A comparison between commercially available artificial intelligence-based and conventional human expert-based digital...

Area of Science:

• Digital dentistry
• Artificial intelligence in prosthodontics
• CAD/CAM in restorative dentistry

Background:

• Limited studies compare deep learning (DL)-based digital workflows with conventional methods for dental design quality and consistency.
• Evaluating DL software's potential in ensuring baseline design quality is crucial.

Purpose of the Study:

• To assess if commercially available DL-based design solutions can yield outcomes comparable to those of a dental laboratory technician.
• Compare DL-driven digital workflows against conventional human-based methods in dental prosthetics.

Main Methods:

• Three digital workflows were used to design 25 anterior complete crowns: two DL-based (DD, DA) and one conventional (CC).
• Outcomes were evaluated for function (incisal path deviations, length, mean inclination) and esthetics (width-to-height ratio, contact length, tooth axis angle).
• Noninferiority analysis was conducted using Welch t and Mann-Whitney U tests (α=.05).

Main Results:

• DL-based workflows (DD, DA) demonstrated noninferiority to the conventional workflow (CC) for incisal path length and width-to-height ratio.
• DD crowns showed noninferiority for mean incisal path inclination and angular radius.
• Both DD and DA workflows were noninferior for proximal contact length and tooth axis angle compared to CC.

Conclusions:

• Digital workflows utilizing DL software (DD, DA) produce anterior crowns with functional and esthetic outcomes similar to conventional laboratory technician designs.
• DL-based software offers a viable alternative for consistent and high-quality anterior crown fabrication.