Intraoperative and experimental evaluation of an optical tactile surface scanning system for autologous bone grafting

We present a system that enables a surgeon to acquire precise information about the shape of a bone part intraoperatively and without the use of 3D imaging Using a navigation system, we track the position of the bone part and the position of a spatula with a known geometry. After initially creating a binary voxel-volume around the bone piece, we remove all voxel where the spatula was moved through. When the spatula is moved along the surface of the bone, we can iteratively create a model of the bone surface from the remaining voxels. This is done interactively with visual feedback to the user. We used three plastic bone replicas in order to access the system's accuracy. We compared the carved model with a high precision laser scanner model. The mean distance of 4.2 Mio measured surface points was 0.57 mm, standard deviation 0.39 mm, the maximum distance was 3.03 mm, 95% of all measured points were closer than 1.34 mm to the reference surface. We tested the applicability of the system during an intervention showing that the system can be used in clinical routine. In future work, the intraoperatively acquired models will be used to determine an optimal position for the bone graft donor site.