LAUSR

Intraoperative brain shift decreases the accuracy of neuronavigation systems based on preoperative images. In this paper, this problem is addressed by calculating an estimation of brain shift which can be employed to update the preoperative brain images. Therefore, the precision of navigation can be improved. In this regard, a brain shift estimation method is proposed using an atlas of brain deformations and constrained Kalman filter (ACKF). In addition, it is proven that the obtained ACKF estimation is the best unbiased minimax estimation when the risk function is the estimation error variance. Furthermore, a comparison is performed between the ACKF and two existing methods, namely, CKF and atlas-based method. The comparison demonstrates that the ACKF results in a more accurate estimation and needs less computation time. Finally, the supremacy of the proposed ACKF method with respect to the CKF and atlas-based method is illustrated through simulation.