基于模糊K-近邻规则的多谱磁共振脑图像分割方法的研究

STUDY ON SEGMENTATION METHOD FOR MULTI-SPECTRUM HEAD MRI BASED ON FUZZY K-NEAREST NEIGHBOR RULE

本文在K 近邻 (K nearestneighbor ,简称KNN)规则的基础上 ,基于模糊C 均值聚类 (FuzzyC meansclustering ,简称FCM)技术 ,提出了模糊K 近邻算法 (FuzzyK nearestneighbor ,简称FKNN) ,并利用该算法对磁共振脑图像进行分割研究。首先对磁共振颅脑图像进行预分割 ,剔除颅骨和肌肉等非脑组织 ,只保留大脑结构 ;然后利用FKNN算法对大脑结构进行分割 ,从脑组织中分别提取出白质、灰质和脑脊液。实验结果表明 ,FKNN方法能有效地从大脑结构中分割出白质、灰质和脑脊液 ,分割效果明显优于KNN方法。

This paper presented fuzzy k nearest neighbor(FKNN, for short) algorithm based on k nearest neighbor(KNN) rule and fuzzy c means clustering(FCM) technique, and the study on multi spectrum brain magnetic resonance images(MRI) segmentation using FKNN algorithm. The non brain tissues such as muscle and skull were removed first with image pre segmentation while the brain structure remained; then, the white matter(WM), gray matter(GM) and cerebral spinal fluid(CSF) could be separated from the brain structure using FKNN rule. The results showed that the WM, GM and CSF could be segmented better from the brain structure using FKNN than using KNN.