OBJECTIVE: To assess the diagnostic accuracy of panoramic radiography (PR), panoramic radiography combined with intraoral radiography (PR+IR), and CT in detecting the supero-inferior extent of tumor invasion of the mandible by gingival carcinoma. METHOD: PR, PR+IR, and CT images of the mandible in 37 patients with gingival carcinoma were evaluated by five oral radiologists for the supero-inferior extent of bone invasion using ROC analysis. The mean ROC curve area (Az) of each observer for the different imaging modalities was analysed by nonparametric two-way ANOVA. P<0.05 was considered statistically significant. RESULTS: The mean Az for the detection of bone invasion were 0.88+/-0.03 for PR, 0.77+/-0.12 for PR+IR, and 0.87+/-0.03 for CT (P=0.0907). The mean Az for the detection of bone invasion beyond the alveolus was 0.89+/-0.07 for PR, 0.85+/-0.08 for PR+IR, and 0.83+/-0.06 for CT (P=0.5438). The mean Az for the detection of bone invasion beneath the mandibular canal were 0.94+/-0.04 for PR, 0.94+/-0.02 for PR+IR, and 0.91+/-0. 04 for CT (P=0.2466). No statistically significant differences were observed in Az between PR, PR+IR, and CT. CONCLUSION: We consider that PR+IR should be adopted as the initial imaging modality to determine the extent of supero-inferior invasion of the mandible in gingival carcinoma.
OBJECTIVE: To assess the diagnostic accuracy of panoramic radiography (PR), panoramic radiography combined with intraoral radiography (PR+IR), and CT in detecting the supero-inferior extent of tumor invasion of the mandible by gingival carcinoma. METHOD: PR, PR+IR, and CT images of the mandible in 37 patients with gingival carcinoma were evaluated by five oral radiologists for the supero-inferior extent of bone invasion using ROC analysis. The mean ROC curve area (Az) of each observer for the different imaging modalities was analysed by nonparametric two-way ANOVA. P<0.05 was considered statistically significant. RESULTS: The mean Az for the detection of bone invasion were 0.88+/-0.03 for PR, 0.77+/-0.12 for PR+IR, and 0.87+/-0.03 for CT (P=0.0907). The mean Az for the detection of bone invasion beyond the alveolus was 0.89+/-0.07 for PR, 0.85+/-0.08 for PR+IR, and 0.83+/-0.06 for CT (P=0.5438). The mean Az for the detection of bone invasion beneath the mandibular canal were 0.94+/-0.04 for PR, 0.94+/-0.02 for PR+IR, and 0.91+/-0. 04 for CT (P=0.2466). No statistically significant differences were observed in Az between PR, PR+IR, and CT. CONCLUSION: We consider that PR+IR should be adopted as the initial imaging modality to determine the extent of supero-inferior invasion of the mandible in gingival carcinoma.