Deep learning-designed implant-supported posterior crowns: Assessing time efficiency, tooth morphology, emergence...

Area of Science:

• Digital dentistry
• Artificial intelligence in prosthodontics
• Implant-supported restorations

Background:

• Conventional computer-aided design (CAD) for implant-supported crowns (ISCs) relies on manual technician input.
• Emerging deep learning (DL) software offers potential for automating and optimizing dental restoration design.
• Evaluating the clinical efficacy and efficiency of DL-based ISC design is crucial for adoption.

Purpose of the Study:

• To compare the design efficiency and clinical outcomes of ISCs created using deep learning (DL) software versus conventional CAD software.
• To assess the impact of technician optimization on DL-generated ISC designs.

Main Methods:

• Retrospective analysis of 20 partially edentulous casts for ISC fabrication.
• ISCs were designed using three methods: DL with no modification (DB), DL with technician optimization (DM), and conventional CAD (NC).
• Comparative analysis of time efficiency, crown contour, occlusal/proximal contacts, and other geometric parameters using statistical methods.

Main Results:

• DL-based ISC design (DB) demonstrated significantly higher time efficiency compared to DM and NC groups.
• DM group showed reduced volume deviations compared to the DB group when compared to the NC group.
• DB and DM groups exhibited similar occlusal table area, cusp angles/heights, and proximal contact intensities to the NC group, except for occlusal contacts.

Conclusions:

• Deep learning-based methods offer significant time efficiency for designing posterior implant-supported crowns.
• DL designs achieve functional properties comparable to conventional methods, including occlusal table area and cusp morphology.
• Technician optimization of DL designs can further refine outcomes like crown contour and emergence profile.