基于内容的医学图像检索

Content-based Medical Image Retrieval

医学图像在临床诊断与治疗中的应用日益广泛,如何利用影像管理系统中大量的图像,辅助医生进行分析与诊断是一个非常重要的问题。传统的基于文本关键字的图像检索方法已不能满足对大型医学图像数据库检索的需要,将基于内容的图像检索方法(CBIR)引入到医学图像数据库中进行研究是一项非常有意义的工作。介绍了基于内容的医学图像检索系统的构成,重点讨论了其中的关键技术问题,包括医学图像分割、特征提取、相似性检索及匹配和相关反馈技术,并分析了国内外的研究现状,对未来发展趋势进行了展望。

Medical Image has been increasingly applied in clinical diagnosis and treatment. It is very important to make use of large numbers of images in medical image management system in order to help clinician to analyze and diagnose. The traditional information retrieval techniques are not fit for retrieving large scale medical image databases. It is a very promising idea to introduce Content-based Image Retrieval（CBIR） technique into indexing medical image databases. The structure of the Content-based Medical Image Retrieval System （CBMIR） is introduced, and the key problems are mainly investigated, which included image segmentation, feature extraction, similarity searching and feedback mechanism. At last, the status and development of CBMIR are discussed.