Radiomic evaluation of mandibular condyles in skeletal Class I, II, and III individuals using CBCT

To compare cone-beam computed tomography (CBCT)-derived radiomic features of the mandibular condyle among adults with Class I, Class II, and Class III sagittal skeletal classes in a Turkish subpopulation of dental patients. This retrospective study included 96 adults, with 32 individuals in each sagittal skeletal class. Both mandibular condyles were segmented, resulting in 192 condyles for analysis. Radiomic features were extracted from CBCT images using PyRadiomics integrated into 3D Slicer. Thirteen predefined features representing shape, first-order intensity, and texture characteristics were evaluated. Generalized estimating equations (GEE) were used to compare radiomic features among skeletal classes while accounting for within-subject correlation between bilateral condylar measurements. Sagittal skeletal class and condylar side were included as main effects, and Bonferroni-adjusted pairwise comparisons were performed using estimated marginal means. Significant skeletal class effects were observed for MeshVolume ( p < 0.001), SurfaceVolumeRatio ( p = 0.046), Mean ( p < 0.001), Median ( p < 0.001), Skewness ( p < 0.001), GLCM_Correlation ( p < 0.001), GLRLM_RunLengthNonUniformity ( p = 0.045), and GLSZM_LargeAreaEmphasis ( p < 0.001). Bonferroni-adjusted pairwise comparisons showed that MeshVolume and Mean were significantly lower in Class II than in Classes I and III, whereas Median differed significantly among all three skeletal classes. Skewness differed significantly between Class I and both Classes II and III, while GLCM_Correlation was significantly higher in Class III than in Classes I and II. GLSZM_LargeAreaEmphasis differed significantly between Classes I and II and between Classes II and III. Although overall skeletal class effects were significant for SurfaceVolumeRatio and GLRLM_RunLengthNonUniformity, none of their Bonferroni-adjusted pairwise comparisons reached statistical significance. No significant skeletal class effects were observed for SurfaceArea, MajorAxisLength, MinorAxisLength, Elongation, or NGTDM_Coarseness. After accounting for within-subject correlation between bilateral condyles, CBCT-based radiomic analysis demonstrated differences in selected mandibular condylar shape, voxel-intensity, and texture characteristics among sagittal skeletal classes. These findings indicate an association between sagittal skeletal class and quantitative imaging-derived characteristics of the mandibular condyle; however, their biological and clinical significance requires further validation. CBCT-based radiomic analysis may complement conventional morphometric assessment by providing additional quantitative information on mandibular condylar characteristics across different sagittal skeletal classes. However, its clinical utility for individual diagnosis, treatment planning, or prediction requires further validation in larger, multicenter studies using standardized imaging protocols.