摘要
有效地实现超声图像的分割依然是临床疾病诊断亟待解决的一个难题。本研究将图像树型框架小波变换、尺度共生矩阵、KL变换主分量分析和自组织神经网络聚类相结合应用于医学超声图像,提出一种分割新方法。实验表明,对于不同的医学超声图像,应用本研究方法均可得到比较清晰的分割结果,且显著地提高了分割图像的对比度,这对于固有对比度较低的医学超声图像来说不啻一种很有效的图像分割新方法,为临床诊断提供新的借鉴。
It has been an urgent and tough problem to implement medical ultrasonic image segmentation effectively in clinical disease diagnosis. This paper proposed a new image segmentation method, which integrated the theory of tree- structured frame-wavelet transform, scale co-occurrence matrix （SCM）, principal component analysis and self- organizing neural network, and applied them to the clinical ultrasonic image finishing image segmentation. Experiment results showed that clearer segmented images with a high contrast were obtained with the proposed method.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2006年第6期650-655,共6页
Chinese Journal of Biomedical Engineering
基金
国家自然科学基金资助项目(No.30170259
No.30570475
No.60372081)
教育部博士点基金资助项目(20050141025)
辽宁省科学技术基金资助项目(No.2001101057)。
关键词
超声图像分割
树型框架小波变换
尺度共生矩阵
主分量分析
自组织神经网络
medical ultrasonic image segmentation
tree-structured frame-wavelet transform
scale co-occurrence matrix
principal component analysis
self-organizing neural network
