Advancing Cortical Bone Mapping with global optimization.

Cortical thickness is important in many domains of bone health research. Cortical Bone Mapping (CBM) is a model-based method for measuring cortical thickness in clinical CT, but has not gained traction in the high-resolution peripheral computed tomography (HR-pQCT) community. In this work, we advance CBM by introducing two novel extensions that fit the model globally rather than locally, allowing for spatial regularization of fitted parameters. With two repeat-measures precision HR-pQCT datasets (of the redproximal tibia, distal radius and distal tibia), we show that both the original and novel CBM methods have equivalent or better precision than current standard methods for measuring mean cortical thickness (e.g. for the distal tibia, 0.42-0.44% versus 0.60 %), and that the new methods have lower precision errors in spatially resolved cortical thicknesses (19%-21%) than the original CBM method (46%). Adopting CBM for HR-pQCT analysis could minimize the need for endosteal contour correction, reducing labour burden and improving reproducibility.