东北大学学报:自然科学版 ›› 2018, Vol. 39 ›› Issue (9): 1226-1231.DOI: 10.12068/j.issn.1005-3026.2018.09.003

• 信息与控制 • 上一篇    下一篇

基于相似性度量的肺结节图像检索算法

魏国辉1,2, 齐守良1, 钱唯1, 张魁星2   

  1. (1. 东北大学 中荷生物医学与信息工程学院, 辽宁 沈阳110169; 2. 山东中医药大学 理工学院, 山东 济南250355)
  • 收稿日期:2017-05-08 修回日期:2017-05-08 出版日期:2018-09-15 发布日期:2018-09-12
  • 通讯作者: 魏国辉
  • 作者简介:魏国辉(1983-),男,山东广饶人,东北大学博士研究生; 钱 唯(1955-),男,美籍华人,东北大学教授,博士生导师.冯明杰(1971-), 男, 河南禹州人, 东北大学副教授; 王恩刚(1962-), 男, 辽宁沈阳人, 东北大学教授,博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(61672146,81671773).国家自然科学基金资助项目(51171041).

Image Retrieval Algorithm of Pulmonary Nodules Based on Similarity Measurement

WEI Guo-hui1,2, QI Shou-liang1, QIAN Wei1, ZHANG Kui-xing 2   

  1. 1. School of Sino-Dutch Biomedical & Information Engineering, Northeastern University, Shenyang 110169, China; 2. School of Science and Engineering, Shandong University of Traditional Chinese Medicine, Jinan 250355, China.
  • Received:2017-05-08 Revised:2017-05-08 Online:2018-09-15 Published:2018-09-12
  • Contact: QI Shou-liang
  • About author:-
  • Supported by:
    -

摘要: 为了克服肺部病变CT表现复杂,极易造成医生误诊的缺点,提出了一种基于相似性度量的医学图像检索算法并用于肺癌的诊断研究,该相似性度量保持了图像的语义相关和视觉相似.首先,根据相似性度量理论构建距离度量学习算法学习一个马氏距离;然后,根据学习的马氏距离度量,提出新的医学图像检索算法,并将提出的算法应用于肺癌的诊断研究.实验结果证明了该检索算法在肺癌诊断应用中的可行性和有效性.

关键词: 医学图像检索, 肺癌, 相似性度量, 距离度量学习, 纹理特征

Abstract: In order to overcome the shortcomings that CT of pulmonary lesions is complex and is very easy to lead to misdiagnosis, a medical image retrieval algorithm based on similarity measurement was proposed to diagnose lung cancer. The similarity measurement maintains the semantic relevance and visual similarity of the image. Firstly, a distance metric learning algorithm was constructed to learn a Mahalanobis distance on the basis of the proposed similarity measurement. Secondly, a novel medical image retrieval algorithm was proposed based on the learned distance metric to diagnose lung cancer. The study results demonstrate the feasibility and effectiveness of the proposed retrieval algorithm in lung cancer diagnosis.

Key words: medical image retrieval, lung cancer, similarity measurement, distance metric learning, texture features

中图分类号: