Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology

Abstract Background Segmentation of airways and soft tissues on panoramic radiographs is a challenging yet crucial task in dental diagnostics, as these regions can often be confused with fractures or other lesions due to superimposition. This study aimed to perform segmentation of both airways and s...

Full description

Saved in:
Bibliographic Details
Main Authors: Aslıhan Şahan Keskin, İlknur Eninanç
Format: Article
Language:English
Published: BMC 2025-06-01
Series:BMC Oral Health
Subjects:
Online Access:https://doi.org/10.1186/s12903-025-06187-9
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849434245524619264
author Aslıhan Şahan Keskin
İlknur Eninanç
author_facet Aslıhan Şahan Keskin
İlknur Eninanç
author_sort Aslıhan Şahan Keskin
collection DOAJ
description Abstract Background Segmentation of airways and soft tissues on panoramic radiographs is a challenging yet crucial task in dental diagnostics, as these regions can often be confused with fractures or other lesions due to superimposition. This study aimed to perform segmentation of both airways and soft tissues on panoramic radiographs simultaneously using an artificial intelligence (AI)-based model. Methods Segmentation masks were created by annotating the nasal, oral, and oropharyngeal airways, along with the tongue, soft palate, and uvula, on 1,004 panoramic radiographs. Data augmentation and image processing techniques were applied to enhance dataset diversity. Of the radiographs, 72% were allocated for training, 18% for validation, and 10% for testing. A custom AI model based on the ResUNet architecture, comprising 74 layers and 24.3 million parameters, was developed utilizing the TensorFlow library. Performance metrics, including accuracy, precision, sensitivity, specificity, F1 score, intersection over union (IoU), and mean average precision (mAP) were evaluated. Results The areas AI model achieved an accuracy of 0.979, precision of 0.869, sensitivity of 0.870, specificity of 0.925, F1 score of 0.870, IoU of 0.777, and mAP of 0.500. Intra-observer agreement values ranged from 0.762 to 0.958. Conclusions To our knowledge, this is the first study to develop an AI -based model for segmentation of airways and soft tissues on panoramic radiographs. The proposed algorithm demonstrated high accuracy in identifying the regions of interest, enabling rapid and efficient radiographic analysis. This model has the potential to enhance decision support systems and reduce the risk of misdiagnosis. Clinical trial number Not applicable.
format Article
id doaj-art-2d6c5e8b85a74b1e82d9f437b4e706a6
institution Kabale University
issn 1472-6831
language English
publishDate 2025-06-01
publisher BMC
record_format Article
series BMC Oral Health
spelling doaj-art-2d6c5e8b85a74b1e82d9f437b4e706a62025-08-20T03:26:43ZengBMCBMC Oral Health1472-68312025-06-0125111410.1186/s12903-025-06187-9Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technologyAslıhan Şahan Keskin0İlknur Eninanç1Department of Dentomaxillofacial Radiology, Faculty of Dentistry, Sivas Cumhuriyet UniversityDepartment of Dentomaxillofacial Radiology, Faculty of Dentistry, Sivas Cumhuriyet UniversityAbstract Background Segmentation of airways and soft tissues on panoramic radiographs is a challenging yet crucial task in dental diagnostics, as these regions can often be confused with fractures or other lesions due to superimposition. This study aimed to perform segmentation of both airways and soft tissues on panoramic radiographs simultaneously using an artificial intelligence (AI)-based model. Methods Segmentation masks were created by annotating the nasal, oral, and oropharyngeal airways, along with the tongue, soft palate, and uvula, on 1,004 panoramic radiographs. Data augmentation and image processing techniques were applied to enhance dataset diversity. Of the radiographs, 72% were allocated for training, 18% for validation, and 10% for testing. A custom AI model based on the ResUNet architecture, comprising 74 layers and 24.3 million parameters, was developed utilizing the TensorFlow library. Performance metrics, including accuracy, precision, sensitivity, specificity, F1 score, intersection over union (IoU), and mean average precision (mAP) were evaluated. Results The areas AI model achieved an accuracy of 0.979, precision of 0.869, sensitivity of 0.870, specificity of 0.925, F1 score of 0.870, IoU of 0.777, and mAP of 0.500. Intra-observer agreement values ranged from 0.762 to 0.958. Conclusions To our knowledge, this is the first study to develop an AI -based model for segmentation of airways and soft tissues on panoramic radiographs. The proposed algorithm demonstrated high accuracy in identifying the regions of interest, enabling rapid and efficient radiographic analysis. This model has the potential to enhance decision support systems and reduce the risk of misdiagnosis. Clinical trial number Not applicable.https://doi.org/10.1186/s12903-025-06187-9Artificial intelligencePanoramic radiographyAirwaysSoft tissues
spellingShingle Aslıhan Şahan Keskin
İlknur Eninanç
Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology
BMC Oral Health
Artificial intelligence
Panoramic radiography
Airways
Soft tissues
title Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology
title_full Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology
title_fullStr Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology
title_full_unstemmed Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology
title_short Segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology
title_sort segmentation of airways and soft tissues on panoramic radiographs using artificial intelligence technology
topic Artificial intelligence
Panoramic radiography
Airways
Soft tissues
url https://doi.org/10.1186/s12903-025-06187-9
work_keys_str_mv AT aslıhansahankeskin segmentationofairwaysandsofttissuesonpanoramicradiographsusingartificialintelligencetechnology
AT ilknureninanc segmentationofairwaysandsofttissuesonpanoramicradiographsusingartificialintelligencetechnology