Radiomics-based quantitative assessment of the mandibular cortical index on panoramic radiographs

Area of Science:

• Dentistry
• Radiology
• Medical Imaging Analysis

Background:

• The Mandibular Cortical Index (MCI) is used to assess mandibular cortical bone status.
• Current MCI assessment relies on subjective interpretation of panoramic radiographs.
• There is a need for objective and reproducible methods for MCI evaluation.

Purpose of the Study:

• To quantitatively evaluate mandibular cortical changes using radiomic features from panoramic radiographs.
• To improve the objectivity and reproducibility of MCI assessment.
• To explore the utility of radiomics in distinguishing different categories of mandibular cortical resorption.

Main Methods:

• Retrospective analysis of 159 panoramic radiographs categorized into C1, C2, and C3 groups.
• Segmentation of the right mandibular cortex and extraction of radiomic features using PyRadiomics.
• Evaluation of five feature classes: shape, first-order, GLCM, GLRLM, and GLSZM.
• Inter-observer reproducibility assessed using intraclass correlation coefficients (ICC).

Main Results:

• Several first-order radiomic features (Energy, Entropy, IQR, MAD, rMAD, Variance) significantly distinguished between normal (C1) and resorbed cortices.
• Texture analysis revealed significant differences in GLCM features (Correlation, Cluster Shade, Joint Entropy) between groups.
• GLRLM (Gray Level Variance) and GLSZM (Zone Entropy) also showed significant group effects.
• No radiomic features remained significant after FDR correction for comparisons between moderately (C2) and severely (C3) resorbed cortices.

Conclusions:

• Radiomics can objectively quantify morphometric and microtextural differences in mandibular cortical bone.
• Radiomic feature analysis shows potential for enhancing the objectivity of MCI assessment on panoramic radiographs.
• Further research may refine radiomics application for precise differentiation of resorption severity.