A methodology for the generation and non-destructive characterisation of transverse fractures in long bones

Long bone fractures are common and although treatments are highly effective in most cases, it is challenging to achieve successful repair for groups such as open and periprosthetic fractures. Previous biomechanical studies of fracture repair, including computer and experimental models, have simplified the fracture with a flat geometry or a gap, and there is a need for a more accurate fracture representation to mimic the situation in-vivo. The aims of this study were to develop a methodology for generating repeatable transverse fractures in long bones in-vitro and to characterise the fracture surface using non-invasive computer tomography (CT) methods. Ten porcine femora were fractured in a custom-built rig under high-rate loading conditions to generate consistent transverse fractures (angle to femoral axis < 30 degrees). The bones were imaged using high resolution peripheral quantitative CT (HR-pQCT). A method was developed to extract the roughness and form profiles of the fracture surface from the image data using custom code and Guassian filters. The method was tested and validated using artificially generated waveforms. The results revealed that the smoothing algorithm used in the script was robust but the optimum kernel size has to be considered.