Biomedical Engineering Reference
In-Depth Information
inhomogeneous, exacerbating the difficulties in direct imaging of
neural activity using MRI. Nonetheless, theoretical models and
initial experiments have been proposed and carried out, which
have helped shed light on the tantalizing possibility of direct and
noninvasive MRI of neurons in action.
An early MRI study in 1989 by Joy et al.
(7)
provided intrigu-
ing results. The authors used spin echo phase imaging to assess
the magnetic field perturbations induced by externally injected
electrical currents, both in phantoms and in vivo. Their results
showed that electrical currents in biological systems on the order
of milliamperes could be detected by acquiring phase maps during
stimulation. Since neuronal action potentials are essentially elec-
trical depolarizations, these results suggested that it may be possi-
ble to image neuronal activity directly using MRI. It first appeared
that the limitation was simply the low signal-to-noise ratio (SNR),
which could be improved by using time-locked averaging of mul-
tiple trials, as is used in techniques such as event-related potential
(ERP/EEG) and event-related field (ERF/MEG) recordings and
with event-related BOLD fMRI. However, many technical chal-
lenges arose in addition to the SNR limitations, including the
spatially incoherent and temporally transient nature of the neuro-
electric activity, and the multiple confounding synchronized sig-
nals reflecting BOLD, CBV, and CBF changes or physiological
noise. These proved to be extremely difficult to address, and as a
result there were virtually no breakthroughs in the decade follow-
ing the initial demonstration.
Nevertheless, interest was renewed in the late 1990's after a
decade of explosive fMRI research based on the hemodynamic
(e.g., BOLD) contrasts. While researchers across various disci-
plines continue to be deeply attracted to such fMRI methods,
neuroscientists and physicists have relatively quickly reached their
intrinsic spatiotemporal limitations to truly interpret the neural
activities. Several groups have thus recently assessed the feasibil-
ity of using MRI for direct imaging of neuronal activation, more
specifically by attempting to detect the minute magnetic field
changes induced either by electrical currents in phantoms
(8-10)
or by neuronal currents during evoked or spontaneous brain activ-
ity in cell cultures or human subjects
(11-22)
. Despite some
encouraging results, many issues remain controversial. For exam-
ple, several simulations
(10, 15)
and experimental
(14, 17)
stud-
ies have shown that phase images are more sensitive to magnetic
field changes induced by neuronal currents than are magnitude
images, yet others claim the opposite
(12, 20, 21)
.Furthermore,
several attempts at reproducing the positive results obtained in
earlier studies have been unsuccessful
(13, 22)
. Whether conclu-
sive or not, all of these studies were intrinsically limited by the
small magnitude of the magnetic field changes induced by neu-
ronal activation. Moreover, and probably more importantly, they
