A comparison of texture quantification techniques based on the Fourier and S transforms

Abstract

Detecting differences in texture has been found to be useful in a variety of medical image analysis applications. One class of methods for texture estimation is based upon analysis of local frequency spectra produced by the S transform. Clinical applications have included detection of multiple sclerosis lesions and identification of brain tumor genotype. This paper describes a software application designed to detect texture differences in medical images, demonstrates and validates the ability of the technique to analyze magnetic resonance images obtained of an in vitro phantom with known textural features, and compares the results to alternative methods. Finally, some examples of texture analysis in several promising biomedical applications are included.