Computed tomography-based radiomics analysis of abscess and cellulitis in odontogenic infections

This study aimed to compare computed tomography (CT) radiomic features of abscesses and cellulitis associated with odontogenic infections in the head and neck region.

Fifty-eight patients who presented with pain and swelling due to odontogenic infections and underwent head and neck CT between July 2020 and August 2024 were retrospectively analyzed. A final diagnosis of abscess or cellulitis was established based on comprehensive clinical, radiological, and surgical findings. Using MaZda version 4.6.2.0, 279 texture features were extracted from three manually placed circular regions of interest positioned within the anterior, central, and posterior portions of the largest axial cross-section of each lesion. Statistical comparisons were performed for all extracted texture features using the Mann-Whitney U test with Bonferroni correction for multiple comparisons. Fisher's coefficient was subsequently used to rank the discriminatory ability of the significant features, and the 10 highest-ranked features were evaluated using receiver operating characteristic (ROC) analysis.

All extracted texture features remained significantly different after Bonferroni correction. Histogram-derived features demonstrated marked differences in the gray-level distribution within the representative circular regions of interest between the two groups. Gray-level co-occurrence matrix correlation values were consistently higher in abscesses than in cellulitis. ROC analysis of the 10 highest-ranked texture features demonstrated excellent discriminatory ability, with AUC values ranging from 0.95 to 0.99.

CT texture analysis identified objective imaging features that may help differentiate abscesses from cellulitis in odontogenic infections. However, further validation is required before clinical application.