Physiol. Meas. 17 (1996) A179–A186. Printed in the UK
Imaging of physiologically evoked responses by electrical
impedance tomography with cortical electrodes in the
anaesthetized rabbit
D S Holder, A Rao and Y Hanquan
Department of Physiology, University College, Gower Street, London WC1E 6BT, UK
Abstract. The purpose of this study was to determine if electrical impedance tomography (EIT)
could be used to image impedance changes of several per cent over tens of seconds, known
to occur during evoked activity of the cerebral cortex. A ring of 16 electrodes was placed on
the exposed superior surface of the brain of anaesthetized rabbits. EIT images were acquired
every 15 s using a Sheffield Mark 1 EIT system. During periods of 2.5–3 min of intense photic
stimulation of both eyes or electrical stimulation of a forepaw, reproducible impedance decreases
of 4:52:7% and 2:72:4% (meanSD) respectively occurred in appropriate cortical areas, with
a time course similar to the period of stimulation. They were accompanied by adjacent smaller
impedance increases. The decreases are probably due to increased blood flow and temperature;
the cause of the adjacent increases may be a shadowing artefact of the reconstruction algorithm
or due to physiological shrinkage of the extracellular space. This demonstrates, for the first time,
that such small changes may be imaged under optimal conditions. These results are encouraging
to the prospect that EIT may eventually be suitable for imaging evoked responses or epilepsy
in human subjects.
1. Introduction
Electrical impedance tomography (EIT) has the potential for imaging a variety of different
processes in the brain, at frequencies of some tens of kHz used currently by single-
frequency imaging systems. These include impedance changes of tens of per cent during
cerebral ischaemia or cortical spreading depression, of several per cent during epilepsy or
physiologically evoked responses over tens of seconds, or even much smaller changes of
about 0.01% over milliseconds during action potential related activity in the cortex (Boone
1995; see Holder 1993a for a review). Until recently, the evidence for such changes was all
from single-channel impedance measurements made with cortical or intracerebral electrodes.
Recently, EIT images of the largest of these changes—cerebral ischaemia (Holder 1992)
and cortical spreading depression (Boone et al 1994)—have been obtained using a ring of
cortical electrodes placed on exposed cerebral cortex.
The mechanism of these changes is well established in the case of both the large changes,
namely cell swelling and consequent decrease in the extracellular space (Hansen and
Olsen 1980), and opening of voltage-sensitive ion channels during neuronal depolarization
(Boone 1995). The nature and causes of impedance changes during physiologically evoked
responses is far less clear. Cerebral impedance decreases of several per cent, lasting
some tens of seconds, have been observed during physiologically evoked activity, such
as exposure to milk or of a female to a male cat (Adey et al 1962) or direct electrical
stimulation (Aladjalova 1964). To our knowledge, no measurements of cerebral impedance
have been made during repetitive evoked responses due to stimulation of the visual, auditory
0967-3334/96/SA0179+08$19.50 c© 1996 IOP Publishing Ltd A179

A180 D S Holder et al
or somatosensory systems. Possible mechanisms include mild degrees of cell swelling with a
consequent decrease in the extracellular space, causing an impedance increase at frequencies
below 100 kHz or so, or increased blood flow, blood volume or temperature, which will
each cause an impedance decrease.
The eventual use of electrical impedance tomography of the head would preferably be
for imaging of pathological or physiological events in the brain in a non-invasive way, with
the use of scalp electrodes. EIT imaging work so far has been restricted to measurement
directly on the cortex because existing reconstruction algorithms cannot produce adequate
images of the brain when scalp electrodes are used, because current is diverted by the highly
resistive skull (Holder 1993b). We are currently designing and testing a portable system
with flexible drive configuration and specially adapted reconstruction algorithm which may
overcome these problems (Cusick et al 1994, Bayford et al 1996).
The purpose of this study was to see if reproducible EIT images could be produced
during physiologically evoked responses. Specifically, we intended to establish if
reproducible impedance changes occurred, if they could be distinguished from baseline
variability, and if they led to reproducible EIT images. Although the eventual intended
application is for imaging through the skull with scalp electrodes, this study was intended
as a first step. Measurements were therefore made under what seemed likely to be optimal
conditions: a ring of 16 spring-mounted electrodes was placed directly on the exposed
cortical surface in anaesthetized rabbits and a well established EIT system, the ‘Sheffield
Mark 1’ (Brown and Seagar 1987), was used, which operates at a single frequency of
50 kHz. Difference images were produced by taking a reference image at rest and comparing
further images taken every 15 s over a period of intense continual stimulation lasting about
3 min. An advantage of the preparation chosen is that the exposed superior surface of
the rabbit brain is almost flat. As impedance changes may be expected to occur mainly
in the cerebral cortex, the anatomical arrangement is therefore largely two-dimensional. It
therefore seemed to be tractable from an imaging point of view, as the available Sheffield
reconstruction algorithm assumes that the imaged object is two-dimensional.
The findings are that it was possible to image the relatively small impedance changes
which occurred during evoked activity; they were a similar size to those expected from
previous single-channel studies in the literature. The direction of the predominant impedance
change was a decrease in the area expected on physiological grounds. A surprising finding
was that each area of decrease was usually accompanied by an adjacent area of impedance
increase. Possible reasons are discussed in the final section.
2. Methods
Eight male adult New Zealand White rabbits (3.5–4 kg) were anaesthetized with 0.5–1.0%
halothane and in a 70%/30% nitrous oxide/oxygen mixture after induction with 0.3 ml kg−1
intramuscular Hypnorm (Fentanyl/Fluanisone, Janssen Animal Health) and 2 mg kg−1
subcutaneous diazepam. Animals were paralysed with pancuronium bromide and artificially
ventilated. The superior surface of the cerebral cortex was exposed by making a circular
craniotomy, 20 mm in diameter; the dura mater was cut away to reveal the pia mater. A
circular array of 16 Ag/AgCl spring-mounted ball electrodes, each about 1 mm in diameter
(figure 1), was placed on the surgically exposed dorsal surface of the cortex (covering most
of the occipital, parietal and frontal cortex). Care was taken to stop any bleeding from
the bone or scalp, and exposed cortex and electrodes were covered with liquid paraffin
to prevent drying. Common-mode, evoked response reference, and earth electrodes were
inserted subcutaneously into the neck. The earth electrode was connected to earth via a