颅内电极定位中的多模图像配准研究

Research on the multi-modality image registration for subdural electrode localization

目的:探讨不同配准方法在颅内电极定位中的表现,寻找适合于颅内电极定位的方法。方法:分别利用均方测量和Mattes互信息方法作为代价函数,将患者植入电极后的CT图像与植入电极前的MRI图像进行配准,对比两种方法的表现。结果:两种方法均能较为准确地定位旋转中心的位置。在同一CT和MRI体数据对中,不同断层图像对进行配准时旋转角度的波动较大;而不同断层配准图像的平均旋转角度与人工配准的结果大致接近。结论:不同断层图像对配准结果的平均值可作为整个CT和MRI体数据的配准参数使用,并得到相对良好的匹配结果。

Objective： To discuss the performance of different methods in multi-modal image registration for subdural electrode localization. Methods：Mean Squares and the Mattes Mutual Information metric methods were employed to align the post-implantation CT volume and the pre-implantation MR volume. Results： Both methods performed well in the determination of the rotation center. Rotation angles of different image pairs of the same volume pair fluctuated a lot. The mean rotation angles of images within one data set are approximate to the manual results that are considered to be the actual result for registration despite their fluctuation range. Conclusion：The average value of the registration results of the images sampled to implement the process could be used to align the whole volume pair and get a relatively good match.