I am a Ph.D. student working with Dr. Peter Cripton in the Orthopaedic and Injury Biomechanics Laboratory at the University of British Columbia (UBC) and Dr. Gunter Siegmund of UBC and MEA Forensic Engineers and Scientists in Vancouver. I was fortunate to be awarded the International Society of Biomechanics travel grant in 2009. This award allowed me to travel to Bern, Switzerland in August to collaborate with researchers at the Institute for Surgical Technology and Biomechanics at the University of Bern. The purpose of this collaboration was to learn and develop a new research method for describing three‐dimensional spinal kinematics using MRI and fluoroscopy. This new technique will be used in my thesis project entitled "Kinematics and muscle activation of the cervical spine during vehicle rollover accidents". The aim of this study is to investigate muscle activation and occupant neck motion during rollover accidents using a human subject rollover simulation. To achieve this, neck muscle activation will be recorded using electromyography and cervical spine motion will be captured using two‐dimensional fluoroscopy. However, with the complex and often coupled 3D motion of the cervical spine, planar analysis is prone to large errors due to x‐ray projection principles and out‐of‐plane vertebral motion. When studying a dynamic event such as a rollover simulation, in which subject motion cannot be constrained to one plane, 3D motion capture is essential. Therefore, the goal of my visit to the University of Bern was to learn 2D to 3D registration techniques and adapt them for the cervical spine for the rollover experiment. During my visit I was fortunate to gain experience with two different 2D‐3D registration techniques and was able to assess their application to the cervical spine.
A comprehensive methodology for 2D‐3D registration of CT and fluoroscopy data has been developed for lumbar spine motion by Dr. Stephen Ferguson and Dr. Paul Thistlethwaite at the University of Bern. This registration technique combines dynamic 2D images from a single fluoroscope with subject‐specific, static, 3D data sets from CT scans to achieve highly accurate and precise 3D kinematics. The accuracy and repeatability of the algorithm was evaluated. The advantages and disadvantages of using MRI instead of CT for the static input data to the kinematic algorithm were also evaluated because unlike CT, MR scanning does not involve additional radiation exposure. Finally, we assessed the feasibility of using this technique with the cervical spine. This feature‐based registration technique was challenging to apply to the cervical spine due to the different size and geometry of the vertebrae compared to the lumbar spine. Adjustments are currently being made to the algorithm and related programs to improve the accuracy and repeatability for application to the cervical spine. Another technique was explored for this 2D‐3D registration that would not rely as heavily on the specific shape and size of the vertebrae. This technique has been developed and validated by Dr. Guoyan Zheng at the University of Bern and uses CT and plain radiographic images to determine acetabular cup orientations after total hip arthroplasty. This hybrid registration technique quantifies rigid transformations by combining iterative landmark‐based registration with a 2D‐3D intensity‐based registration. Preliminary work we completed in applying this technique to the lumbar spine was promising. Therefore, we are presently working to quantify the accuracy of this method and its potential application to the cervical spine.
This successful collaboration was the critical starting point for two very promising applications of 2D to 3D registration with the cervical spine. Current work is aimed at further developing and evaluating both models in a dynamic cervical spine task using MRI scans from the University of British Columbia. Without the valuable interactions that I had with Drs. Ferguson and Zheng at the University of Bern it is unlikely that I would have identified both of these promising techniques for 2D‐3D registration. This occurred as a direct result of the ISB travel grant and I would like to thank the ISB for funding this invaluable experience. I would also like to thank the Mathematics of Information Technology and Complex Systems Centre of Excellence and the Natural Sciences and Engineering Research Council of Canada for scholarship and project grant funding for supporting my research.
Robyn Newell
