Objective
The purpose of this study is to evaluate the effectiveness of a deep learning approach for automatically detecting pulp stones using panoramic imaging. Pulp stones are mineralized structures within the pulp cavity of teeth and can serve as important diagnostic indicators for dentists.
Dataset
The researchers utilized a comprehensive dataset consisting of 2409 panoramic radiography images. These images were collected from different patients. The presence or absence of pulp stones was manually labeled in these images.
Model and Results
A deep learning model based on the YOLOv5 architecture was trained. The following metrics were obtained on the test dataset:
- F1 score: 0.7892
- Sensitivity: 0.8026
- Precision: 0.7762
These results demonstrate that deep learning-based artificial intelligence algorithms are successful in detecting pulp stones in panoramic radiographs. It is believed that such algorithms can enhance the effectiveness and efficiency of clinical decision support systems for dentists.
