Development of a clinically applicable tool for bone density assessment.

OBJECTIVES: To assess the accuracy and reliability of new software for radiodensitometric evaluations.
METHODS: A densitometric tool developed by MevisLab was used in conjunction with intraoral radiographs of the premolar region in both in vivo and laboratory settings. An aluminum step wedge was utilized for comparison of grey values. After computer-aided segmentation, the interproximal bone between the premolars was assessed in order to determine the mean grey value intensity of this region and convert it to a thickness in aluminum. Evaluation of the tool was determined using bone mineral density (BMD) values derived from decalcified human bone specimens as a reference standard. In vivo BMD data was collected from 35 patients as determined with dual X-ray absorptiometry (DXA). The intra and interobserver reliability of this method was assessed by Bland and Altman Plots to determine the precision of this tool.
RESULTS: In the laboratory study, the threshold value for detection of bone loss was 6.5%. The densitometric data (mm Al eq.) was highly correlated with the jaw bone BMD, as determined using dual X-ray absorptiometry (r = 0.96). For the in vivo study, the correlations between the mm Al equivalent of the average upper and lower jaw with the lumbar spine BMD, total hip BMD and femoral neck BMD were 0.489, 0.537 and 0.467, respectively (P < 0.05). For the intraobserver reliability, a Bland and Altman plot showed that the mean difference +/- 1.96 SD were within +/-0.15 mm Al eq. with the mean difference value small than 0.003 mm Al eq. For the interobserver reliability, the mean difference +/-1.96 SD were within +/-0.11 mm Al eq. with the mean difference of 0.008 mm Al eq.
CONCLUSIONS: A densitometric software tool has been developed, that is reliable for bone density assessment. It now requires further investigation to evaluate its accuracy and clinical applicability in large scale studies.