In order to provide real-time intraoperative guidance, computer assisted surgery (CAS) systems often rely on
computationally expensive algorithms. The real-time constraint is especially challenging if several components such as
intraoperative image processing, soft tissue registration or context aware visualization are combined in a single system.
In this paper, we present a lightweight approach to distribute the workload over several workstations based on the
OpenIGTLink protocol. We use XML-based message passing for remote procedure calls and native types for transferring
data such as images, meshes or point coordinates. Two different, but typical scenarios are considered in order to evaluate
the performance of the new system. First, we analyze a real-time soft tissue registration algorithm based on a finite
element (FE) model. Here, we use the proposed approach to distribute the computational workload between a primary
workstation that handles sensor data processing and visualization and a dedicated workstation that runs the real-time FE
algorithm. We show that the additional overhead that is introduced by the technique is small compared to the total
execution time. Furthermore, the approach is used to speed up a context aware augmented reality based navigation
system for dental implant surgery. In this scenario, the additional delay for running the computationally expensive
reasoning server on a separate workstation is less than a millisecond. The results show that the presented approach is a
promising strategy to speed up real-time CAS systems.