Detailed knowledge about the cortical bone geometry and thickness can be used for many applications within orthopedic and trauma care. Three-dimensional (3D) image data, contemporary image processing and analysis as well as Artificial Intelligence techniques could be used for complex bone models to be efficiently generated and analyzed. Such models potentially enable generation of information relevant for clinical applications, research and development, or educational purposes. The generation of highly accurate bone models for the distal radius is of interest.
To develop a method for generation of a highly accurate 3D statistical model of the distal radius and a learning system to assess its osteoporosis status.
Using 60 high-resolution peripheral quantitative computed tomography (HR-pQCT, 82µm image resolution) scans of the distal radius, a highly accurate 3D statistical cortical bone model of the latter has been successfully generated. In addition, an accurate learning system (accuracy 98.3%, AUC values > 0.96) has been developed and validated using 10-fold cross-validation. It allows to determine the osteoporosis status based on thickness data at given locations.
Teunis T, (MD) University Medical Center Utrecht, the Netherlands