Texture is an image property which is difficult to grasp. It can be described as a "homogeneous visual pattern"l. but there exists no formal definition of texture. Intuitively people can discriminate between different textures. referring to visual clues like coarseness, orientation. periodicity, and regularity. Using such concepts, several authors have tried to quantify these aspects of texture'. However. texture encompasses more than these more or less random aspects to which the human eye is sensitive. Therefore, the majority of texture analysis algorithms is based on an image model. in which certain characteristics of the image texture are condensed. Using this image model, texture features can be derived, most of which cannot be related to visual image features. Texture analysis methods are able, in contrast to a human observer, to quantIfy textures objectively. Therefore. texture features can be used for the purpose of characterization, discrimination, and segmentation of textures in. for example, aerial and satellite imagery. Most texture analysis methods have been developed and tested on textures from the collection of texture images in Brodatz' before putting them mto use in a more realistic environment. Since the early seventies, texture analysis methods have also been applied In medical images. For example, Sulton et a!. tried to categorize different stages of pulmonary disease in radiographs4 Since then, the field of application of texture analysis methods in radiology has expanded from chest radiographs to mammograms and bone radiographs. The goal of our study is twofold: in the first place to assess the suitability of different texture analysis methods for usc in radiographs, secondly to select or develop texture features which are able to quantify the changes in the radiographic trabecular pattern occurring in osteoporosis. Osteoporosis is defined as "a disease characterized by low bone mass and microarchitectural changes of bone tissue, leading to enhanced bone fragility and a consequent increase in fracture risk." (WHO, 1994)5

, , , , ,
KPN Research
E.S. Gelsema , H.E. Schütte (Henri)
Erasmus University Rotterdam
Erasmus MC: University Medical Center Rotterdam

Veenland, J. (1999, January 13). Texture analysis of the radiographic trabecular bone pattern in osteoporosis. Retrieved from http://hdl.handle.net/1765/19723