An Effective Edge Detection Methodology for Medical Images Based on Texture Discrimination

Abstract

As medical images are fuzzy, edge detection based on texture characteristics is comparatively effective than intensity based techniques. A new methodology is described for texture edge detection in medical images that is applicable across modalities. We use a multi-scale filter to capture texture edge information. An experimental prototype based on the proposed methodology provides a test bed for comparison with a popular edge detection technique.

Texture edge statistics for efficient retrieval of biomedical images

“…We present a texture edge based generic workflow for deriving edge based features from medical images. Its application for image structure based retrieval spans across medical imaging modalities due to high relevance of texture content within medical images [3]. The steps involved in the workflow are preprocessing, texture edge detection followed by edge based feature extraction (see figure 1).…”

“…The method used here for detection of texture edges is motivated from our earlier work [3] using 2-D Gabor filters. Application of Gabor filters to describe the global textural properties in medical images is well known [7].…”

“…Both these methods primarily focus on using intensity gradients for describing image structure. The texture edges are known to represent the tissue boundaries better than intensity edges in medical images [3]. Therefore it is imperative to research on the use of structural information pertaining to texture edges for describing medical images.…”

Figure Retrieval in Biomedical Literature

“…Biomedical images are used to capture the structural/functional aspects of human body in the form of image or measurements. Structures observed in medical images have relatively fuzzy boundary [6] when compared to other three classes that mainly comprise fairly definite structures due to distinct shape of geometrical constructs used in them. It is evident that biomedical figures in all the four classes exhibit distinct structural behavior.…”

A model-based edge estimation method with increased edge localization accuracy for medical images

“…These methods, which produce good results, require the regulation of some parameters based on trial and error. Other methods try to detect the edges either by modeling the morphological structure of the object or recognizing the tissues that belong to the object [7,8]. There are also methods that use a genetic algorithm or artificial intelligence [9].…”