Deep learning for automatic segmentation of bone sequestra on orthopantomographs: a UNet-Based approach

NCT ID NCT07809048

What the study statuses mean

This study's is highlighted.

Recruitment status, easiest to join first

Recruiting now
This trial is taking on new participants right now.
Not yet recruiting
Registered, but not yet taking participants.
By invitation only
Not open to general applications. Only people the study team invites can take part.
Paused
Paused for now. It may or may not start again.
Ongoing
Running, but no longer taking on new participants.
Completed This study
The trial has finished. Results may not be published yet.
Stopped early
Stopped early, before it reached the end. That can be for many reasons, including safety.
Cancelled
Cancelled before anyone took part.

Expanded access (not trials)

Expanded access
Not a trial. This treatment can be requested outside a study, case by case, for people who qualify.
Expanded access (paused)
Not a trial. The treatment can normally be requested outside a study, but is unavailable right now.
Expanded access (ended)
Not a trial. The treatment could once be requested outside a study, but no longer can.
Approved
The treatment has been approved, so it is available normally rather than through this programme.

When the status isn't known

Details not published
The full record has not been published yet, so there is little to show here.
Status unknown
This status has not been confirmed recently, so it may be out of date.

First seen Sep 09, 2026 · Last updated Sep 09, 2026

Summary

This study developed and evaluated a UNet-based deep learning model for automatic segmentation of bone sequestra on orthopantomographs (OPGs), with the goal of improving diagnostic efficiency and reducing inter-observer variability in osteomyelitis, osteoradionecrosis, and medication-related osteonecrosis of the jaw. A total of 120 anonymized OPGs with 134 annotated sequestra were used for training. Images were preprocessed with min-max normalization and augmented to enhance robustness. Manual expert annotations served as ground truth. A UNet architecture, optimized with Dice loss and the AdamW optimizer, was trained for 300 epochs. Performance was assessed using Dice coefficient, Intersection over Union (IoU), precision, recall, F1-score, and accuracy. ROC analysis and confusion matrix evaluations were performed. Agreement with clinicians was quantified using the Intraclass Correlation Coefficient (ICC). The model achieved a best-checkpoint validation Dice of 0.79 and IoU of 0.93. On the test set, performance included a Dice of 0.79, IoU of 0.74, recall of 0.81, precision of 0.81, and F1-score of 0.81. ROC analysis showed balanced discriminative performance with an AUC of 0.76. The confusion matrix indicated strong lesion/background classification, with minor under-segmentation. ICC analysis showed excellent agreement with an experienced surgeon (ICC = 0.85), though lower with a less experienced dentist (ICC = 0.57). The proposed UNet-based model enables efficient segmentation of sequestra, reducing annotation time by \>90% while maintaining diagnostic accuracy. The model highlights the clinical utility of AI-assisted decision support in maxillofacial radiology.

This is an AI summary of the original study and may miss details. Read our disclaimer.

Study facts

What this study's own registry entry says, in plain language.

Participants

120 people

The number who actually took part.

Started

Mar 2024

Finished

Mar 2025

Lead sponsor

Other sponsor

The registry's catch-all category, for sponsors it does not file as a company, a government agency, or a research network.

Who can take part

This study's own entry requirements. Only the study team can say for certain whether you qualify.

Who is studied

Patients will be selected among the patients with osteomyelitis with sequestrum

Ages

18 to 99 years

Sex

Anyone

Healthy volunteers

Not accepted

This study is not open to healthy volunteers. The entry requirements below say who it is open to.

Show the full entry requirements

Copied word for word from the study's registry entry, so the wording is the study team's rather than ours.

Inclusion Criteria: * Patient with exposed bone over 8 weeks with a history of bisphosphonates or antiangiogenic drugs, * Patient with a history of previous radiotherapy, * Patients with necrotic exposed bone with or without a history of trauma * Patients with osteomyelitis. Exclusion Criteria: * Patients with no clear vision of the borders of the sequestra, * Panoramic images with artifacts in and around the lesion.

Get updates

Get notified about this study

Sign up to get updates when this study changes or when new studies for Patient with A history of previous radiotherapy are added.

Our safety recommendation!

By submitting, you agree to our Terms of use

Conditions

The condition(s) this trial relates to.

As listed by the trial registrant

The condition terms exactly as the trial's registrant entered them.

Contacts and locations

Locations

  • Izmir Katip Celebi University

    Izmir, Çiğli, 35640, Turkey (Türkiye)