Phase Imaging and Susceptibility Reconstruction

The importance of MR phase imaging has strongly increased during the last years. The investigation of phase data enables detailed insights into microstructure and magnetic properties of tissue, in particular field information and magnetic susceptibility. Phase, field and susceptibility currently evolve into potential future diagnostic tools.

Phase Preparation

The calculation of field maps requires a proper preparation of the measured phase data. The recombination of phase data from multi-channel receive coils is a more delicate procedure than processing only magnitude data and requires careful consideration. Unwrapping of the raw (ambiguous) phase is essential to access the true phase evolution within the sample. Operating in multiple dimensions, this procedure is complicated and implicates high computing demands. We are devloping methods to significantly reduce computing time whilst ensuring high quality of the unwrap result.

Field Maps

Maps of the static magnetic field are derived from the evolution of the unwrapped phase in time, measured with a multiple-echo gradient echo sequence. One focus of our current research is to properly decompose these field maps into (a) a part that represents distortions of external origin and (b) a local tissue-induced component. Identification and removal of the external contributions enables insights into microscopic tissue properties that cannot be imaged by other modalities. Techniques used to achieve this aim are, for example, low-pass filters (such as Gaussian), expansion on spherical harmonics (e.g. the SPHINX algorithm, developed in our group [1]) or our recently developed multistage filter chain, addressing the diverse sources of field distortions in consecutive steps.

Susceptibility Reconstruction

The reconstruction of susceptibility maps is of high interest in medical imaging, allowing for a classification of tissue types complementary to that based on the conventional NMR parameters. Furthermore, pathological changes in the structure of tissue (e.g. MS lesions) can be visualised. An unequivocal interpretation of the images, however, suffers from the dependence of the phase contrast on the orientation of the structures of interest and adjacent geometry with respect to the applied magnetic field. Reconstruction algorithms employ mainly deconvolution-based approaches to estimate the susceptibility distribution responsible for an observed field map (e.g. [2]).
The effect of local susceptibility contrast within the tissue increases with the strength of the static magnetic field. In addition, the linear-to-nearly-quadratic SNR increase with field allows small signal changes to be readily distinguished from the effect of noise. The characterisation of subtle changes within the tissue becomes feasible at high fields such as 9.4T human and animal scanners.

Example for phase imaging workflow: 


The field map of a representative transversal slice from a post mortem brain (upper image) is decomposed into long scale field inhomogeneities (lower left) and local detail (lower right).

• The topmost image shows recombined phase data from an in vivo measurement at 3T
• Phase unwrapping enables calculation of the field map (left)
• Field distortions of external origin are removed (bottom)
• Susceptibility is reconstructed from the local field (right)