使用图像分割技术可以找到图像中的目标区域,目标区域可以定义为具有特定值的单个区域,也可以定义为具有相同值的多个区域。目前图像分割已经融入到生活中的方方面面,在遥感领域,它应用于航拍图中的地形、地貌的分割;在医学图像领域,它可以用来帮助医生评估后续病情、计划手术等。医学图像分割技术在医学领域有着不可小觑的作用,它可以使病理结构的变化在图片上得到更清晰的展示,有利于提高临床医生的诊断效率和准确率。
在早期的医学诊断中,医学细胞图像首先通过显微镜或者其他光学仪器获 取,再通过病理学专家凭借个人经验进行定性标注分析,这种人工诊断方式 主要有两个不足,一是容易受到个人主观因素影响,二是针对复杂的医学图像 诊断时间长,消耗医生大量的精力。早期的医学分割方法有机器学习、边缘检 测和模版匹配等。这些方法虽然取得了一定的进展,但是医学图像在提取特征 时经常会出现噪声、对比度低、模糊等问题,因此它比普通原彩图像提取特征 更困难,而且人工干预设计特征还存在着一定的局限性。不仅如此,对于医学 细胞图像来说,细胞染色不均、密度较高的细胞群、微生物杂质等都会导致无 法准确地分割出目标细胞。
鉴于此,提出一种基于区域椭圆拟合的细胞分割方法,运行环境为MATLAB R2018A,压缩包=数据+代码+参考文献,部分代码如下:
close all;
clear all;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Method Selection
%Set METHOD = 0 to only test the Segmentation Stage,
%Set METHOD = 1 to run DEFA method,
%else you run EMAR method
%Set METHODSEG = 1 to run OTSU method
%Set METHODSEG = 2 to run Adaptive Thresh method,
%Set METHODSEG = 3 to run Adaptive Thresh+extra method ,
%Set METHODSEG = 4 to run the proposed ICIP 2018 method
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
METHOD = 1;
METHODSEG = 4;
AICBIC_SELECTION = 1; %Set AICBIC_SELECTION = 1, to use AIC is selected else BIC is used
set(0,'DefaultFigureColormap',jet);
DataDirD{1} = 'Dataset_NIH3T3//'; %NIH3T3 nucleus dataset
DataDirD{2} = 'Dataset_U20S//'; %U20S cell images
ResultsDirD{1} = 'RES_NIH3T3//';
ResultsDirD{2} = 'RES_U20S//';
filesD{1} = ['dna-0-0 dna-1-0 dna-10-0 dna-11-0 dna-12-0 dna-13-0 dna-14-0 dna-15-0 dna-16-0 dna-17-0 dna-18-0 dna-19-0 dna-2-0 dna-20-0 dna-21-0 dna-22-0 dna-23-0 dna-24-0 dna-26-0 dna-27-0 dna-28-0 dna-29-0 dna-3-0 dna-30-0 dna-31-0 dna-32-0 dna-33-0 dna-34-0 dna-35-0 dna-36-0 dna-37-0 dna-38-0 dna-39-0 dna-4-0 dna-40-0 dna-41-0 dna-42-0 dna-43-0 dna-44-0 dna-45-0 dna-46-0 dna-47-0 dna-48-0 dna-49-0 dna-5-0 dna-6-0 dna-7-0 dna-8-0 dna-9-0 '];
filesD{2} = ['dna-0-0 dna-1-0 dna-10-0 dna-11-0 dna-12-0 dna-13-0 dna-14-0 dna-15-0 dna-16-0 dna-17-0 dna-18-0 dna-19-0 dna-2-0 dna-20-0 dna-21-0 dna-22-0 dna-23-0 dna-24-0 dna-25-0 dna-26-0 dna-27-0 dna-28-0 dna-29-0 dna-3-0 dna-30-0 dna-32-0 dna-33-0 dna-34-0 dna-35-0 dna-36-0 dna-37-0 dna-38-0 dna-39-0 dna-4-0 dna-40-0 dna-41-0 dna-42-0 dna-44-0 dna-45-0 dna-46-0 dna-47-0 dna-48-0 dna-49-0 dna-5-0 dna-6-0 dna-7-0 dna-8-0 dna-9-0 '];
出图如下:
完整代码:MATLAB环境下基于区域椭圆拟合的细胞分割方法
工学博士,担任《Mechanical System and Signal Processing》审稿专家,担任《中国电机工程学报》优秀审稿专家,《控制与决策》,《系统工程与电子技术》,《电力系统保护与控制》,《宇航学报》等EI期刊审稿专家。
擅长领域:现代信号处理,机器学习,深度学习,数字孪生,时间序列分析,设备缺陷检测、设备异常检测、设备智能故障诊断与健康管理PHM等。