脑梗塞患者MR图像分割的研究

MR Image Segmentation of Cerebral Infarction Patients

目的研究一种可实现脑梗塞患者MR图像脑组织分割的算法。方法根据脑梗塞患者MR图像中脑组织的区域和边缘的特性,对传统水平集算法进行改进,实现了对特定目标体分割的能力,降低了边界泄漏发生的可能性。结果通过体膜和大量脑梗塞患者MR图像实验和SPM5对比,实验证实了改进算法对MR图像分割的准确性和鲁棒性。结论该算法为脑梗塞患者的脑图像分析和脑组织测量提供了一种有效的分割方法。

Purpose A algorithm was presented based on the brain tissue segmentation in MRI of cerebral infarction patients. Methods According to the brain regions and margin features of cerebral infarction patients,the traditional Level Set method was improved with object grey information and Unsharp Mask on MR image. The improved Level Set segmented the specified objects without edge leakage. Results Experiments were completed for the segmentation on Phantom and real MR data by improved algorithm and SPM5. The results showed the validity and accuracy in image segmentation. Conclusions This paper provides a valuable segmentation method for cerebral infarction patients＇ MRI analysis and measurement on brain tissues.