Echo-planar imaging (EPI) is a fast nuclear magnetic resonance imaging (MRI
) method. Unfortunately, local magnetic field inhomogeneities induced mainl
y by the subject's presence cause significant geometrical distortion, predo
minantly along the phase-encoding direction, which must be undone to allow
for meaningful further processing. So far, this aspect has been too often n
eglected,
In this paper, we suggest a new approach using an algorithm specifically de
veloped for the automatic registration of distorted EPI images with corresp
onding anatomically correct MRI images. We model the deformation field with
splines, which gives us a great deal of flexibility, while comprising the
affine transform as a special case. The registration criterion is least squ
ares. Interestingly, the complexity of its evaluation does not depend on th
e resolution of the control grid. The spline model gives us good accuracy t
hanks to its high approximation order. The short support of splines leads t
o a fast algorithm, A multiresolution approach yields robustness and additi
onal speedup,
The algorithm was tested on real as well as synthetic data, and the results
were compared with a manual method, A wavelet-based Sobolev-type random de
formation generator was developed for testing purposes. ii blind test indic
ates that the proposed automatic method is faster, more reliable, and more
precise than the manual one.