Primary methods and applications of image texture research in computer-aided diagnosis
10.3969/j.issn.1673-8225.2009.39.027
- VernacularTitle:计算机辅助诊断技术中图像纹理研究的主要方法及其应用
- Author:
Chun LIU
;
Tao YANG
;
Juan WANG
;
Lemin TANG
- Publication Type:Journal Article
- From:
Chinese Journal of Tissue Engineering Research
2009;13(39):7721-7727
- CountryChina
- Language:Chinese
-
Abstract:
OBJECTIVE: To analyze recent developments in image texture research both from methodology and from medical image analysis. DATA SOURCE: With the key words of "medical image, texture research, image analysis, application", we retrieved PubMed database (http://www.ncbi.nlm.nih.gov/sites/entrez/), ScienceDirecr database (http://www.sciencedirect.com/) from 1983 to 2009 and CNKI database (http://www.cnki.net/) from 2004 to 2009. DATA SELECTION: Original research thesis, and reviews with clear opinion, sufficient data and reliable conclusion were included. Repetitive studies and studies concerning unrelated to the objective were excluded. MAIN OUTCOME MEASURES: A total 104 literatures were selected, including 10 Chinese literatures and 94 English literatures. These literatures were primarily collected by reading titles and abstracts. A total of 33 literatures with unrelated objective, 18 literatures with repetitive studies were excluded. Finally, 53 Chinese and English literatures were included for further analysis. RESULTS: Primary methods used in texture analysis are structural, statistical, model-based and transform-based-method. When we are interested in identifying texture primitives and their distribution to analyze regular texture, structural approaches are suited. Characteristics of texture like smoothness and coarseness are well analyzed by statistical approaches. Model-based-method is based on the construction of an image model that can be used not only to describe texture, but also to synthesize it. Digital features of texture are got by using some signal processing theories in transformation domain. Texture applications have been widely used in medical imaging domain. CONCLUSION: Because of the specific and complication of medical image and texture, not all texture measure can be used for medical image analysis. One of the development directions of medical image texture research is how to integrate and educe advantages of different methods to fully extract texture features and exactly attribute medical image texture and the relation between its changes and pathological state, resulting in an important component of computer-aided diagnosis.
