Electrical impedance tomography is a technology for producing images of int
ernal body structures based upon electrical measurements made from electrod
es on the body surface. Typically a single plane of electrodes is used, see
king to reconstruct a cross-section of the body. Yet the majority of image
reconstruction algorithms ignore the three-dimensional (3D) characteristics
of the current Row in the body. Actually, a substantial amount of current
Rows out of the electrode plane, creating distortions in the resulting imag
es. This paper describes a reconstruction algorithm, ToDLeR, for solving a
linearized 3D inverse problem in impedance imaging. The algorithm models th
e body as a homogeneous cylinder and accounts for the 3D current flow in th
e body by analytically solving for the current Row from one or more layers
of electrodes on the surface of the cylinder. The algorithm was implemented
on the ACT3 real-time imaging system and data were collected From a 3D tes
t phantom using one, two and four layers of electrodes. By using multiple p
lanes of electrodes, improved accuracy in any particular electrode plane wa
s obtained, with decreased sensitivity to out-of-plane objects. A cylindric
al target located vertically more than 8 cm below a single layer of 16 elec
trodes, and positioned radially midway between the centre and the boundary,
produced an image that had 35% of the value obtained when the target was i
n the electrode plane. By adding an additional layer of 16 electrodes below
the first electrode plane, and using 3D current patterns, this artefact wa
s reduced to less than 10% of the peak value. We conclude that the 3D algor
ithm, used with multiple planes of electrodes, reduces the distortions from
out-of-plane structures in the body.