A kind of segmentation method of linear structures from medical images
Yajun Yang, Xiufen Ye · 2016
In many medical imagery applications, extracting linear structures, such as blood vessels or dendrites, from medical images is crucial. Since the extraction of linear structures from medical images using global threshold algorithm accompanies with the problem of linear discontinuities. We propose a connected domain based segmentation method of linear structures from medical images, which takes full advantage of the linear or ridge smoothing and enhancement characteristics of steerable filter, as well as the better linear or ridge structure connectivity when processing images with smooth boundary using local threshold segmentation method. The corresponding experimental results show that the methods proposed in this paper can guarantee the linear connection characteristics well with inhibited noise greatly.