Spatial specificity of the enhanced dip inherently induced by prolonged oxygen consumption in cat visual cortex: implication for columnar resolution functional MRI.

Abstract

Since changes in oxygen consumption induced by active neurons are specific to cortical columns, the small and transient "dip" of deoxyhemoglobin signal, which indicates an increase in oxygen consumption, has been of great interest. In this study, we succeeded in enhancing and sustaining the dip in the deoxyhemoglobin-weighted 620-nm intrinsic optical imaging signals from a 10-s orientation-selective stimulation in cat visual cortex by reducing arterial blood pressure with sodium nitroprusside (a vasodilator) to mitigate the contribution of stimulus-induced blood supply. During this condition, intact spiking activity and a significant reduction of stimulus-induced blood volume changes (570-nm intrinsic signals) were confirmed. The deoxyhemoglobin signal from the prolonged dip was highly localized to iso-orientation domains only during the initial approximately 2 s; the signal specificity weakened over time although the domains were still resolvable after 2 s. The most plausible explanation for this time-dependent spatial specificity is that deoxyhemoglobin induced by oxygen consumption drains from active sites, where spiking activity occurs, to spatially non-specific downstream vessels over time. Our results suggest that the draining effect of pial and intracortical veins in dHb-based imaging techniques, such as blood oxygenation-level dependent (BOLD) functional MRI, is intrinsically unavoidable and reduces its spatial specificity of dHb signal regardless of whether the stimulus-induced blood supply is spatially specific.

Full text

Page 1

Spatial specificity of the enhanced dip inherently induced by
prolonged oxygen consumption in cat visual cortex: Implication for
columnar resolution functional MRI
Mitsuhiro Fukuda,aPing Wang,aChan-Hong Moon,aManabu Tanifuji,band Seong-Gi Kima,*
aBrain Imaging Research Center, Department of Neurobiology, University of Pittsburgh, 3025 East Carson Street, Pittsburgh, PA 15203, USA
bLaboratory for Integrative Neural Systems, Brain Science Institute, The Institute of Physical and Chemical Research (RIKEN), 2-1 Hirosawa, Wako-shi,
Saitama 351-0198, Japan
Received 22 December 2004; revised 13 July 2005; accepted 15 September 2005
Available online 27 October 2005
Since changes in oxygen consumption induced by active neurons are
specific to cortical columns, the small and transient ‘‘dip’’ of
deoxyhemoglobin signal, which indicates an increase in oxygen
consumption, has been of great interest. In this study, we succeeded
in enhancing and sustaining the dip in the deoxyhemoglobin-weighted
620-nm intrinsic optical imaging signals from a 10-s orientation-
selective stimulation in cat visual cortex by reducing arterial blood
pressure with sodium nitroprusside (a vasodilator) to mitigate the
contribution of stimulus-induced blood supply. During this condition,
intact spiking activity and a significant reduction of stimulus-induced
blood volume changes (570-nm intrinsic signals) were confirmed. The
deoxyhemoglobin signal from the prolonged dip was highly localized to
iso-orientation domains only during the initial ¨2 s; the signal
specificity weakened over time although the domains were still
resolvable after 2 s. The most plausible explanation for this time-
dependent spatial specificity is that deoxyhemoglobin induced by
oxygen consumption drains from active sites, where spiking activity
occurs, to spatially non-specific downstream vessels over time. Our
results suggest that the draining effect of pial and intracortical veins in
dHb-based imaging techniques, such as blood oxygenation-level
dependent (BOLD) functional MRI, is intrinsically unavoidable and
reduces its spatial specificity of dHb signal regardless of whether the
stimulus-induced blood supply is spatially specific.
D 2005 Elsevier Inc. All rights reserved.
Keywords: BOLD; Brain mapping; CBF; CBV; Hemodynamic response;
Hypotension; Sodium nitroprusside; Intrinsic signal; Optical imaging;
Ocular dominance column; Orientation column
Introduction
Non-invasive mapping of functional cortical columns with
common neural properties is critical to understanding how the
brain works. Since neural activity induces a focal increase in
glucose and oxygen consumption rates, the metabolic response is
supposed to be localized to the active site (Lowel et al., 1987;
Thompson et al., 2003; Woolsey et al., 1996). Therefore, focal
increases of deoxyhemoglobin (dHb) produced by the oxygen
consumption change may precisely indicate the location of neural
active sites. This expectation was confirmed with intrinsic optical
imaging spectroscopy studies (Malonek and Grinvald, 1996),
which showed that the early increase of dHb signal (referred to as
the ‘‘dip’’, i.e., metabolic response) is better localized to
orientation-selective columns in the cat visual cortex than the
late decrease of dHb signal due to an over-compensation of
cerebral blood flow (CBF). Similarly, cortical columns in the cat
visual cortex were successfully mapped only by the early dip in
blood oxygenation level-dependent functional magnetic resonance
imaging (BOLD fMRI (Ogawa et al., 1990)) but not by late
responses at high magnetic fields (Kim et al., 2000a). Thus, the
early dip has been of great interest to the functional imaging
community (Hennig et al., 1994; Hu et al., 1997; Menon et al.,
1995).
However, due to the small and transient nature of the early
negative BOLD dip (Buxton, 2001; Duong et al., 2000a; Jezzard et
al., 1997; Kim et al., 2000b; Logothetis, 2000) and consequently
the low signal-to-noise ratio (SNR), it is difficult to routinely map
sub-millimeter-scale functional columns from this dip. Alterna-
tively, brain activity has been commonly mapped from the positive
BOLD signal, which indicates a decrease in dHb (i.e., an over-
compensated hemodynamic response), although the spatial specifi-
city of this signal is poor compared to metabolic response. The
poor spatial specificity of the late dHb responses (positive BOLD)
can be explained by two hypotheses: (I) widespread CBF and
cerebral blood volume (CBV) responses beyond the site of spiking
1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2005.09.026
www.elsevier.com/locate/ynimg
NeuroImage 30 (2006) 70 – 87
* Corresponding author. Fax: +1 412 383 6799.
E-mail address: kimsg@pitt.edu (S.-G. Kim).
Available online on ScienceDirect (www.sciencedirect.com).

Page 2

activity (Malonek and Grinvald, 1996) and (II) draining of dHb
from the active site to non-specific downstream vessels (Duong et
al., 2000a). Under normal physiological conditions, the increased
oxyhemoglobin supply (i.e., CBF increase) and dHb drain occur
simultaneously, making separation non-trivial. In order to decouple
these two hypotheses, we propose to modulate the CBF and CBV
responses induced by neural activity but without changing neural
activity and oxygen consumption rates.
In this paper, the spatial specificity of dHb signals to iso-
orientation columns is investigated in the absence of evoked CBF
and CBV responses (i.e., conditions matching those of the early
dip); these studies were performed in cat visual cortex by using
optical imaging of intrinsic signals (OIS), which measures
changes in light reflection from the cortical surface (Blasdel and
Salama, 1986; Grinvald et al., 1986) accompanied by hemody-
namic response to neural activity. To this end, dHb-weighted 620-
nm OIS and CBV-weighted 570-nm OIS (for review, see
Bonhoeffer and Grinvald, 1996) evoked by orientation-selective
gratings were measured at normal and low blood pressure (BP)
induced by sodium nitroprusside (sNP). Since blood vessels are
considerably dilated at low BP (Endrich et al., 1987; Kontos et
al., 1978), any further increase in CBF and CBV due to visual
stimulation (i.e., 570-nm OIS) will be significantly reduced. Thus,
spatiotemporal dynamics of 620-nm OIS inherently induced by
oxygen consumption change can be measured. If, during
stimulation at low BP, the 620-nm OIS is not localized to active
sites, then the spatial localization of the dHb signal is unlikely due
to non-specific blood supply but rather due to the draining of
dHb. This investigation will determine the intrinsic spatial
specificity of dHb-based functional imaging techniques such as
conventional BOLD fMRI and OIS. The companion work studied
with fMRI under similar low BP conditions is reported elsewhere
(Nagaoka et al., in press).
Materials and methods
Ten cats (1.3–2.0 kg, 11–19 weeks) were used in this study
(8 cats for optical imaging and 2 cats for multiple-unit record-
ing). All experimental procedures were approved by the Institu-
tional Animal Care and Use Committee at the University of
Pittsburgh.
Animal preparations and physiological conditions
Each cat was initially treated with atropine sulfate (0.05 mg/kg,
sub-cutaneous injection) and anesthetized with a sub-cutaneous
injection of a mixture of ketamine (20 mg/kg) and xylazine (1 mg/
kg). The cat was then intubated and mechanically ventilated (1–
2% isoflurane in a mixture of initially 50% N2O and 50% O2, and
subsequently 70% N2O and 30% O2). The femoral artery and vein
were cannulated for monitoring arterial blood pressure and
injection of drugs, respectively. An intravenous catheter was also
inserted into the cephalic vein for continuous infusion of
pancuronium bromide (0.2 mgIkg?1Ih?1) mixed in 5% dextrose
Ringer’s solution by an infusion pump (Product No. 55-3333,
Harvard Apparatus, MA). The cat was then placed in a stereotaxic
apparatus (SN-3N, Narishige, Japan). Electroencephalogram
(EEG) electrodes were implanted in the frontal bone, and a metal
head post and a recording chamber (a 17.5-mm inner diameter)
were mounted on the skull with dental acrylic cement. The
chamber was placed so that it included A10-P5 and L0-L5 in
Horsley–Clarke coordinate; this region roughly corresponds to the
location of area 18 in the lower quadrant of the visual field, which
is approximately between the vertical meridian and ¨25- into the
contralateral field, and between the horizontal meridian and ¨25-
into the lower field (Tusa et al., 1979). After performing a
craniotomy inside the chamber, the dura mater was resected. The
inside of the chamber was then filled with ¨1.5% agarose (Product
No. 05065, Fluka BioChemika, Switzerland) and sealed with a
round glass cover slip. After surgery, the stereotaxic frame was
removed, and the cat was secured by the head post under a
macroscope objective for optical imaging. Recordings commenced
1–2 h after the surgery.
To assess the anesthetic level of the cats throughout surgery
and experiments, we continuously monitored their rectal temper-
ature, EEG, electrocardiogram, arterial blood pressure, arterial
oxygen saturation (SpO2) and expired-CO2. These parameters
were recorded with BIOPAC (MP150, BIOPAC Systems Inc.,
CA) for later analyses. Rectal temperature was maintained
between 38.1 and 38.7-C with a feedback regulated heating pad
system (FHC, Inc., ME). Expired-CO2was monitored by capno-
meter (Capnomac Ultima, Datex Omeda, Finland) and maintained
in the range of 3.1–3.9% by adjusting the volume or rate of the
ventilator (RSP-1002, Kent Scientific, CT) or both. SpO2
measured with a pulse oximeter (8600 V, NONIN Medical, Inc.,
Denmark) was almost 100%. Arterial blood gas analysis (Stat
profile pHOx, Nova Biomedical, MA) was performed under
normal blood pressure conditions for nine cats (these measure-
ments were not performed for one of the cats used for multiple-
unit recording) as follows (mean T one standard deviation (SD)):
PCO2(24.3 T 1.9 mm Hg, n = 9), PO2(204.9 T 19.1 mm Hg, n =
8 (the measurement could not be performed for one of the cats
used for optical imaging)), and pH (7.46 T 0.03, n = 9). The
isoflurane concentration in a gas mixture of 70% N2O and 30%
O2was maintained at the same level during each set of recording
for individual cats (usually 0.85–1.0%), which consisted of
control (before vasodilator injection), low BP (during injection),
and recovery (post-injection) conditions. The range of mean
arterial blood pressure (MABP) for control was 76.1–100.9 mm
Hg, and that for recovery was 81.6–105.1 mm Hg. These were
within the normal range of MABP for cats under ¨1% isoflurane
anesthesia (Harel et al., 2002a; Zhao et al., 2004a). The range of
MABP for low BP was 40.1–55.4 mm Hg.
Sodium nitroprusside-induced low blood pressure
To reduce arterial BP, a vasodilator (sodium nitroprusside
dehydrate (sNP), Abbott Laboratories, IL) was injected intra-
venously for ¨25 min (total amount of injection: 0.32–0.59 mg/
kg). The sNP (2.5, 5.0 or 7.5 mg) was dissolved in 6 ml of either
saline or a mixture of saline and Dextran 40 or 70 (Baxter, IL).
The vasodilator infusion rate was dynamically adjusted (between
0.0003 ml/min (initial) and max. 0.08 ml/min to avoid cyanide
toxicity) by manually controlling an infusion pump (PHD2000,
Harvard Apparatus) to maintain MABP at ¨45 mm Hg for ¨25
min during optical imaging and for ¨15 min during multiple-unit
recording. The initial drop in MABP tended to reach ¨40 mm Hg,
and in two cases, EEG signals became flat at that period. After
termination of the vasodilator injection, MABP returned to levels
similar to control values (before the vasodilator injection) within
¨5 min.
M. Fukuda et al. / NeuroImage 30 (2006) 70–87
71

Page 3

Visual stimuli
All visual stimuli were presented binocularly. Square-wave
high-contrast gratings were generated by a graphic video board
(VSG2/5, Cambridge Research Systems, United Kingdom) con-
trolled by a custom software written in MATLAB 6.5 (MathWorks,
Inc., MA) or Microsoft Visual C++ 6.0 and were presented on a
21-in. CRT monitor (800 ? 600 pixels and 100 Hz refresh rate,
GDM-F520, Sony). The spatial and temporal frequencies of the
gratings were 0.15 cycle (c) per degree and 2 c/second (s)
respectively for selective stimulus of area 18 (Bonhoeffer et al.,
1995; Issa et al., 2000). The direction of the motion of the gratings
was reversed every 0.5 s during visual stimulation. The cat’s pupils
were dilated with ophthalmic solutions of atropine (1%), phenyl-
ephrine hydrochloride (2.5%), and proparacaine hydrochloride
(0.5%). Contact lenses (T0.0 D, Danker Laboratories Inc., FL)
were fitted to the eyes to prevent corneal drying. The visual field
center was roughly estimated by projecting images of optic disks
and retinal vessels onto the CRT monitor (Bishop et al., 1962)
whose distance was then placed where the best focus of the optic
disks and the retinal vessels were obtained (19–31 cm from eyes;
visual field 70-–115- wide and 54-–88- high).
Optical imaging
The intrinsic signals were recorded with a custom-made
imaging system consisting of tandem lens optics (Ratzlaff and
Grinvald, 1991) (i.e., face-to-face combination of a projection lens
(50 mm F/1.2, Nikon, Japan) and an objective lens (50 mm F/1.2,
Nikon)), a charge-coupled device (CCD) camera (CS8310, 640 ?
480 pixels, Tokyo Electric Industry, Japan) and a 10-bit frame
grabber board (either Pulsar or Corona-II, Matrox Graphic Inc.,
Canada). The imaging area was 8.8 ? 6.6 mm2. For seven of the
eight cats, the camera was focused on the cortical surface to detect
vascular responses on pial vessels. The diaphragm of the projection
lens was set to F/2.8 for a large depth of field. For the remaining
one cat, the camera was refocused 500 Am below the cortical
surface, and the diaphragm of the projection lens was set to F/1.2
for a shallow depth of field as with conventional optical imaging.
Since the results obtained from both depth of field were essentially
the same, both data were pooled together for analyses. Images were
captured at 1/30 s with the frame grabber board controlled by
custom software written in Microsoft Visual C++ 6.0. To improve
SNR and also reduce data size, temporal averaging of 15
consecutive images and spatial binning of 2 ? 2 pixels were
performed before saving the acquired images. Thus, the final
temporal and spatial resolutions of the saved images were 0.5 s/
image and 27.5 ? 27.5 Am2/pixel, respectively. The cortical
surface was illuminated through an interference filter (540, 570, or
620 T 10 nm, Oriel Instruments, CT) using a bifurcated fiber optic
bundle (SP77533, Oriel Instruments, CT) that was connected to a
tungsten–halogen light source (Model No. 66881 and 68931, Oriel
Instruments, CT).
Experimental design
Experimental protocol was outlined in Table 1. In all studies,
cortical surface imaging with 540-nm wavelength and four-
orientation 620-nm OIS were performed. Additionally, single-
orientation 620-nm and/or 570-nm OIS were measured from the
same cortical surface, or multiple neural recordings were
performed as follows.
Cortical surface imaging with 540 nm
Initially, an image of pial vascular pattern was taken with 540-
nm wavelength, at which light absorption by hemoglobin is largest.
Four-orientation 620-nm OIS
After acquiring an image of surface vascular patterns, iso-
orientation domains for a particular orientation were mapped with
620-nm OIS, which is empirically known to give excellent
contrast. To obtain the iso-orientation map efficiently and quickly
with the conventional cocktail blank method (described in OIS data
analysis), four orientations (0-, 45-, 90-, and 135-) were used with
short stimulus duration (2 s) and inter-stimulus interval (2 s). Each
stimulation run consisted of 1-s of pre-stimulus homogenous gray
(i.e., 2 images) and 2-s stimulus with a moving grating presented at
one of the four orientations followed by 1-s of post-stimulus
homogenous gray. Twenty runs were repeated for each orientation
stimulus (i.e., total recording time, ¨2.7 min). The four orienta-
tions were presented at a sequential order (from 0- to 45-, 90- and
135-). Based on our experience, the orientation presentation order
(e.g., random or sequential presentation) does not affect iso-
orientation maps in this short time recording.
Table 1
Experimental design
Experiment PurposeMABPStimulationRemarks
Cortical surface
imaging with 540 nm
Four-orientation 620-nm OIS
Visualization
of vessels
Mapping iso-orientation
domains
NormalNone –Separation of tissue
and vessel ROIs
–‘‘Reference’’ for single-condition
iso-orientation maps
–Assignment of active and inactive
domains for quantitative analyses
–Determination of whether functional
CBV response is decreased at low BP
Normal4 orientations, 2-s
stimulation duration
2-s inter-stimulus
interval (ISI)
One orientation,
10-s stimulation
64-s ISI (49 s in one case)
One orientation,
10-s stimulation
64-s ISI
4 orientations
10-s stimulation
5-s ISI
Single-orientation
570-nm OIS
Spatiotemporal dynamics
of CBV responses
Normal and
Low
Single-orientation
620-nm OIS
Spatiotemporal dynamics
of dHb responses
Normal and
Low
–Enhancement of the dip at low BP
–Spatial specificity of the prolonged dip
Multiple-unit recordingFidelity of neural activity Normal and
Low
–Selectivity of neural activity during low BP
M. Fukuda et al. / NeuroImage 30 (2006) 70–87
72

Page 4

Single-orientation 570- and 620-nm OIS
To determine contribution of CBV response or spatial
specificity of dHb signal, 570-nm or 620-nm OIS induced by
10-s visual stimulation were recorded during normal and low BP
conditions respectively (n = 5 cats for each wavelength, 2 of
them for both wavelengths). The 10-s stimulation duration is
relevant to typical block-design fMRI studies. To achieve
efficient signal averaging in a limited experimental time during
the low BP, only single orientation (0-, 45-, 90-, or 135-) was
presented. Twenty single-orientation stimulation runs were con-
tinuously acquired at each of normal BP, low BP, and recovery
sessions; each run consisted of 10 s of pre-stimulus stationary
grating and a 10-s moving grating followed by 54 s (39 s for one
570-nm OIS) of post-stimulus stationary grating to allow the
signals to return to pre-stimulus baseline (i.e., inter-stimulus
interval was 64 or 49 s). Since results obtained from the two
stationary grating periods (64 and 49 s) were essentially the same,
both data were pooled together for analyses. For low BP sessions,
OIS recordings were started after the MABP reached a desired
level (¨45 mm Hg). One hour after termination of vasodilator
injections, the recovery of the signals was confirmed.
Multiple-unit recording
Spikingactivityofmultipleneuronswasrecordedextracellularly
from single tungsten microelectrodes (Catalog #UEWLFFL-
MNN1E, Frederick Haer and Co. Brunswick, ME) during control
(normal BP), low BP, and recovery conditions. The electrode was
insertedusingahydraulicmicrodrive(NarishigeMO-81,Japan)into
an iso-orientation domain pre-determined with the Four-orientation
620-nm OIS recording. The spikes whose amplitudes exceeded a
certain threshold, which was set near background activity, were
detected at 40 kHz using a spike acquisition system (Plexon Inc,
Dallas, Texas). Stimulus consisted of a 2-s homogeneous gray and a
10-s duration moving grating with one of four orientations (0-, 45-,
90-, and 135-) followed by 3-s homogenous gray. For each
orientation stimulus, five runs were averaged and spike firing rates
in 100-ms bins during stimulation were calculated to determine the
selectivity of multiple neurons recorded.
OIS data analysis
All data were analyzed with MATLAB 6.5. After averaging
across stimulation runs for each experiment, three sorts of maps
(single-condition raw activity, differential iso-orientation, and
single-condition iso-orientation maps) were generated as described
in detail below. All the values below are reported as mean T 1 SD.
Statistical significance was examined by ANOVA and two-tailed
paired t test. P values of less than 0.05 were considered to be
statistically significant.
Averaging across stimulation runs
To increase SNR, OIS runs were averaged. For the experiment
of Four-orientation 620-nm OIS, all 20 OIS runs (4 s/each run; 1-s
pre-stimulus and 3-s post-stimulus onset) were averaged for each
orientation. For the experiment of Single-orientation 570 and 620-
nm OIS, OIS runs were first chopped into a period of 40 s (10-s
pre-stimulus and 30-s post-stimulus onset) and screened based on
MABP during each 40-s run before averaging. To obtain the
MABP during each run, MABP at every cardiac cycle was
calculated from the sum of two-thirds of diastolic BP and one-third
of systolic BP, and all the MABP for 40 s were averaged. Then one
SD of the mean for MABPs across 20 runs (15 runs for low BP
sessions, the first five runs were excluded from analyses because
the reflected light intensity was not yet stable (see Fig. 2B right
panel)) was calculated. Finally, OIS runs whose MABP were
within a range of the one SD were averaged. At each BP condition,
one averaged run was determined.
Generating single-condition raw activity maps
Then the first frame analysis (for review, see Bonhoeffer and
Grinvald, 1996) was performed on the averaged run to reduce the
contribution of slow signal fluctuations related with vasomotion
(e.g., ¨0.1 Hz (Mayhew et al., 1996)) to OIS time courses; the
average of two images (i.e., 1 s) just before stimulus onset (pre-
stimulus image, R) was subtracted from all images (R(t), t is time
from the stimulus onset). The subtraction images (DR = R(t) ? R)
were then divided by the pre-stimulus image (R), so that the pixel
values indicate a relative change of reflectance (DR/R; 0, no
change of light reflection; ? (negative sign), decrease of light
reflection (i.e., increase of light absorption); + (positive sign),
increase of light reflection (i.e., decrease of light absorption)). This
first frame analysis was applied to both the Four-orientation 620-
nm OIS and Single-orientation 570 and 620-nm OIS experiments;
it is equivalent to a percentage signal change induced by
stimulation, which is commonly used in functional MRI. The
image generated by the first fame analysis is referred to as a
‘‘single-condition raw activity map’’ (Fig. 1B). All images in the
Results section are shown as gray-scale code, where brightness of
gray shows intensity of light reflection from pre-stimulus baseline
(i.e., darkening and lightening represent decrease and increase in
light reflection, respectively).
Generating differential iso-orientation maps
As seen in the single-condition raw activity map (Fig. 1B) or in
the map profile (Fig. 1C), OIS is induced not only in the domain
specific to a particular orientation but also in all other domains for
other orientations, resulting in orientation-specific and -non-
specific components (see also Frostig et al., 1990; Grinvald et
al., 1986; Malonek and Grinvald, 1996). To remove the
orientation-non-specific component and to quantitatively deter-
mine iso-orientation domains specific to a particular orientation,
we used the cocktail blank data analysis method, which is
commonly used to generate iso-orientation maps (for review, see
Bonhoeffer and Grinvald, 1996). To this end, first, single-
condition raw activity maps for the four orientations (0-, 45-,
90-, and 135-) were obtained from the experiment of Four-
orientation 620-nm OIS (i.e., 4-s recording for each orientation).
Each map was averaged from 0.5 to 3 s after the stimulus onset
(the image at 0 s was not included in the average because the
signal did not emerge clearly) to increase SNR. Then the resulting
four single-condition raw activity maps were averaged to generate
a cocktail blank image. Finally, the cocktail blank image was
subtracted from the raw activity map with one of the four
orientations, and the spatial filter (Fig. 1D) was applied (see
below). We will refer to the generated map as a ‘‘differential iso-
orientation map’’ (Fig. 1F). This map was used to quantitatively
determine iso-orientation domains (see ROI analysis).
Generating single-condition iso-orientation maps
The local modulations specific to a particular orientation
(orientation-specific component) can also be extracted from global
modulations, which are independent of stimulus orientation
M. Fukuda et al. / NeuroImage 30 (2006) 70–87
73

Page 5

(orientation-non-specific component), by simply applying a spatial
filter (Fig. 1D) to the single-condition raw activity map. We will
refer to this spatial filtered map as a ‘‘single-condition iso-
orientation map’’ (Fig. 1E). The map was obtained from the
experiment of Single-orientation 570- and 620-nm OIS. This map
was used to demonstrate whether iso-orientation domains deter-
mined by single-orientation are consistent with those determined
by four-orientation stimuli (Figs. 6 and 9). Properties of the filter
were determined by the difference of two Gaussian functions
?x2
1 ? e
0.008 for high pass frequency). The spatial domain image (Fig. 1B)
was first transformed into a frequency domain image and then
multiplied by the spatial filter (Fig. 1D) to remove low and high
frequency components. Finally, the filtered frequency image was
inversely transformed into the spatial domain image (Fig. 1E).
Black patches in this map represent preferential activation by a
particular stimulus orientation.
(f x ð Þ ¼ e
2r2
?x2
2r2
2, where r1= 0.4 for low pass frequency, r2=
ROI analysis
To determine spatiotemporal responses induced by a stimulus
with a single-orientation, ‘‘tissue’’ and ‘‘vessel’’ regions were
chosen (Fig. 1G); the separation between tissue and vessel regions
was determined on the cortical surface image taken at 540 nm (see
Fig. 1A). Since vessels absorb more light than tissue at this
wavelength, vessels appear dark in cortical surface images at 540
nm. Pixels containing signal from pial vessels (a pixel size was
27.5 ? 27.5 Am2) were assigned to the vessel region of interest
(ROI), and the remaining pixels were the tissue ROI (Fig. 1G). The
ratio of vessel ROI to tissue ROI was 0.67 T 0.06 (n = 7 cats). In
one cat, due to blurring of pial vessels caused by defocusing
camera from cortical surface, the vessel ROI was larger than tissue
ROI (the ratio was 1.96), however, this does not affect the result.
Tissue ROI was further sub-divided into ‘‘active’’ and ‘‘inactive’’
domains based on the differential iso-orientation map. The pixels
having negative values (?DR/R%) in the differential iso-orienta-
tion map (Fig. 1F) were assigned to active domains, and the
remaining pixels (0% and +DR/R%) were assigned to inactive
domains (Fig. 1H). The ratio of active domains to inactive domains
was 0.76 T 0.20 (n = 8 cats). Average time courses were plotted
from all pixels within each ROI (i.e., vessel, tissue, active, and
inactive domains) in the single-condition raw activity maps.
Results
Effects of sNP administration on arterial blood pressure and
cortical reflected light intensity
A decrease in arterial blood pressure (BP) causes an increase in
CBV to keep CBF rather constant (i.e., autoregulation). Thus, the
increase in light absorption (decrease in reflected light) induced by
the increase in CBV is expected at low BP. To test this prediction,
570-nm reflected light intensities between normal and low BP were
compared. Fig. 2 shows continuous recordings of arterial BP (A)
and cortical reflected light at 570-nm wavelength (B) during
normal BP (left panel) and during low BP, induced by sNP
injection (right panel). The sNP reduced MABP, systolic and
diastolic blood pressure, and also pulse pressure (difference
between systolic and diastolic blood pressures) (compare left and
right panels of Fig. 2A and insets). Target MABP during sNP
injections was at ¨45 mm Hg (right panel of Fig. 2A), which may
be slightly below the lower limit of CBF autoregulation (Chillon
and Baumbach, 2002). Under this condition, vessels are consid-
erably dilated (Kontos et al., 1978), and thus CBV (Schumann-
Bard et al., 2005) and total hemoglobin (Ferrari et al., 1992) are
increased. In accordance with our expectation, the reflected light
intensity gradually decreased from ¨800 arbitrary units during
normal BP to ¨700 units during low BP (Fig. 2B).
To quantitatively compare properties of reflected light between
the two BP conditions, mean of the reflected light intensity for a
10-s pre-stimulus period was calculated for each of the selected
runs (see OIS data analysis in Materials and methods) for both BP
conditions. The mean reflected light intensity of 570-nm wave-
length during low BP (MABP, 48.3 T 5.8 mm Hg) for all five cats
was 8.0 T 6.7% less than that during normal BP (MABP, 81.2 T 5.0
mm Hg). Similarly, the average reflected light intensity at 620 nm
during low BP (MABP, 48.1 T 3.3 mm Hg) for all five cats was
reduced by 2.2 T 1.5%, compared to that with normal BP (MABP,
82.1 T 3.3 mm Hg). The decreases in reflected light intensity were
Fig. 1. Cortical activation map generation for analysis of OIS data. (A) A
cortical surface image (540 nm) showing vessels for correlation with the
maps (620 nm) in panels B, E, F, G, and H. (B) A single-condition raw
activity map. The map was obtained by averaging relative change of
reflectance images after stimulus onset (0.5–10 s, the image at 0 s was not
included in the average because the signal did not emerge clearly). Global
absorption increase was observed as a darkening of the entire region
(grating at 45- orientation). (C) The profile along the line a-b in map B. (D)
Properties of the spatial filter used to remove the global absorption change.
(E) A single-condition iso-orientation map. The global absorption change
was filtered from the single-condition activity map in panel B. Black
patches are now responses specific to the grating at 45- orientation. (F) A
differential iso-orientation map. Locations of black patches were deter-
mined on this map and marked with white crosses on all maps. (G)
Separation of pial vessels (black) and tissue (yellow) based in image A. (H)
Separation of active (red) and inactive (blue) domains. Based on the
differential iso-orientation map F, the ROI was sub-divided into active and
inactive domains for the grating at 45- orientation. Maps were demonstrated
with normal BP data (average for 12 runs). See Materials and methods for
processing details.
M. Fukuda et al. / NeuroImage 30 (2006) 70–87
74