基于可拓检测的医学图像分割  

Segmentation of Medical Images Based on Extension Detecting Technology

作　　者：刘林[1] 黄英[1] 贺振华[1] 

机构地区：[1]广东工业大学自动化学院,广东广州510006

出　　处：《广东工业大学学报》2013年第3期18-22,共5页Journal of Guangdong University of Technology

基　　金：广州市海珠区科技计划项目(2011-YL-05)

摘　　要：颅内血肿尤其是急性颅内血肿是危害人生命健康的颅内损伤之一.准确分割颅内血肿区域具有重大的临床应用价值.血肿区域医学图像分割技术是实现颅内血肿三维重构及体积计算的关键技术,分割中的难题是如何提高分割精度.本研究建立了颅内血肿医学图像物元模型,提出了将可拓检测物元聚焦与模糊C均值聚类算法相结合的研究方法,解决了模糊C均值聚类算法在颅内血肿医学图像分割中容易陷入局部最优解的问题,从而有效地提高了分割精度.The intracranial hematoma, especially acute intracranial hematoma, is one of intracranial inju- ries which do harm to human life and health, so accurate segmentation of intracranial hematoma area has significant clinical value. Segmentation of hematoma regional medical images is a key technology to real- ize intracraniat hematoma 3D reconstruction and volume calculation, and the problem with segmentation is how to improve the accuracy. It established a matter-element model for medical images of intracranial hematoma, and proposed a research method which combined extension detecting technology of focusing on matter with the fuzzy C-means （FCM） clustering algorithm, which prevents the FCM clustering algorithm from falling into local optimization in segmentation of intracranial hematoma medical images. Therefore, this method effectively improves the accuracy of segmentation.

关 键 词：可拓检测 物元聚焦 模糊C均值聚类 颅内血肿 分割方法 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]