Surgical simulation with computer assisted neurosurgical system

该项目的目的是通过CT或MR图像生成患者大脑的三维实体模型，并使其成为可能进行模拟手术作为术前排练。这将大大提高神经外科手术的准确性和安全性。我们首先开发了半自动图像分割，能够从CT或MR图像中区分解剖结构的区域。传统的解剖结构识别方法是简单地根据图像的黑白水平或像素值、组织的不均匀性或噪声漂移进行区分，但这不利于严格的组织分割。为了解决这个问题，我们分析了感兴趣区域的图像特征，并确定了分割的最佳参数。然后，我们依次对CT或MR图像应用多个图像滤波器获得组织轮廓曲线，这些轮廓曲线由操作员立即检查并存储在图像处理计算机中。利用组织轮廓数据，可以方便地在屏幕上显示患者头部的三维线框图像。其次，我们制作了实体模型。等高线印在与CT或MR切片厚度相同的蜡粘土板上。板沿等高线切割并编译以重建实体模型。最后，我们将实体模型固定在桌面上，用小刮刀刮刀，在神经外科导航器“CANS navigator”下进行手术模拟。这些过程都记录在导航仪上，并在实际操作时重播。

The aim of the project is to produce 3 dimensional solid model of the patient brain from CT or MR images and to make it possible to perform simulation surgery on it as the preoperative rehearsal. This will enhance neurosurgical accuracy and safeness greatly.We first developed the semi-automated image segmentation that enable to discriminate the area of anatomical structures from CT or MR images. Conventionally, anatomical structures were simply discriminated according to the black and white levels or pixel values, tissue inhomogeniety or noise drifts of the images, however, hamper the strict tissue segmentation. To solve this problem, we analyzed the image characteristics of the region of interest and determined optimal parameters for segmentation. Then, we sequentially applied several image filters to the CT or MR images to obtain tissue contour curves, which were checked immediately by the operator and stored in the image processing computer. The 3 dimensional wire-frame image of the patient head was easily displayd on the screen using tissue contour data. Secondly, we produced the solid model. The contour lines are printed on the wax clay plates with the same thickness as CT or MR slices. The plates are cut along the contour line and compiled to reconstruct the solid model. Finally, we fixed the solid model on the table and shaved with a small spatulas to performed surgical simulation under a neurosurgical navigator, "CANS Navigator". The procedures were recorded with the navigator and replayd at the actual operation.