANOVA, Fisher’s exact test, and Pearson’s correlation were used in statistical analysis (α=0.05).
Results: AI-generated crowns presented significantly higher average positive profile discrepancy, SD, RMSestimate, z-difference and volume/area profile discrepancy than CAD-designed crowns (p<0.05). The cusp angle values of group BI and TD were both significantly higher than the original teeth (p<0.05), but there was no significant difference between two groups. No significant difference was found in fracture loads, while group BI has a significantly higher percentage of restorable substrate damage (p<0.05).
Conclusions: Discrepancies in occlusal morphology exist between AI-generated and CAD-designed crowns, as the latter showed higher similarities to the original teeth. Nevertheless, both AI-generated and technician’s CAD-designed crowns can achieve clinically acceptable fracture resistance.

2021 South East Asian Division Meeting (Hong Kong)
Hong Kong
2021
119
Prosthodontics Research
  • Chen, Yanning  ( The University of Hong Kong , Hong Kong , Hong Kong )
  • Pow, Edmond  ( The University of Hong Kong , Hong Kong , Hong Kong SAR , Hong Kong )
  • Tsoi, James Kit Hon  ( The University of Hong Kong , Hong Kong , Hong Kong )
  • NONE
    GRF (1720220)
    Oral Session
    AI in dentistry and diagnostic science
    Thursday, 12/09/2021 , 02:00PM - 03:30PM