Dawit Assefa, Harald Keller, and David A. Jaffray
Magnetic resonance imaging, medical image processing, multi-parametric, quaternion, trinion
This study presents a novel colour transformation scheme for use in processing multi-parametric magnetic resonance images (MRI) expressed as two-dimensional multi-component (colour) signals in higher order algebraic spaces. The traditionally used monochromatic analysis of such images has a main drawback in that it misses the information as to how the different parameters are correlated to each other. In the current work, multiple MR parameters are combined to one entity and analysed holistically using higher order Fourier transforms. The scheme is a hybrid approach that incorporates other image processing techniques such as principal component analysis (PCA) and ways of extracting statistical features. The approach has been implemented in terms of identification of brain tumours in MRI taken from a cohort of patients treated for glioblastoma multiforme. Our results show that the idea of combining the information from different MRI techniques goes much beyond what can be achieved by using any single parameter, thus allowing an improved image processing and pattern recognition work.
Important Links:
Go Back
