上海大学学报(自然科学版) ›› 2012, Vol. 18 ›› Issue (3): 240-245.doi: 10.3969/j.issn.1007-2861.2012.03.005

• 论文 • 上一篇    下一篇

基于分数阶Fourier变换的乳腺肿瘤纹理特征提取

钟瑾1,施俊1,周世崇2,胡文婕1   

  1. (1.上海大学 通信与信息工程学院,上海 200072;2.复旦大学附属肿瘤医院 超声科,上海 200032)
  • 出版日期:2012-06-30 发布日期:2012-06-30
  • 通讯作者: 施俊(1977~),男,副教授,博士,研究方向为医学图像处理、医学信号处理. E-mail:junshi@staff.shu.edu.cn
  • 基金资助:

    上海市教委科研创新基金资助项目(09YZ15);上海市教委重点学科建设资助项目(J50104)

Extraction of Texture Features of Breast Tumor Based on  Fractional Fourier Transform

ZHONG Jin1,SHI Jun1,ZHOU Shi-chong2,HU Wen-jie1   

  1. (1. School of Communication and Information Engineering, Shanghai University, Shanghai 200072, China;
    2. Department of Ultrasound, Cancer Hospital of Fudan University, Shanghai 200032, China)
  • Online:2012-06-30 Published:2012-06-30

摘要: 分数阶Fourier变换(fractional Fourier transform, FrFT)是传统的Fourier变换在分数级次上的推广,它同时包含了信号时频域的信息.对乳腺肿瘤超声图像进行分数阶Fourier变换,提取分数阶Fourier变换域相位信息的统计纹理特征,并结合AdaBoost分类算法,实现乳腺肿瘤的良恶性分类.实验结果表明,基于分数阶Fourier变换提取的纹理特征能较好地应用于乳腺肿瘤的良恶性分类,说明该方法可用于乳腺肿瘤的计算机辅助诊断.

关键词: AdaBoost, 超声图像, 分数阶Fourier变换, 乳腺肿瘤, 纹理特征

Abstract: Fractional Fourier transform (FrFT) is a generalized form of the traditional Fourier transform, which includes information of both the time domain and the frequency domain. FrFT is used to extract phasebased features of breast tumor from ultrasound images. These features are used as input to the AdaBoost algorithm to distinguish benign tumors from malignant tumors. Experimental results indicate that the texture features extracted with the proposed FrFTbased algorithm can well classify breast tumor, and can be used in computeraided breast diagnosis.

Key words: AdaBoost, breast tumor, fractional Fourier transform (FrFT), texture feature, ultrasound image

中图分类号: