MR multi-spectral texture analysis using space-frequency information
Hongmei Zhu, Yunyan Zhang, Xing‐Chang Wei, Luanne M. Metz, Anwesha Law, Joseph Ross Mitchell · 2004
A new way to extract texture information from magnetic resonance (MR) data sets is proposed. The approach is based on the recently developed Stockwell transform (ST), a multi-scale localized extension of the Fourier transform. Since the ST provides a space-frequency representation of an image, this property may be exploited to derive texture images using local spatial frequency content of multi-spectral MR exams of multiple sclerosis (MS) patients. As illustrated, the new information can aid detection of sub-clinical abnormalities in cerebral normal appearing white matter in patients with MS.