摘要
探讨纹理特征分析算法在超声图像定量分析中的应用。提取了正常肝脏与脂肪肝B超图像的近远场平均灰度比特征,以及用于超声医学图像分析的纹理特征模型—灰度共生矩阵和灰度游程矩阵,并用基于假设检验的特征选择方法进行了对比研究。结果表明:近远场灰度比(MIR)、灰度共生矩阵的角二阶矩(ASM)、对比度(CON)、熵(ENT)和反差分矩(IDM)、灰度游程矩阵的灰度分布的不均匀性特征(GLN)6个特征量可以描述肝脏超声图像。说明全面分析超声图像的纹理特征,对B超图像的定量分析有很大帮助。
To disucss the application of the image texture characteristic analysis algorithm in quantitative analysis of ultrasonic medical image Some parameters about texture characteristic of the gray level co-occurrence matrices （GLCM） and the gray level run-length （GLRL） matrices and near far field mean intensity ratio （MIR） from normal and fatty liver ultrasonic medical image were got. Texture characteristic analysis was compared with near far field mean intensity ratio. Result showed that six image features about MIR. ASM. CON. ENT IDM. GLN could describe ultrasonic liver medical image. It proves that the analysis of texture characteristic is available for quantitative analysis of ultrasonic medical image.
出处
《生物医学工程研究》
2007年第2期166-169,共4页
Journal Of Biomedical Engineering Research
关键词
超声图像
脂肪肝
纹理分析
灰度共生矩阵
灰度游程长度
Ultrasonic medical image
Fatty live
Texture characteristic analysis, Gray level co-occurrence matrices, Gray level runlength matrices
