MRI立体定向丘脑底核和杏仁核个体化、数字化、可视化图谱的研究

目的探讨建立个体化MRI立体定向丘脑底核（STN）和杏仁核（AM）数字化、可视化图谱的可行性。方法对1例成年男性健康自愿者应用1．5TMRI,在标准的脑立体定向空间做1mm层厚的MRI轴位脑扫描,在MRI上对STN和AM进行识别、手动分割、提取、存储和三维重建。结果三维重建的STN和AM表面光滑、形态逼真,清晰的显示了个体STN和AM在标准立体定向空间中的位置,可以直观的观察STN和AM的形态结构。重建后的STN和AM可以任意旋转、缩放及变换颜色。STN前后径较长,上下径和左右径相对较短,表现出典型的“双透镜”结构特点,而AM则表现出典型的“杏仁”形状。结论MRI立体定向个体化的STN和AM数字化、可视化具有可行性,如果要达到指导立体定向功能神经外科临床应用的目的,需解决计算机自动化识别、自动配准和分割问题。

The study on digitalization and visualization of STN and AM in individual brain atlas based on stereotactic MRI

Objective To evaluate the feasibility of digitalization and visualization of 3D reconstruction of individual brain atlas based on stereotactic MR images in order to explore approach for the individual therapy of functional neurosurgery. Methods The whole brain of one adult male was scanned in the standard stereotactic space with 1mm slice thickness using a 1.5 tesla MRI system, and then subthalamic nulcleus ( STN ) and amygdaloid nucleus ( AM ) were identified, segmented, extracted, saved, and reconstructed in 3D. Results The surfaces of 3D reconstruction of STN and AM based on the MRI were smooth and their shapes were clearly delineated. The morphological feature of STN and AM on the individual brain could be visualized directly in 3D. 3D reconstruction of STN and AM could be rotated, zoomed, and changed color arbitrarily. The anteroposterior diameter of the STN nucleus was longer than the diameters of upper-down and left-right in 3D. The 3D reconstruction of STN manifested typical structure of "dual lens" , and AM nucleus displayed the typical anatomical characteristics of "amygdaloid". Conclusion The 3D reconstruction of individual cranial structure is favorable for pretreatment localization. The digitalization and the visualization of individual brain atlas based on MRI are feasible, but software for automatic segmentation, extraction, registration of MRI images need to be developed, because of semi-automatic or manual data acquisition methods are not sufficient to analyses and deal with a great quantity of imaging data.