基于伪超声和互信息的磁共振与超声图像配准

摘要

The registration of intra-operative ultrasound (US)and pre-operative magnetic resonance (MR)data is very impor-tant in surgical navigation.A novel registration method of MR and US based on their imaging theories was presented.The proposed method employed pseudo-ultrasound derived from MR data and mutual information to implement the registration of two modal images. Additionally,multi-resolution analysis and Powell algorithm were also used in the searching of the largest mutual information,which were effective to reduce the probability of falling into local minima while improved the registration accuracy.The experimental results show that the proposed registration method has higher accuracy than that of mark point method which is widely used in the current surgi-cal navigation system.%手术中超声图像与术前磁共振图像的配准在手术导航系统中非常重要。我们利用磁共振和超声成像原理，提出了基于伪超声和互信息，并结合多分辨率技术与Powell优化算法对两种模态图像进行配准的方法，该方法可以有效降低寻优过程中陷入局部极值收敛的概率，提高两种模态图像的配准精度。实验结果表明，我们提出的基于伪超声和互信息的配准方法比目前手术导航系统中普遍采用的标记点方法具有更高的配准精度。