基于线性滤波的脑CT图像颅内结构的提取

Extraction of intracranial structures from cerebral CT-images based on linear filtration

针对现有的提取颅脑CT图像颅腔内结构算法自动性较差的问题,提出了一种新算法。该算法首先根据颅脑颅骨组织宽度大、灰度值高的特点,利用线性空间滤波方法提取颅脑颅骨的粗略轮廓;然后利用图像数学形态学方法和基于阈值的水平左右扫描算法,实现颅内组织的自动化分割;最后,通过对100例颅脑CT检查病例的图像进行实验,来验证算法的可行性。实验结果表明,该算法使99%具有完整颅骨环的颅腔内结构实现计算机自动化分割,分割结果准确。

As the existing extraction algorithms for intracranial structures on cerebral computed tomography have poor automatic ability, a novel algorithm is presented. At first, the skull image from CT is characterized by great width and high gray-level. Therefore, the proposed algorithm uses linear filtration to extract the outline of the skull. Then mathematic morphology and horizontal scanning algorithm are employed for automatic segmentation of intracranial structures on cerebral CT images. At last, the segmentation results of 100 cases of cerebral CT images demonstrate that satisfactory results are achieved in 99% cases with closed skull. The segmentation results are accurate.