Loading...
Sinus Pathology Detection in Panoramic Radiographs Verified by CBCT Imaging
Yang, Kitty
Yang, Kitty
Citations
Altmetric:
Genre
Thesis/Dissertation
Date
2025-05
Advisor
Committee member
Group
Department
Oral Biology
Permanent link to this record
Collections
Research Projects
Organizational Units
Journal Issue
DOI
https://doi.org/10.34944/8xpg-vp24
Abstract
Introduction: The accurate identification of maxillary sinus pathology plays a critical role in dental diagnostics, as these conditions can significantly impact treatment outcomes. Panoramic radiographs (PR) are frequently used due to their accessibility and lower radiation dose compared to cone-beam computed tomography (CBCT). However, questions remain about their reliability in detecting various sinus pathologies.
Objective: This study seeks to evaluate the diagnostic accuracy of PR for identifying maxillary sinus abnormalities, using CBCT and expert oral radiologist evaluation as reference standards. Specific maxillary sinus conditions evaluated include mucosal thickening, mucous retention cysts, sinus opacification, antral polyposis, sinus border disruption, and periapical lesions.
Methods: PR and CBCT from patients seen at TUKSOD were extracted from PACS® software, additional information was collected via the axiUm® database. Inclusion criteria consisted of patients age ≥18, having PR and CBCT taken within six months apart. Exclusion criteria involved PR with technique error, severe artifacts, low image quality, or missing sinus anatomy on the CBCT. Quality control was performed by removing duplicate images, blurred images, technique errors, and artifacts. All images had a unique identifier and were viewed on Weasis DICOM Viewer, version 4.5.1. Three residents were calibrated and served as readers for PR diagnoses, with CBCT diagnoses confirmed by two expert radiologists. After reviewing each PR, residents documented the presence of absence of sinus diseases and characterized the type of disease in an Excel form without time constraints. Inter-rater reliability was assessed using Kappa statistics and intraclass correlation coefficients (ICC). Diagnostic performance metrics, including sensitivity, specificity, precision, F1 score, accuracy, and AUC, were calculated to account for data imbalance.
Results: 79 PR were identified, amounting to a total of 158 sinuses. After excluding 16 uncertain responses, 142 sinuses were included in the final analysis. The analysis showed varying agreement levels among reviewers, with an overall inter-rater reliability of ICC=0.40. When comparing PR evaluations to the CBCT reference standard, kappa values varied by feature, with the highest agreement for mucosal thickening (K=0.51 for Reader 1) and lower agreement for rarer findings like antral polyposis (Κ=0.14). Overall, PR showed variable ability to detect sinus abnormalities, with precision values below 15% for subtle findings such as retention pseudocysts and antral polyposis.
Conclusion: Maxillary sinus conditions were able to be diagnosed in PR to an extent. Results demonstrate disagreement on maxillary sinus pathology while using PR. There is a promising outlook of developing an AI algorithm that can improve sinus pathology detection accuracy using PR without the necessary corresponding CBCT.
Description
Citation
Citation to related work
Has part
ADA compliance
For Americans with Disabilities Act (ADA) accommodation, including help with reading this content, please contact scholarshare@temple.edu