基于新的形态学梯度参数的DTI图像分割算法

Segmemtation Algorithm of DTI Image Based on New Morphological Gradient Parameters

为了解决传统DTI图像分割中更细致边缘信息的丢失问题,提出了新的张量形态学梯度参数,并基于张量相似性形态学梯度和各向异性形态学梯度,采用标记的分水岭算法对DTI图像进行分割。通过对人脑胼胝体图像的分割实验表明,利用新参数TMG-l2和TMG-RA能够更加快速、准确地对DTI图像进行细致边缘轮廓的定位和分割,保护了重要分割区域的边缘信息。

To solve the problem of losing detailed edge intormation inherent in traditional DTI image segmentation methods, some new morphological gradient tensor parameters are proposed. Based on the tensor similarity morphological gradients anti tensor morphologieal anisotropie gradients, the tag based on watershed algorithm is applied in DTI image segmentation. The human brain corpus eallosum image segmentation experiments show that this algorithm with the TMG-12 and TMG-RA parameters can quickly and aeeurately locate and segment the outline of the image, and the edge information of the important legion is preserved.