Degenerative changes in the spine as well as back pain can be considered a common ailment. Incorrect loading of the lumbar spine structures is often considered as one of the factors that can accelerate degenerative processes, leading to back pain. For example, a degenerative change could be the occurrence of spinal stenosis following spondylolisthesis. Surgical treatment of spinal stenosis mainly focuses on decompressing the spinal canal with or without additional fusion through dorsal spondylodesis. There are differing opinions on whether fusion along with decompression provides potential benefits to patients or represents an overtreatment. Both conventional therapies and surgical methods aim to restore a “healthy” (or at least pain-free) distribution of load. Surprisingly little is known about the interindividual variability of load distribution in “healthy” lumbar spines. Since medical imaging does not provide information on internal forces, computer simulation of individual patients could be a tool to gain a set of new decision criteria for these cases. The advantage lies in calculating the internal load distribution, which is not feasible in in-vivo studies, as measurements of internal forces in living subjects are ethically and partially technically unfeasible. In the present research, the forward dynamic approach is used to calculate load distribution in multi-body models of individual lumbar spines. The work is structured into three parts: (I) Load distribution is quantified depending on the individual curvature of the lumbar spine. (II) Confidence intervals of the instantaneous center of rotation over time are determined, with which the motion behavior of healthy lumbar spines can be described. (III) Lastly, the effects of decompression surgeries on the load distribution of lumbar spines are determined.

One of the greatest goals in computer graphics is the aesthetic representation of objects. In addition to conventional methods, another field focuses on non-photorealistic renderings. The so-called example-based rendering is an area where users can transfer their art style to a pre-computed 3D rendering, using a hand-painted template. There are some algorithms that already provide impressive results, but their problem is that most of these procedures count as offline methods and are not able to produce results in real-time. For this reason, this work show a method that satisfies this condition. In addition, the influence of the run-time reduction on the results is investigated. Requirements are defined, to which the method and its results are examined. Other methods in this field are referenced and compared with their results.