Accuracy of Different Image Modalities in Detection of Bone Lesions

Methods: 7 dried human mandibles were used for this study. The mandibles were put inside water bath made of Plexiglass plates. Then PSP and CBCT imaging were performed. Mandibles were struck by a coping saw in buccolingual shape and dimensions of the oval lesions in 6.1 °Á 6.1 mm, 3 °Á 6.1 mm and 4 °Á 4 mm were created by milling into the spongy bone. After fixing the two parts together (buccal and lingual) imaging was repeated again. Presence or absence of lesions on images were obtained and then evaluated and recorded by two observers. Using SPSS 16, compatibility level, sensitivity, and specificity and ROC analysis were determined for each observer.

Results: Internal agreement of observers in all three imaging methods was low to moderate (kappa≤0.613). The external agreement in all three imaging methods was moderate (kappa=0.406). The area under the curve (AZ) of imaging techniques in each observation was not significantly different. In addition, the area under the lesion size curve (AZ) for the two observers was not significantly different.

Conclusions: The lesions confined to spongy bone could be identified on film, PSP, CBCT. But interpretation of PSP and CBCT images needed more skills and training.