基于改进主动形状模型的前列腺超声图像分割算法

发布时间：2018-03-27 21:39

  本文选题：超声图像分割　切入点：Gabor特征　出处：《东南大学学报(自然科学版)》2017年05期

【摘要】：为了提高前列腺超声图像的分割精度,提出了一种基于改进主动形状模型的前列腺超声图像分割算法.首先,提取前列腺超声图像的特征集合,该特征集合由Gabor纹理特征和局部二值模式(LBP)特征组成.然后,通过利用k均值算法对提取的特征集合进行聚类分析,得到超声图像的聚类表示图.最后,在聚类表示图上应用ASM获取超声图像中前列腺的形状信息.结果表明,该算法可以准确地定位前列腺边界信息,与医生手动标记的前列腺轮廓相比,平均绝对距离仅为1.559 6 mm,戴斯相似度系数最高可达93.88%.利用超声图像的聚类表示图可以获得更加精确的前列腺轮廓信息,可用于海扶高聚焦超声(HIFU)手术中的精准导航.
[Abstract]:In order to improve the accuracy of prostate ultrasound image segmentation, an algorithm based on improved active shape model is proposed. Firstly, the feature set of prostate ultrasound image is extracted. The feature set is composed of Gabor texture feature and local binary pattern feature. Then, by using k-means algorithm to cluster the extracted feature set, the clustering representation diagram of ultrasonic image is obtained. The shape information of prostate in ultrasonic image is obtained by using ASM on the cluster representation map. The results show that the algorithm can accurately locate the boundary information of prostate, compared with the prostatic contour marked manually by doctors. The average absolute distance is only 1.559 mm, and the highest similarity coefficient of Deiss can reach 93.88. By using the clustering representation of ultrasonic images, more accurate information of prostate contour can be obtained, which can be used for accurate navigation of HIFU in HIFU.
【作者单位】： 东南大学计算机科学与工程学院;东南大学计算机网络和信息集成教育部重点实验室;东南大学中法生物医学信息研究中心;
【基金】：国家自然科学基金资助项目(31571001,61201344,61271312,61401085,81530060) 江苏省自然科学基金资助项目(BK2012329,BK2012743,BK20150647,DZXX-031,BY2014127-11)
【分类号】：R737.25;TP391.41
，

本文编号：1673316