Reliability of HR-pQCT derived cortical bone structural parameters when using uncorrected instead of corrected automatically generated endocortical contours in a cross-sectional study: the Maastricht study

Most HR-pQCT studies examining cortical bone use an automatically generated endocortical contour (AUTO), which is manually corrected if it visually deviates from the apparent endocortical margin (semi-automatic method, S-AUTO). This technique may be prone to operator-related variability and is time consuming. We examined whether the AUTO instead of the S-AUTO method can be used for cortical bone analysis. Fifty scans of the distal radius and tibia from participants of The Maastricht Study were evaluated with AUTO, and subsequently with S-AUTO by three independent operators. AUTO cortical bone parameters were compared to the average parameters obtained by the three operators (S-AUTOmean). All differences in mean cortical bone parameters between AUTO and S-AUTOmean were < 5%, except for lower AUTO cortical porosity of the radius (− 16%) and tibia (− 6%), and cortical pore volume (Ct.Po.V) of the radius (− 7%). The ICC of S-AUTOmean and AUTO was > 0.90 for all parameters, except for cortical pore diameter of the radius (0.79) and tibia (0.74) and Ct.Po.V of the tibia (0.89), without systematic errors on the Bland–Altman plots. The precision errors (RMS-CV%) of the radius parameters between S-AUTOmean and AUTO were comparable to those between the individual operators, whereas the tibia RMS-CV% between S-AUTOmean and AUTO were higher than those of the individual operators. Comparison of the three operators revealed clear inter-operator variability. This study suggests that the AUTO method can be used for cortical bone analysis in a cross-sectional study, but that the absolute values—particularly of the porosity-related parameters—will be lower.