Neurosurgical navigation systems using preoperative images have a big problem in their accuracy which is caused byBrain Shift. To this problem we have developed an image modification technique using brain mechanics. In this study a method is introduced in which the intraoperative brain shape is estimated through Finite Element Method. In the process the nearest model shape to intraoperative brain surface is firstly estimated by modifying an FEM result through adjusting mechanical property, then the residual between the result and the actual shape is obtained by a virtual load on the brain surface.
The accuracy of our method is evaluated with phantom experiments.We prepare two kinds of phantom experiments for investigating the most adaptive conditions in terms with bindings, initial model parameters, etc. One is that using mathematical model whose mechanical property is elastic, and the other is that using soy bean curd whoes mechanical property is non-linear.