Background
Adolescent idiopathic scoliosis (AIS) with a Cobb angle larger than 45° is a spinal deformity that needs surgical treatment to stop progression and reduce pain. Spinal fusion is a surgical procedure where rods are attached to the spine through screws to diminish the Cobb angle.
Objective
The metal rods are pre-bent to correct the spinal deformity as much as possible. This rod shape is currently based on experience of the orthopedic surgeon. The aim of this thesis is to use Fi-nite element analysis (FEA) to predict a suitable rod shape for AIS patients that need to un-dergo surgery. 
Method
An FE model of an AIS spine was created based on high quality CT images. Rotation of ver-tebrae led to the optimal shape of the spine in the coronal plane. Vertebral bodies in this model were one by one rotated to a new position from top to bottom, to bring the spine to a new state and align it as much as possible to a normal physiological shape. However, alignment of the spine is an iterative process and rotation of every subsequent vertebra affects the shape of the entire spine. For each vertebral rotation, the reaction moments to maintain the new position is determined in the FE spine model. The acquired reaction moments are transferred to an FE rod model to provide the pre-bending of a rod that can bring the spine in its new state.
Results
The FE spine model was corrected to reduce the Cobb angle as much as possible. The highest stresses were found at the spinal apex. The reaction moments were transferred to a rod that became deformed in such a manner that straightened the spine model and corrected the Cobb angle to its pre-determined position with 33% reduction.
Discussion
This is an exploratory study that examines the possibility to predict ideal rod shapes pre-operatively to reduce surgical duration and improve the quality of the procedure. An accessory rod shape was accomplished but could not be validated due to lack of comparable studies. Improved FE spine models will lead to higher reliability of FE pre-bent rod models. This can be accomplished by, for example, inclusion of ligaments, muscles and other structures in the FE model and use of time-dependent material properties. The current model did not lead to a perfect aligned physiological spine and optimization algorithms should be developed in future models to further improve pre-bending of the rods.
Conclusion
Pre-operative prediction can improve the quality and reduce the time of spinal fusion. Im-provements to the FE spine- and rod model, such as automation, generalization, inclusion of more structures, and optimization algorithms will lead to optimal pre-bending of rods and bet-ter and quicker surgical outcome.