We introduce a new magnetic field geometry, B-x(x, y) = g(y)y cos(q(x)x), t
o spatially encode magnetic resonance imaging (MRI). The field is called th
e PERL field since it is PERiodic in x and Linear in y. A technique is prop
osed to acquire two-dimensional (2D) data without switching the encoding fi
elds. The time-domain PERL signal and image are not related by a two-dimens
ional Fourier transform (2DFT). They are related by a 1DFT in the x-dimensi
on and a new Bessel function integral transform, called the PERL transform,
in the y-dimension. We numerically solve this equation and develop a recon
struction algorithm. The algorithm is evaluated by assuming a known spin de
nsity which is then used to calculate the PERL signal. This signal is the i
nput for the algorithm. We show the reconstructed image matches the convolu
tion of the initial assumed spin density and a point spread function (PSF).
We identify the PSF and describe the conditions under which it approaches
delta. (C) 1999 John Wiley & Sons, Inc.