Texture based classification and segmentation of tissues using DT-CWT feature extraction methods

Dogu Baran Aydogan, Markus Hannula, Tuukka Arola, Prasun Dastidar, Jari Hyttinen
June 2008
Cite DOI

Abstract

In this study, four different dual-tree complex wavelet (DT-CWT) based texture feature extraction methods are developed and compared to segment and classify tissues. Methods that are proposed in this study are based on local energy calculations of sub-bands. Two of the methods use rotation variant texture features and the other two use rotation invariant features. The methods are tested on two texture compositions from the Brodatz texture database and two actual magnetic resonance (MR) images. Results show that there is not a significant difference between using rotation variant or invariant features. On the other hand, for the same Brodatz textures, all DT-CWT based feature extraction methods are competitive with other filtering approaches.

Type
Conference paper
Publication
2008 21st IEEE International Symposium on Computer-Based Medical Systems

Add the full text or supplementary notes for the publication here using Markdown formatting.

Biomedical computing Biomedical imaging Computed tomography Discrete wavelet transforms feature extraction Image databases Image segmentation Magnetic resonance Spatial databases Testing
Dogu Baran Aydogan
Dogu Baran Aydogan
Group leader, Academy Research Fellow

I am interested in computational neuroimaging, connectivity of the brain and brain stimulation