DBS术前图像与ICBM-152图谱的配准算法

Registration Algorithm of DBS Preoperative Image and ICBM-152 Atlas

脑深部电刺激(Deep Brain Stimulation,DBS)手术治疗是帕金森病患者重要治疗手段。脑图谱与术前图像之间的配准是实现DBS术前导航和核团识别的一种有效方法。本文对术前图像与ICBM-152图谱的配准问题进行了研究,首先将术前临床图像标准化至图谱的坐标空间,随后使用分段线性配准对图谱进行粗配准,最后利用互信息作为相似性测度,并基于B样条弹性形变模型实现脑图谱与术前数据的非刚性配准。上述方法可以为每个病人计算出个性化的脑图谱,实验结果表明,在原始图像与图谱的互信息测度为0.558的情况下,该方法使二者的互信息提高至1.217。

Deep brain stimulation(DBS)surgical treatment is an important treatment for patients with Parkinson’s disease.One of the effective ways of preoperative navigation and nuclei identification for DBS is to perform registration between brain atlases and preoperative images.In this paper,we initially implemented the registration of preoperative images and ICBM-152 atlas.We firstly normalized the preoperative clinical image to the coordinate space of the atlas,and then used piecewise linear registration to coarsely map the spectrum,at last,non-rigid registration of the brain atlas with preoperative data was achieved using mutual information as a measure of similarity.Thereby a personalized brain atlas could be calculated for each patient,the experimental results showed that the mutual information between the image and the atlas was 0.558 in the initial stage of the experiment,while it increased to 1.217 when the proposed method was employed.