Observation of two distinct modes of BOLD signal in rat somatosensory fMRI data

Introduction: The BOLD contrast used in fMRI has been successfully exploited
to map the human brain in various tasks. However, there are still many
questions concerning the relationship between cerebral hemodynamics and the
changes in the BOLD signal. In fMRI BOLD experiments, a complex array of
mechanisms may affect image contrast, including alternations of blood flow,
blood volume, oxygen metabolic rate and intravascutar magnetic susceptibility.
The purpose of this study was to identify the most significant characteristic
patterns in the temporal behavior of the BOLD signal in a blocked designed
fMRI somatosensory rat study.
Statistical Method: To obtain, model free, temporal characteristic patterns of
the data, the principal component analysis (PCA) method was used. Inspection of the first few PCs showed two patterns persistent in most of the data. The rest
of the PCs did not show clear temporal behavior and the spread of their
expansion coeffficients over the brain had a diffuse and noisy nature. Since the
PCA analysis yielded mild PCs variance values, correlation analysis was used
with the above persistent patterns as models.
Method: Male Sprague-Dawley rats 1250-350 g) were anesthetized with urethane i.p. (150mg/100 g) and put in a Bruker-Biospec 4.7T. The rats' heads were
secured in a home built head holder and body temperature was kept constant.
A pair of small needle-electrodes were inserted under the skin of the right or the
left hindlimb. The stimulation consisted of 0.3 ms rectangular pulses with
frequency of 3 Hz and amplitude between 10 and 25 V. A 20 mm diameter
surface coil was used to transmit rf pulses and receive the signal. The surface coil
was placed over the skull and centered over the midline of the animal. Coronal
slices positioned between bregma 0.5 and -2.5 were taken. MR images were
acquired using the gradient echo (GE) sequence (TE = 40 ms, TR = 80 ms) with
field of view of 2.56 cm, 1 mm slice thickness and resolution of 64 x 64. The flip
angle was ~ 35 ~ in the is common to all activation maps of the increasing
function mode observed in this study.
To ensure that the increasing function mode is not an artifact, the experiment
was repeated with an additional 6.8 min (extra 80 points) with no stimulation
during that time. In Fig. 1, we show the first PC obtained in this experiment.
It exhibits a mixture of both modes: the higher correlation with the stimulus
mode is seen on top of the increasing mode. We notice two different phases of
returning to rest value: a slow signal decay process lasting for ~ 3.5 rain, and
a fast decay process.
Conclusion: Understanding the mechanism of the increasing function mode as
part of the complex bemodynamic response to stimulation, is essential for all
neuroimaging methods. Even if this mode is not directly related to neuronal
activity but presents a global hemodynamic response to the stimulation, one has
to appreciate its contribution to the BOLD signal in order to correctly map the
neuronal activity