Abstract
Differentiating mandibular osteoradionecrosis (ORN) from tumor recurrence remains challenging on preoperative imaging. We conducted a multicenter retrospective study including 24 patients with histologically confirmed ORN who underwent surgery and had preoperative CT available. Radiomic features were extracted using an IBSI-compliant pipeline and compared between pure ORN (n = 21) and ORN with recurrence (n = 3). Three features showed nominal uncorrected differences between groups. Given the rarity of histologically confirmed recurrence, this study was designed as an exploratory hypothesis-generating analysis. These findings suggest that CT radiomics may capture subtle imaging differences and warrant validation in larger cohorts.