根据脑内深部髓质静脉低对比度、管径细小、偏水平分布、受干扰严重等特征,提出一种基于海赛矩阵的脑内深部髓质静脉主干提取方法,该方法通过固定尺度海赛矩阵的特征值构造出一种筛选器初步筛选出静脉主干,并从形态学上连接断开的主干、剔除伪主干。用手动分割金标准进行对照实验,实验结果表明本方法与金标准相比达到一个较高的重合度,但是主干末梢与背景融合处没能很好地处理。
According to the characteristics of DMVs,including low contrast,small diameter,horizontal distribution and severely disrupted,this paper presented a novel DMVs trunk extraction method based on the Hessian matrix. Firstly,it extracted the trunk by a preliminary screening filter constructed by the eigenvalues of a fixed scale Hessian matrix,and then connected broken trunk and eliminated artifacts morphologically. Experimental results show that the proposed method extracts the DMVs trunk quickly and effectively,and the result has a high similarity with manual segmentation. But the mixed place of trunk endings and the background can not be handled well.