Abstract
Mandibular cortical width (MCW) on dental panoramic radiograph (DPR) is considered as a guideline for screening osteoporosis. We have previously developed the automated scheme for measuring MCW based on the profile of cortical bone. However, in cases of the presence of linear structures in cortex, the MCW may not be accurately measured. In this study, we propose a computerized scheme that identifies the risk of osteoporosis by measuring the dense part of cortical bones on the basis of the extracted linear structure. Our proposed scheme was evaluated by two sets of DPR obtained at Asahi University Hospital and dental clinics. Asahi cases are composed of 100 DPR including 26 osteoporotic cases based on dual energy X-ray absorptiometry (DXA). Clinic cases are composed of 458 DPR from multiple dental clinics, including 21 suspected osteoporotic cases based on the image findings by dental radiologists. Experimental results on the Asahi cases showed that the sensitivity and specificity for identifying osteoporotic patients were 80.8% and 94.6%, respectively, whereas those on the clinic cases were 90.5% and 79.2%, respectively. The automated MCW measurement technique using extraction of linear structures on DPR has a potential to identify asymptomatic osteoporotic patients and generalizability in different imaging environments.
