Application of a calibration method provides more realistic results for a finite element model of a lumbar spinal segment

Background An important step in finite element modeling is the process of validation to derive clinical relevant data. It can be assumed that defect states of a finite element model, which have not been validated before, may predict wrong results. The purpose of this study was to show the differences in accuracy between a calibrated and a non-calibrated finite element model of a lumbar spinal segment for different clinical defects. Methods. For this study, two geometrically identical finite element models were used. An in vitro experiment was designed, deriving data for the calibration. Frequently used material properties were obtained from the literature and transferred into the non-calibrated model. Both models were validated on three clinical defects: bilateral hemifacetectomy, nucleotomy and interspinous defects, whereas in vitro range of motion data served as control points. Predictability and accuracy of the calibrated and non-calibrated finite element model were evaluated and compared. Findings. Both finite element models could mimic the intact situation with a good agreement. In the defects, the calibrated model predicted motion behavior with excellent agreement, whereas the non-calibrated model diverged greatly. Interpretation. Investigating the biomechanical performance of implants and load distribution of different spinal structures by numerical analysis requires not only good agreement with the intact segment, but also with the defect states, which are initiated prior to implant insertion. Because of more realistic results the calibration method may be recommended, however, it is more time consuming. (c) 2006 Elsevier Ltd. All rights reserved.