模式识别在医用超声数字图像特征提取中的应用研究

目的:设计超声数字图像特征提取及辅助分析系统,以提高超声图像分析效率和临床诊断效果。方法:基于图像模式识别方法,建立图像预处理、特征提取和辅助分析工具;采用提取图像降噪、增强、均匀化、特征提取及分析等方法,构建标准化超声图像辅助分析流程。结果:超声数字图像特征提取及辅助分析系统实现了标准化超声图像分析流程,构建的图像预处理、特征提取和辅助分析工具,在心脏占位病变鉴别、乳腺组织肿瘤分割、骨骼肌量化评估等场景测试中能够增强图像典型特征显示,有助于提高临床诊断质量及效率。结论:超声数字图像特征提取及辅助分析系统可将模式识别方法引入超声图像分析中,通过图像预处理提取图像宏观特征和感兴趣区域(ROI)特征,构建一种超声图像模式识别工具,对于类似的超声图像分析系统具有参考意义。

Study on the application of pattern recognition in characteristics extraction of medical ultrasound digital images

Objective:To design a system with characteristics extraction and auxiliary analysis of ultrasound digital images so as to improve the analysis efficiency of ultrasound image and the effects of clinical diagnosis.Methods:Based on image pattern recognition method to construct a tool that could implement image preprocessing,feature extraction and auxiliary analysis.And reduced noise of extraction image,enhancement,homogenization,feature extraction and analysis methods were adopted to construct a standardized ultrasound image auxiliary analysis process.Results:This system with characteristics extraction and auxiliary analysis of ultrasound digital images could realize standardized ultrasound image analysis process.And the constructed tool that could implement image preprocess,characteristics extraction and auxiliary analysis could enhance the display of typical characteristics of images in scenario tests included the identification of space occupying lesion of heart,tumor segmentation of mammary tissue,quantitative assessment of skeletal muscle and so on.It contributed to enhance clinically diagnostic quality and efficiency.Conclusion:This system with characteristics extraction and auxiliary analysis of ultrasound digital images can introduce pattern recognition method into ultrasound image analysis.And through image preprocessing to extract gross feature and region of interest(ROI)feature of images,and to construct a pattern recognition tool of ultrasound image,which have reference significance for similarly ultrasound image analysis system.