Review of Segmentation for Medical image analysis
成像方法:X射线,CT,MRI,SPECT,PET等
分割的定义: Image segmentation is a procedure for extracting the region of interest (ROI) through
an automatic or semi-automatic process【1】.
应用: border detection in angiograms of coronary冠状动脉血管造影, surgical planning, simulation of surgeries, tumor detection and segmentation肿瘤检测与分割, brain development study, functional mapping, blood cells automated classification, mass detection in mammograms, image registration, heart segmentation and analysis of cardiac images。
分割方法(4类):
1)region-based methods, Here we explain two most popular regionbased approaches: thresholding and region growing。
1,1)阈值法
缺点:没有考虑图像的空间信息,导致噪声敏感
局部阈值法(基于局部的均值方差信息)和Otsu阈值化(找最优全局阈值,极小化类内方差)
局部阈值法(基于局部的均值方差信息)和Otsu阈值化(找最优全局阈值,极小化类内方差)
1.2)区域生长法,一种交互分割方法,会产生hole或不连通区域
2)clustering methods,
2.1)K-means
2.2)Fuzzy c-means
2.3)EM算法
3)classifier methods(模式识别), k近邻(KNN,非参数)和极大似然(参数),缺点没有利用空间信息,训练数据需要人工分割。
4)hybrid methods.
4.1)Gruph cut
4.2)
结果评价:
Dice Similarity Index(DSI)度量自动与人工分割的重叠程度。
实验数据:
参考文献:
【1】Norouzi, A., Rahim, M.S.M., Altameem, A., Saba, T., Rad, A.E., Rehman, A., Uddin, M., 2014. Medical Image Segmentation Methods, Algorithms, and Applications. IETE Technical Review 31, 199-213.