甲状腺结节超声图像分割算法研究

Ultrasound image segmentation algorithm for thyroid nodules

针对甲状腺结节超声图像易被噪声污染、对比度低、灰度不均匀等特点,提出基于改进的LIF模型与CV模型相结合的分割算法。针对LIF模型在演化过程中易陷入局部最小值的问题,融入了局部梯度能量信息,从而避免了演化时局部最优的问题;同时结合了CV模型对初始化位置不敏感的优点,从而使得该模型不仅能实现对灰度不均匀图像的分割,而且降低了对初始轮廓位置的敏感性。对比实验结果表明,该算法既能有效克服噪声的影响,又能实现对灰度不均匀图像的精确分割。

As ultrasound images have the characteristics of large noise, low contrast and uneven gray, the segmentation algorithm combing improved local image fitting（LIF） model and Chan Vese（CV） model is proposed in this paper. The traditional LIF model is easy to fall into local minimum during evolution process. In order to avoid the local optimal problem during evolution, this paper introduces global gradient energy information. Combining the advantages of global segmentation of CV model that it is not sensitive to the initial position,the proposed model can not only segment the images with non-uniform gray distribution, but also weaken the sensitivity of the active contour to the initial position. Experimental results show that the proposed algorithm can overcome the influ- ence of noise and realize the segmentation of the uneven gray images.