Related Experiment Video
Updated: Aug 14, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Perilesional and Contextual Radiomic Features Improve Differentiation of Odontogenic Cysts on Panoramic Radiographs
Barbara Obuchowicz1, Joanna Zarzecka1, Marzena Jakubowska2
1Department of Conservative Dentistry with Endodontics, Institute of Dentistry, Jagiellonian University Medical College, Montelupich 4, 31-155 Krakow, Poland.
Journal of Clinical Medicine
|August 13, 2026
Summary
Radiomics analysis of panoramic radiographs shows promise in differentiating odontogenic cysts, improving classification accuracy when combining texture, morphological, and perilesional features for better diagnosis.
Area of Science:
- Oral and Maxillofacial Radiology
- Medical Imaging Informatics
- Quantitative Imaging Biomarkers
Background:
- Conventional radiographic assessment of odontogenic cysts faces challenges due to overlapping imaging features and limited specificity.
- Accurate differentiation of cyst subtypes is crucial for appropriate clinical management and follow-up strategies.
- Radiomics offers a quantitative approach to extract imaging biomarkers for improved non-invasive lesion characterization.
Purpose of the Study:
- To evaluate the efficacy of radiomics analysis on panoramic radiographs for differentiating odontogenic cysts.
- To compare the classification performance of different radiomic feature groups (texture, morphological, perilesional).
- To assess the impact of integrating various feature types on diagnostic accuracy.
Main Methods:
- Histopathologically confirmed odontogenic cysts (n=63) including odontogenic keratocyst (OKC), radicular cyst (RC), and dentigerous cyst (DC).
- Extraction of radiomic features from segmented panoramic radiographs: agnostic texture, morphological/intensity, and perilesional features.
- Classification using logistic regression with leave-one-out cross-validation for pairwise and four-class analyses.
Main Results:
- Radiomic features showed significant differences between cyst and anatomical regions.
- Pairwise classification performance varied, with highest balanced accuracy for DC vs. OC-NOS (0.838).
- Integrating morphological and perilesional features improved four-class balanced accuracy from 0.416 (texture-only) to 0.487 (combined, p=0.002).
Conclusions:
- Radiomics analysis of panoramic radiographs provides moderate, preliminary differentiation of odontogenic cysts.
- Incorporating perilesional and spatial features enhances classification accuracy.
- Substantial overlap necessitates multimodal diagnostic approaches and further validation in larger cohorts.
