Development and evaluation of a deep learning segmentation model for assessing non-surgical endodontic treatment outcomes on periapical radiographs: A retrospective study
2024

Deep Learning Model for Predicting Endodontic Treatment Outcomes

Sample size: 1200 publication Evidence: high

Author Information

Author(s): Dennis, Suebnukarn Siriwan, Vicharueang Sothana, Limprasert Wasit

Primary Institution: Thammasat University

Hypothesis

The integration of the Mask R-CNN model with clinicians would improve the accuracy of predicting endodontic treatment outcomes on preoperative periapical radiographs compared to predictions made by clinicians alone.

Conclusion

The deep learning-based Mask R-CNN model demonstrated high performance in classifying endodontic treatment outcomes and improved clinician accuracy when used as a decision-support tool.

Supporting Evidence

  • The Mask R-CNN model achieved a mean average precision (mAP) of 0.88.
  • Clinician performance improved significantly with the help of the Mask R-CNN model.
  • The model's predictions were evaluated against a test set of 120 images.

Takeaway

This study created a smart computer program that helps dentists predict how well root canal treatments will work by looking at X-ray images of teeth.

Methodology

The study used a retrospective design to develop and evaluate a deep learning model based on preoperative periapical radiographic images, employing a Mask R-CNN segmentation algorithm.

Potential Biases

Potential biases may arise from the retrospective nature of the study and the limited dataset from a single institution.

Limitations

The study was limited to retrospective data from a single hospital and did not include other important preoperative patient history.

Participant Demographics

The sample included 1200 cases, with 46.3% male and 53.7% female, primarily aged 25-64 years.

Statistical Information

Confidence Interval

95% CI 0.83–0.93

Digital Object Identifier (DOI)

10.1371/journal.pone.0310925

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication