In only one year, from July 1997 to July 1998, three partners in the Esprit funded KneesUp project have produced an accurate 3D model of the human knee, including soft tissues, to demonstrate the biomechanical environment of the knee joint. The team applied finite element modelling and high performance computing and networking (HPCN) technology to build the KneesUp simulation model for implant testing and injury prediction in car crash scenarios. For this purpose, a series of four concrete simulation cases have been developed and studied. These demonstrations include a normal gait cycle and a heel strike impact as showcases for the orthopaedic and implant industry, and a lateral pedestrian and toe panel intrusion foot impact as illustrative examples for the automotive industry. In addition, the partners used the model to observe the behaviour of a first prototype for a meniscal implant.
The knee joint constitutes one of the most complex structures in the human body, easily susceptible to injury due to its unfortunate anatomy and position at a great distance from the body's centre of gravity. Former attempts to simulate the intricate movement patterns of the knee joint have failed to render the dynamic changes in position, the contact area and the loading of the menisci in three dimensions. The KneesUp model is based upon the highly detailed geometric data acquired from Magnetic Resonance Imaging (MRI) scans of a cadaveric specimen. Algorithms were designed to identify the tissue type within the MRI scans. The partners generated a 3D surface model and remeshed the vital soft tissues to solid representations. In order to link the material properties of the specimen with the model, most of the physical parameters were retrieved from published research, leaving only a few to obtain from laboratory experimentation.
The final result consists of a high quality finite element model in PAM-SAFE code to perform a variety of simulations. The PAM-SAFE code was provided by Engineering Systems International, a member of the ESI-Group, which is specialized in the field of numerical simulation packages applied in virtual prototyping and biomechanical modelling. Kinematic data from the specimen was gathered in a special test rig by means of an advanced optical tracking system for later comparison with the simulation data, in order to validate the model. The entire process of accurate 3D knee modelling implies extremely large computing resources, which can only be delivered by HPCN. Therefore, the German National Research Center for Information Technology (GMD) has actively supported the KneesUp project as TTN (Technology Transfer Nodes) centre of excellence. Both computational cost and time could be held within limits thanks to the parallel computing capability available at ESI.
Third partner and co-developer of the KneesUp model is the University of Sheffield which has shown an ongoing interest in meniscal problems and in qualitative prosthetic alternatives to total removal of the menisci, the two horseshoe shaped pieces of fibrocartilage in the knee. The menisci assist in transferring load evenly from the thigh bone to the shin. Once damaged, they do not repair naturally but their final removal leads to osteoarthritis. Accurate prosthetic replacement could solve the problem but requires a deep insight in the forces inflicted on the implant by complex body movements. The KneesUp model provides the opportunity to execute preliminary tests on knee joint implants with the gait cycle simulation, since walking constitutes the predominant load case for knee joint function. This type of simulation is used to study the action of the muscles on the hamstring and quadriceps tendons.
A second simulation involves the landing on the slightly flexed knee, referred to as the heel strike, which frequently occurs in all kinds of sports. Useful data can be acquired from observing the impact force from the heel striking the floor up to the knee joint for the shoe and sports equipment and training industries. On the other hand, the development cycle for implants in the human knee joint will be shortened in the orthopaedic industry whereas the chances of failure implant in surgery will be reduced. Two other simulations have been developed for the automotive industry to help safety engineers in the design of vehicle structures which could substantially minimize the knee injuries. For this purpose, a frontal car crash with typical foot impact and a lateral pedestrian impact, in which a car bumper hits a restrained leg, have been selected.
Most of the simulations were run on Silicon Graphics and Cray hardware equipment at ESI and required 137 hours or nearly one week of wall clock time for a 50ms simulation on a 4-way Cray J90. With superior scaling, the partners anticipate that the run time can be reduced to 1.5 days on an 8-way J90 and to less than a day on the SGI 02000. Since highly advanced PAM-SAFE options were applied to the KneesUp model, the simulations were executed by means of the shared-memory code. The used codes are not yet available in the distributed-memory code version. The overall costs of the KneesUp project amounted to 262 Kecu, of which 205 Kecu were funded by the European Commission. For more detailed background information on the project, we refer to the medical area of the Technology Transfer Node at GMD.