Background field removal using spherical mean value filtering and Tikhonov regularization

Abstract

Purpose: To introduce a new method for removing background artifacts in field maps and apply it to enhance the accuracy of susceptibility mapping.

Methods: A field artifact removal method is introduced that is based on the sophisticated harmonic artifact reduction for phase data (SHARP) method exploiting the harmonic mean value property. The new method uses Tikhonov regularization at the deconvolution stage and is referred to as regularization enabled SHARP (RESHARP). RESHARP was compared with SHARP in a field-forward susceptibility simulation and in human brain experiments, considering effects on both field maps and the resulting susceptibility maps.

Results: From the simulation, RESHARP was able to reduce error in the field map by 17.4% as compared with SHARP, resulting in a more accurate single-angle susceptibility map with 6.5% relative error (compared with 48.5% using SHARP). Using RESHARP in vivo, field and susceptibility maps of the brain displayed fewer artifacts particularly at the brain boundaries, and susceptibility measurements of iron-rich deep gray matter were also more consistent than SHARP across healthy subjects of similar age.

Conclusion: Compared with SHARP, RESHARP removes background field artifact more effectively, leading to more accurate susceptibility measurements in iron-rich deep gray matter.