Introduction: Quantitative computed tomography (QCT) is a powerful imaging that allows 3D evaluation of bone quality according to apparent density. Its clinical potential is well recognized and it continuously gains popularity as an important diagnostic tool. However, reproducibility of this measure is still unclear.
Objective: The purpose of this study is to evaluate the repeatability and the reproducibility of a semi-automated method to assess bone mineral density using QCT.
Materials and Methods: QCT images of two patients affected by thoracic fracture were analyzed three times by three observers. Radiological phantoms were also imaged to establish relationships between grey level and bone density. Vertebral bodies adjacent to the fractured vertebra were identified and segmented according to 4 region of interest (ROI) using an active contour method. ROI's average density was calculated using the established relationship. Repeatability and reproducibility of the method were determined from a variance analysis (p=0,05).
Results: ROI's average density varies between 118 and 191 mg/cm3. The standard deviation is less than 12 mg/cm3 for all observers. The observer accounts for less than 12% of the total variability, which is not statistically significant (p>0,100).
Conclusion: The results show that the proposed bone density assessment method is reliable and offers a good inter- and intra-observer reproducibility.
Significance: The evaluation of bone quality could affect the choice of the treatment of fractures. The development of a semi-automated segmentation method allows reducing uncertainties related to the human intervention and to obtain reproducible results.