2016 Volume 65 Issue 3 Pages 256-263
Objective: It is known that an altered morphology of the mandibular cortex on panoramic radiographs is significantly correlated with osteoporosis. We developed a computer program that scans the mandibular inferior cortex and defines the mandibular cortical index (MCI) automatically.
Methods: The procedure of MCI analysis is as follows. As the first step, the mandibular contour was extracted. Next, bilateral measurement points were set in the mental foramen region. Then the thickness and the roughness of the mandibular cortex were calculated to determine the MCI classification: normal cortex (Class 1), mildly to moderately eroded cortex (Class 2), and severely eroded cortex (Class 3). When the measurement point is not correct, the observer can choose an appropriate measurement point in the mandibular cortex manually. A panoramic image database, in which the MCI classification was determined based on agreement among three experienced dental radiologists, was used to test the computer program. The database consisted of 78, 67, and 60 cases of Class 1, Class 2, and Class 3, respectively.
Results: The accuracy of computer-based classification was 94.9% for Class 1, 92.5% for Class 2, and 90.0% for Class 3. The diagnostic sensitivity to detect possible osteoporosis (Class 2 and 3) was 94.5%, and the specificity was 94.9%.
Conclusions: It was suggested that the present computer program can analyze MCI correctly.