AI-driven crown generation: A comparative analysis of point cloud completion models for mandibular first molar...

Area of Science:

• Artificial Intelligence in Dentistry
• 3D Point Cloud Processing
• Prosthodontics and Implantology

Background:

• Developing anatomically accurate dental prostheses is crucial for successful implant surgery.
• Artificial intelligence (AI) offers potential for automating and improving the accuracy of dental crown generation.
• Evaluating existing AI models for specific clinical applications like prosthetically guided implant surgery is necessary.

Purpose of the Study:

• To adapt and evaluate three AI-driven point cloud completion models (PF-Net, PCN, PoinTr) for generating anatomically accurate dental crowns.
• To assess the geometric fidelity, clinical applicability, and computational efficiency of these AI models for mandibular first molar crowns.
• To advance prosthetically guided implant surgery planning through accurate virtual prosthesis generation.

Main Methods:

• A dataset of 120 intraoral scans was used, with partial dentition models as input and natural crowns as ground truth.
• Three AI models (PF-Net, PCN, PoinTr) were trained using Chamfer Distance loss.
• Models were evaluated on geometric accuracy (Chamfer Distance, Hausdorff Distance, RMSE, M.N.D.), clinical dimensions, and generation time.

Main Results:

• PoinTr demonstrated superior clinical efficacy and geometric accuracy, with the lowest mesiodistal diameter deviation (0.14 ± 0.33 mm).
• PoinTr achieved minimal surface errors (Chamfer Distance: 0.10 mm; Hausdorff Distance: 0.22 mm) and balanced accuracy with efficiency (5.52 seconds/crown).
• PF-Net was fastest (3.22 seconds/crown) but had unacceptable morphological deviations; PCN showed lower performance than PoinTr.

Conclusions:

• PoinTr generates dental crowns with optimal morphological fidelity and precise dimensional control for prosthetically guided implant surgery.
• The PoinTr-based approach enhances digital workflow efficiency and accuracy in implant planning and conventional fixed prosthodontics.
• This AI tool improves surgical planning, surgical guide design, and reduces technician dependency in restorative dentistry.