Elsevier

Clinical Radiology

Volume 59, Issue 12, December 2004, Pages 1061-1069
Clinical Radiology

Review
Texture analysis of medical images

https://doi.org/10.1016/j.crad.2004.07.008Get rights and content

The analysis of texture parameters is a useful way of increasing the information obtainable from medical images. It is an ongoing field of research, with applications ranging from the segmentation of specific anatomical structures and the detection of lesions, to differentiation between pathological and healthy tissue in different organs. Texture analysis uses radiological images obtained in routine diagnostic practice, but involves an ensemble of mathematical computations performed with the data contained within the images. In this article we clarify the principles of texture analysis and give examples of its applications, reviewing studies of the technique.

Introduction

The texture of images refers to the appearance, structure and arrangement of the parts of an object within the image. Images used for diagnostic purposes in clinical practice are digital. A two-dimensional digital image is composed of little rectangular blocks or pixels (picture elements), and a three-dimensional digital image is composed of little volume blocks called voxels (volume elements); each is represented by a set of coordinates in space, and each has a value, representing the grey-level intensity of that picture or volume element in space. Since most medical images are two-dimensional we will restrict the discussion to pixels, bearing in mind that the extension to voxels and volumetric images is straightforward.

We may attribute the texture concept in a digital image to the distribution of grey-level values among the pixels of a given region of interest in the image. One way of depicting this is to display the digital data as a three-dimensional map based on the pixel values, as shown in Fig. 1. Thus, texture analysis is in principle a technique for evaluating the position and intensity of signal features, i.e. pixels, and their grey-level intensity in digital images. Texture features are, in fact, mathematical parameters computed from the distribution of pixels, which characterize the texture type and thus the underlying structure of the objects shown in the image.

According to the methods employed to evaluate the inter-relationships of the pixels, the forms of texture analyses are categorized as structural, model-based, statistical and transform methods.1

This represents texture by the use of well-defined primitives. In other words, a square object is represented in terms of the straight lines or primitives that form its border. The advantage of these methods are that they provide a good symbolic description of the image. On the other hand, it is better for the synthesis of an image than for its analysis. The theory of mathematical morphology3 is a powerful tool for structural analysis.

Here an attempt is made to represent texture in an image using sophisticated mathematical models (such as fractal or stochastic). The model parameters are estimated and used for the image analysis. The disadvantage is the computational complexity involved in the estimation of these parameters.

These are based on representations of texture using properties governing the distribution and relationships of grey-level values in the image. These methods normally achieve higher discrimination indexes than the structural or transform methods.

The texture properties of the image may be analyzed in a different space, such as the frequency or the scale space. These methods are based on the Fourier,4 Gabor5 or Wavelet transform.6 The Wavelet transform is the most widely used because of the ease with which it may be adjusted to the problem in question.

Section snippets

Texture parameters

Medical images possess a vast amount of texture information relevant to clinical practice. For example, current magnetic resonance (MR) images of tissues are not capable of providing microscopic information that can be assessed visually. However, histological alterations present in some illnesses may bring about texture changes in the MR image that are amenable to quantification through texture analysis. This has been successfully applied to the classification of pathological tissues from the

Important considerations

The parameters described above give an idea of the type of information that texture analysis may produce from an image, depending on which texture parameters provide the information sought. Most applications use texture measures as a way of classifying regions of interest in images, for example to differentiate between healthy and pathological tissue, or in order to separate different anatomical structures. Therefore, the procedure generally adopted is to compute a large set of texture

Applications

Texture analysis may be applied in a series of studies of medical images. One application is the segmentation of a given anatomical structure, based on the texture characteristics of the structure. However, texture analysis is most important for those cases in which change cannot be detected by direct inspection of the image. For example, in some conditions the tissue of associated anatomical structures suffers alterations. These can normally be detected by histological examination, but

Conclusions

We have described here the technique of texture analysis in medical images. Texture parameters are simply a mathematical representation of image features that can be characterized in words as smooth, rough, grainy and so on. This implies that in principle, texture analysis may be applied to any set of image regions that may be differentiated by such description.

In outlining the main categories of texture parameters, and the several uses of each technique, MRI applications have been emphasized

Acknowledgements

This work was partially supported by a grant from Fundação de Ampara à Pesquisa do Estado de São Paulo (FAPESP) (Proc. N. 02/00275-5).

References (26)

  • E.R. Dougherty et al.

    Hands-on morphological image processing

    (2003)
  • R. Bracewell

    The Fourier transform and its applications

    (1999)
  • S. Qian et al.

    Discrete Gabor transform

    IEEE Trans Signal Process

    (1993)
  • Cited by (871)

    View all citing articles on Scopus
    View full text