一种三维规则数据场中复杂组织快速分割方法

Tissue segmentation based on 3D morphology and 3D seed filling

提出了一种三维规则数据场中复杂组织快速分割方法:首先熵阈值二值化三维医学图像,然后用三维形态学腐蚀操作,断开复杂组织与其它组织间的弱连接,并对复杂组织进行连通标记;接着提取出腐蚀后的复杂组织模板,并对此模板进行三维形态学扩张操作,恢复先前三维形态学腐蚀操作消除的部分;最后采用改进的快速三维种子填充算法精确地分割出复杂组织。复杂组织分割实验结果表明了该方法的有效性。

A method based on 3D mathematical morphology theory and 3D seed filling algorithm is present for the segmentation of complicated tissue. Firstly, two thresholds were chosen to make the original 3D regular data set to be binary dataset ; Secondly, the 3 D mathematical morphology erosion operator was applied to cut off the weak linkage relation between the complicated tissue and other tissues, and the adjacent connection relationship of the voxel （volume element） in the complicated tissue was labeled, then the complicated tissue template was extracted. Thirdly, the 3 D mathematical morphology dilation operator was applied to the extracted tissue template to recover the tissue from the erosion template. Finally, an improved effective 3D seed fill algorithm was proposed to obtain the accurate complicated tissue. Experimental results show that the segmentation method is effective.