OralViewer: 3D Demonstration of Dental Surgeries for Patient Education with Oral Cavity Reconstruction from a 2D Panoramic X-ray

Title of the Paper

OralViewer: 3D Demonstration of Dental Surgeries for Patient Education with Oral Cavity Reconstruction from a 2D Panoramic X-ray

Paper Information

• Conference: 26th International Conference on Intelligent User Interfaces (IUI ’21), April 14–17, 2021
• Authors: Yuan Liang et al.
• Affiliation: University of California, Los Angeles, etc.
• Keywords: Deep learning, 3D visualization, patient education, single-view 3D reconstruction, interactive user interface

Research Background and Problem

• Problems or Challenges:
  1. Current methods for dental surgery education (e.g., verbal explanations, hand-drawn diagrams, videos) often fail to effectively reduce patient anxiety or help them understand complex surgical procedures.
  2. The three-dimensional anatomy required for dental surgeries lacks widely accessible and cost-effective solutions for educational purposes, as most existing 3D modeling methods rely on expensive scans like CT.
• Significance: Helping patients understand surgical procedures can effectively reduce anxiety and improve treatment satisfaction and outcomes. Providing intuitive and patient-specific 3D surgical demonstrations can bridge the knowledge gap between patients and dentists.
• Research Motivation and Related Work:
  • Existing 3D surgical demonstration technologies have shown potential in other fields (e.g., cardiac surgery) but remain unexplored in dental surgery education.
  • Current 3D modeling technologies heavily rely on costly CT scans, making them difficult to adopt widely in dental clinics.

Solution

• Methods and Approach: OralViewer is an interactive application that allows dentists to demonstrate dental surgery steps using patient-specific 3D oral cavity models.
  1. Proposes a novel deep learning-based method to reconstruct patient-specific 3D dental structures from 2D panoramic dental X-rays.
  2. Registers the reconstructed dental model with predefined gum and jawbone templates to form a complete 3D oral cavity model.
  3. Simulates surgical steps in real-time using virtual dental tools.
• Innovations:
  1. Constructs a 3D oral cavity model from single-view 2D dental images (reducing cost and radiation exposure).
  2. Provides interactive virtual dental tools to simulate surgical steps, enhancing patient understanding of complex procedures.
• Implementation Steps:
  1. 3D Dental Reconstruction:
     • Uses a deep convolutional neural network to predict the 3D shape of teeth.
     • Segments and localizes tooth regions from X-rays, generating voxel-level 3D models via back-projection.
     • Extracts the dental arch curve using an unsupervised method from intraoral photographs.
  2. Complete Oral Cavity Modeling:
     • Combines predefined gum and jawbone templates with patient-specific dental models through registration and assembly to form a complete oral cavity structure.
  3. Surgical Demonstration Tools:
     • Implements six virtual dental tools, including a scalpel, drill, treatment head, syringe, curette, and artificial crown/implant, to simulate various steps of dental surgeries.

Research Outcomes

• Specific Results:
  1. Technical evaluation shows an average Intersection over Union (IoU) of 0.771 for 3D dental reconstruction.
  2. Patient studies indicate that OralViewer significantly improves patient understanding of surgical procedures compared to traditional educational methods.
  3. Expert evaluations validate the system's clinical effectiveness and suggest improvements in the operation of virtual tools for greater intuitiveness.
• Advantages Compared to Existing Methods:
  1. Compared to existing 3D modeling technologies reliant on CT scans, OralViewer reduces cost and radiation exposure.
  2. Enhances patient engagement and understanding through interactive simulation of surgical steps.
• Experimental or Evaluation Results:
  1. Technical evaluation revealed varying reconstruction accuracy across different tooth categories, with slightly lower performance for wisdom teeth due to sample diversity and limited training data.
  2. In patient studies, the group using OralViewer scored significantly higher in describing surgical steps compared to the control group.
  3. Experts acknowledged that the oral cavity model and simulation tools effectively improved patient education efficiency, though user experience in tool operation requires further refinement.
• Limitations and Future Directions:
  1. The model does not include root canal structures, limiting its effectiveness in demonstrating certain procedures (e.g., root canal treatments).
  2. The current gum model is relatively coarse and could be improved by incorporating soft tissue scan templates.
  3. Recommends adding more virtual tool modules to cover all dental surgery scenarios.
  4. Suggests enhancing the interactive experience of virtual tools, such as implementing more intuitive controls on touchscreen devices and dynamically adjusting tool appearances to reflect parameter changes.

References

The paper extensively cites literature from related fields, including dental radiology, deep learning, user interaction design, and CAD tools, referencing a total of 54 works to support its methods and arguments.