Related Experiment Video
Updated: Jun 26, 2026
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
1.8K
An Artificial Intelligence model for implant segmentation on periapical radiographs
Niha Adnan1, Muhammad Hanif2, Khurram Khan2
1Department of Surgery, Aga Khan University Hospital.
Summary
A deep learning (DL) algorithm effectively segmented dental implants on PA radiographs, achieving high accuracy comparable to human annotators. This demonstrates DL
Area of Science:
- Radiology
- Artificial Intelligence
- Biomedical Imaging
Background:
- Dental implant placement requires precise visualization.
- Radiographic interpretation is crucial for assessment.
- Manual segmentation of dental implants can be time-consuming.
Purpose of the Study:
- To develop and evaluate a Deep Learning (DL) algorithm for segmenting dental implants on PA radiographs.
- To compare the DL algorithm's performance against human-annotated ground truth.
Main Methods:
- A dataset of 300 PA radiographs was annotated for implants and teeth.
- Data augmentation increased the dataset to 1294 images for training.
- A U-net architecture was trained for implant segmentation.
Main Results:
- The DL algorithm achieved 93.8% accuracy, 90% precision, and 83% recall.
- The F-1 score was 86%, with an Intersection over Union of 86.4%.
- Performance was evaluated on 130 unseen images.
Conclusions:
- The DL algorithm demonstrated high performance in segmenting dental implants on PA radiographs.
- The algorithm's accuracy is comparable to human expert performance.
- This technology holds promise for improving radiographic analysis.
