Y Iwashita1. 1. Department of Dental Radiology, Kagoshima University Dental School, Kagoshima City, Kagoshima, Japan.
Abstract
OBJECTIVES: To develop an algorithm for automatic separation of cortical from cancellous bone in the mandible and for measuring mandibular bone mineral content (BMC). METHODS: A dried mandible was scanned together with a reference phantom by spiral CT. A new algorithm was developed specifically for the mandible and cross-sectional images reconstructed in the buccolingual plane. The cortical bone was extracted with a threshold of 80% of the peak height in the line-scan profile and the cancellous bone extracted separately. The volume and BMC were estimated. Images of cortical and cancellous bone were displayed in a plane approximating the orientation used for dental radiography and BMC expressed as aluminium-equivalents (Al-eqs). RESULTS: The ratio of cortical to cancellous bone by volume was 22:78, and the ratio of the BMC 60:40. Comparison of BMC from reconstructed CT with Al-eqs showed good correlation but was 1.5 times higher with CT. CONCLUSIONS: It is possible to separate cancellous from cortical bone automatically from CT images and to measure the BMC.
OBJECTIVES: To develop an algorithm for automatic separation of cortical from cancellous bone in the mandible and for measuring mandibular bone mineral content (BMC). METHODS: A dried mandible was scanned together with a reference phantom by spiral CT. A new algorithm was developed specifically for the mandible and cross-sectional images reconstructed in the buccolingual plane. The cortical bone was extracted with a threshold of 80% of the peak height in the line-scan profile and the cancellous bone extracted separately. The volume and BMC were estimated. Images of cortical and cancellous bone were displayed in a plane approximating the orientation used for dental radiography and BMC expressed as aluminium-equivalents (Al-eqs). RESULTS: The ratio of cortical to cancellous bone by volume was 22:78, and the ratio of the BMC 60:40. Comparison of BMC from reconstructed CT with Al-eqs showed good correlation but was 1.5 times higher with CT. CONCLUSIONS: It is possible to separate cancellous from cortical bone automatically from CT images and to measure the BMC.