中文摘要：

目的探索适用于阿尔茨海默病（Alzheimer’s disease，AD）患者MR图像脑组织的分割的方法。方法结合阿尔茨海默病患者MR图像中组织区域和边缘的特性对传统水平集进行改进，利用同态滤波对图像进行偏差场修正，增加了UNSHARP MASK处理方法，有效避免了水平集边界泄漏问题。结果标准体膜和真实数据实验证实，该改进算法分割结果优于SPM5。结论利用修正偏差场和添加UNSHARP MASK方法有可能提高AD患者MR图像脑组织分割的准确性和鲁棒性，本研究为MR图像脑组织的精确分割和进一步准确测量作了有益探索。

英文摘要：

Purpose To explore a medical image segmentation method based on the brain tissue segmentation in MR image of Alzheimer＇s disease patients. Methods According to brain region and margin features of patients with AD, the traditional level set was improved. Applying unsharp mask process and bias field correction with .homorphic filter can avoid edge leakage effectively. Results Experiments for the segmentation on phantom and real data by improved algorithm, showed the validity and accuracy in image segmentation. The results were better than those of SPM5. Conclusions The developed level set showed the validity and accuracy in image segmentation. This method is useful for Alzheimer＇s disease patients＇ MR image analysis and measurement on brain tissues in future.