PURPOSE: To examine the accuracy of computed tomography (CT) in predicting the presence of surgically confirmed abscess in patients presenting with deep maxillofacial infections (DMIs). MATERIALS AND METHODS: The medical records of 36 patients with clinically suspected DMI were reviewed. All patients underwent preoperative CT scan to determine whether abscess or cellulitis was present. The operative reports of all patients were compared with their preoperative CT scan findings. RESULTS: Of 36 patients, 30 had abscess predicted by preoperative CT scan. The positive predictive value for DMI based on CT findings interpreted by an oral surgeon and confirmed as an abscess at operation was 90%, whereas the negative predictive value was 33%. When findings were interpreted by a radiologist, the positive predictive value was 94.1% whereas the negative predictive value was 26.3%. Thirty patients had abscess predicted by preoperative CT scan when interpreted by an oral surgeon, whereas seventeen had abscess predicted by a radiologist. At operation, 3 of 30 patients had cellulitis whereas 27 had abscess, for a false-positive rate of 60% and false-negative rate of 12.9%. On the basis of preoperative radiography, 1 of 17 patients had cellulitis whereas 16 had abscess, for a false-positive rate of 16.7% and a false-negative rate of 46.7%. The agreement between CT and operative findings in predicting abscess by an oral surgeon was 80.6% and by a radiologist was 58.3%. CONCLUSION: CT yielded high sensitivity for the detection of DMI abscess but poor specificity, likely because of the paucity of cellulitis.
PURPOSE: To examine the accuracy of computed tomography (CT) in predicting the presence of surgically confirmed abscess in patients presenting with deep maxillofacial infections (DMIs). MATERIALS AND METHODS: The medical records of 36 patients with clinically suspected DMI were reviewed. All patients underwent preoperative CT scan to determine whether abscess or cellulitis was present. The operative reports of all patients were compared with their preoperative CT scan findings. RESULTS: Of 36 patients, 30 had abscess predicted by preoperative CT scan. The positive predictive value for DMI based on CT findings interpreted by an oral surgeon and confirmed as an abscess at operation was 90%, whereas the negative predictive value was 33%. When findings were interpreted by a radiologist, the positive predictive value was 94.1% whereas the negative predictive value was 26.3%. Thirty patients had abscess predicted by preoperative CT scan when interpreted by an oral surgeon, whereas seventeen had abscess predicted by a radiologist. At operation, 3 of 30 patients had cellulitis whereas 27 had abscess, for a false-positive rate of 60% and false-negative rate of 12.9%. On the basis of preoperative radiography, 1 of 17 patients had cellulitis whereas 16 had abscess, for a false-positive rate of 16.7% and a false-negative rate of 46.7%. The agreement between CT and operative findings in predicting abscess by an oral surgeon was 80.6% and by a radiologist was 58.3%. CONCLUSION: CT yielded high sensitivity for the detection of DMI abscess but poor specificity, likely because of the paucity of cellulitis.