Neuroimaging Data Processing/Processing/Steps/Field Map Correction

Neuroimaging Data Processing/Processing/Steps
Coregistration and Normalization Field Map Correction Physiological Noise Regression

Magnetic field inhomogeneities


Tissue differences in the brain result in static field inhomogeneities which can lead to signal distortion in the fast and sensitive EPI sequences used for fMRI. Such distortions become severe in regions where air-filled sinuses border with bone or tissue, such as the frontal lobes and temporal lobes. The result can be geometric distortion of the image or even signal drop-outs, making it difficult to achieve an accurate registration between an activation map calculated from an fMRI time series and an undistorted anatomical image (which is acquired by a less sensitive sequence). Field inhomogeneities can be minimized during the scan through shimming. By adjusting many first-, second-, and higher-order magnetic field gradients generated by shimming coils, field distortions can be corrected. However, it might be necessary or recommendable to account for inhomogeneities that could not be overcome by shimming during preprocessing.[1]

Field mapping


A field map is an image of the intensity of the magnetic field across space. A field map can be obtained by getting two images of the signal phase with slightly different echo times. The difference between the phase images is proportional to the strength of the field at any given location. If the field is completely uniform, then the phase difference induced by the difference echo times will be the same in all voxels, and the resulting image will be a uniform gray. A field map can be acquired as part of the scanning procedure or on a phantom and then used to correct for any geometric distortions.[2]

Bias field correction


If there is no prior knowledge on the magnetic field distribution, intensity variations can be estimated from the collected images themselves. The image is assumed to be a combination of the true data without bias along with the distorting effects of the bias field. Correction algorithms commonly relying on Markov random field models take are used to determine the most likely pattern of distortions, and to reconstruct the assumed trued image by removing the calculated distortions from the biased image. Bias field correction can (or does?) also take in to account tissue segmentation in that different type of tissues are modeled and then the distributions of intensities in different tissues is made equal.[3]



SPM provides two ways for geometric distortion correction. When the distortions occur along anterior-posterior axis and the intrinsic symmetry of the brain is not affected, Realign & Unwarp is designed for such benign distortion. However, if the distortions happen on the other axes, such as the right-left orientation, the FieldMap toolbox and VDM utility should be applied to overcome these severe distortions.


  1. Hutton, Chloe; Bork, Andreas; Josephs, Oliver; Deichmann, Ralf; Ashburner, John; Turner, Robert (2002). "Image Distortion Correction in fMRI: A Quantitative Evaluation". NeuroImage. 16 (1): 217–40. doi:10.1006/nimg.2001.1054. PMID 11969330.
  2. Huettel, S. A., Song, A.W., & McCarthy, G. (2008). Functional Magnetic Resonance Imaging (2nd edition). Sinauer Associates, Inc: Sunderland, Massachusetts, USA.
  3. Guillemaud, R.; Brady, M. (1997). "Estimating the bias field of MR images". IEEE Transactions on Medical Imaging. 16 (3): 238–51. doi:10.1109/42.585758. PMID 9184886.

Further reading