In accordance with previous studies, we confirmed the capability of LASCA to image functionally active areas in a rat model of somatosensory activation (Dunn et al. 2003; Durduran et al. 2004; Lau et al. 2005; Weber et al. 2004). We were able to show that the activation contrast of this method allows a time-resolved relative CBF imaging that enhances microcirculatory areas. Part of this enhancement can be attributed to the rather long exposure time used in our setup (Yuan et al., 2005). However, MR studies have shown a tight spatial coupling of perfusion-weighted methods (Duong et al., 2001) that outperforms BOLD-fMRI even at high field strength (Pfeuffer et al., 2002). Duong et al. concluded that neurovascular coupling has a high spatial resolution if the derived signal can be confined to the capillary compartment. In FAIR-MRI, this is achieved by exploiting the fact that the uninverted water that gives rise to the T1 contrast diffuses out of the capillaries into the brain parenchyma. Thus, this contrast agent accumulates (and gradually vanishes) at the center of activation instead of being dispersed by the draining veins after seconds as it is the case in permanent contrast agents (including intrinsic deoxygenated hemoglobin in BOLD-fMRI) that do not cross the blood brain barrier. Our study shows that – apart from this unquestionable advantage gained by the technically inherent features of FAIR-MRI – CBF imaging per se enhances microcirculatory areas over the pial vasculature. This high localizing capability can be ascribed to a minimization of vascular compartment heterogeneities. On the one hand, CBF imaging is superior to BOLD/Deoxy-Hb imaging because it is not influenced by the vascular oxygenation gradient that can seriously affect BOLD-fMRI by shifting the main signal to the poorly localized venous compartment. On the other hand, CBF imaging is superior to CBV imaging because the latter amplifies signals from vascular compartments with high compliance, which, again, results in a high venous contribution. However, for CBV imaging, this effect can be overcome by focusing on the “early response” before the local contrast agent rise reaches the venous compartment. Several groups, including our own, have shown that under these circumstances CBV imaging can image single activated cortical columns (Lindauer et al. 2001; Sheth et al. 2004; Vanzetta et al. 2005). In this study, we have shown that for CBF measurement the analysis does not need to be confined to the early response in order to deliver high resolution functional images: the relative magnitude of LASCA measured CBF changes was highest in the capillary compartment whereas the contribution of larger vessels was smaller. This can be attributed to the fact that in the passively distended postarteriolar compartment (the “windkessel” or “balloon”) CBF is determined by both changes of CBV and red blood cell velocity (CBF=CBV/t=A*s/t=A*v where A is the vessel cross-section, s the length of the blood vessel and v the blood velocity). If the vessel cross-section increases, the red blood cell velocity decreases and vice versa. This renders CBF imaging rather insensitive to the high compliance of the pial veins. Even though CBV increases (leading to a relative high venous contribution in CBV-weighted imaging), CBF is not affected as much since the red blood cell velocity decreases. This antagonizes the relative high CBV signal from the pial veins. Although CBF imaging is also affected by CBV changes, it is not as sensitive to compliance effects during passive vasodilation.#

We conclude that LASCA imaging is an excellent method for high resolution functional brain imaging in animal studies that enhances microcirculatory areas due to a relative insensitivity to vascular heterogeneities of compliance and oxygenation.#

During functional challenge by electrical forepaw stimulation, we found a tight correlation between CBF changes measured by LASCA and LDF. When put into relation to the underlying neuronal activity (revealed by SEPs), the relative amplitudes of LASCA and LDF were comparable. There is evidence that LDF measurements overestimate relative blood flow changes compared to autoradiographic CBF measurements, but predominantly in high flow states (Dirnagl et al. 1989; Fabricius and Lauritzen 1996; Goadsby 1991). Durduran et al. have found peak CBF responses on somatosensory stimulation with LASCA that were about 30% lower than their usual LDF measurements (Durduran et al., 2004). However, other reasons than a systematic difference between the two methods might be responsible for the lower CBF amplitudes in LASCA: CBF was not measured in the same animal with both methods and electrophysiological reference data were not included into this comparison. Our results confirm the findings of Dunn et al. for functional neuroimaging. They report approximately equal relative CBF changes for both methods during ischemia and spreading depression (Dunn et al., 2001).#

In summary, LDF and LASCA measured relative CBF values show a tight correlation. We conclude that LASCA is a reliable CBF imaging method with an enormous capability to resolve functional CBF responses with high spatial and temporal resolution.#

In agreement with other groups, we found a correlation between CBF increases and underlying neuronal activity that can be described by a nonlinear function of the form y=ax^b. However, although the estimated b was greater than unity (1.83 and 1.76 for LASCA and LDF measurements, respectively), the confidence intervals of these estimations had lower bounds of 1.146 and 1.014. Thus, a linear relationship cannot be excluded. In addition to that, part of the observed nonlinearity might be due to differences of signal to noise ratios that are inherent in all method comparisons. In order to remain below the threshold where pain reactions set in we did not increase the stimulation intensity above 2.0 mA. These data are consistent with the data shown by Ureshi et al.: using electrical hindpaw stimulation, they found a nonlinear relationship between LDF CBF response and LFP (Ureshi et al., 2005). The studies by Devor et al. also showed a nonlinear relationship between LFP and CBF increases in rat somatosensory cortex upon mechanical whisker deflection (Devor et al., 2003). Jones et al. found a pronounced sigmoid relationship between CBF increases and summed local field potentials (Jones et al., 2004). Though the dynamic range of stimulation intensities is comparable, these latter data were gathered during electrical stimulation of the rat whisker pad. This facial region has a large representation in the posterior barrel subfield and it can be assumed that it is more sensitive than the forepaw system when using the same stimulus current. For this reason, with the same range of stimulus current intensities, the dynamic range of evoked CBF responses might be larger in the whisker system compared to the forepaw system. A similar study of Nielsen et al. compared CBF responses with LFP during infraorbital nerve stimulation. They also fitted a sigmoid function to this relationship (Norup and Lauritzen, 2001). However, their data could most likely be described by a power law function as well. In summary, our data show a nearly identical neurovascular coupling relationship between CBF measured by LASCA vs. LDF. This relationship can be described by a nonlinear function. This confirms findings by other groups and strengthens the assumption that LASCA is a promising functional imaging method that essentially adds imaging capability to the well established LDF method.#

After fixation in a stereotactic frame, a closed cranial window was implanted over the right somatosensory cortex. The center of this window was 5 mm lateral and 1 mm caudal to bregma. After removal of the skin, the cranial bone was thinned to translucency with a drill over an area of size 4×5 mm. To prevent thermal injury to the cortex, saline solution was applied. The bone layer was then carefully removed with a forceps, without disrupting the dura. To measure somatosensory evoked potentials, a ball-tip silver electrode was placed on the dural surface and connected to an EEG recorder (Neuropack 2, Nihon Kohden, Tokyo, Japan). The reference electrode was inserted into the skin of the neck. Around the cranial window, a wall was formed with bone wax. Physiological saline solution was filled on top of the dura. For LASCA measurements, the window was closed with a cover slip. Two needle electrodes were then inserted into the skin over the left carpal joint for later delivery of stimulus currents. After surgery, anesthesia was switched to an i.v. application of α-Chloralose/Urethane (40 mg α-Chloralose and 400 mg Urethane per kg body weight as bolus, followed by the same dosage continuous i.v. infusion per hour within the first hour and then reduction of the infusion rate as necessary). Anesthesia level was assessed periodically by testing responses to tail pinch and monitoring MABP. To minimize transition effects, functional activation studies were done after a 45-min waiting period. The physiological values during the experiment were kept within normal limits (Table 1).#

Somatosensory stimulation was performed by electrical forepaw stimulation (Neuropack 2, Nihon Kohden, Tokyo, Japan) via two needle electrodes that were inserted subcutaneously over the left carpal joint. Each stimulation train had a duration of 4 s, with 0.2 ms rectangle pulses delivered at a frequency of 4 Hz. To test for a linearity of the neurovascular coupling and for the correlation of LASCA and LDF measurements, a range of different amplitudes were obtained by varying the current intensity between 0.4, 0.8, 1.2, 1.6 and 2.0 mA. Each stimulation block consisted of 10 stimulations with the same current delivered with an interstimulus interval of 60 s. The order of the stimulation intensity applied was randomized. However, stimulation block order was maintained for both LDF and LASCA measurements within one experiment.#

To obtain a quantitative parameter of the underlying neuronal activity, an epidural EEG was measured within the cranial window parallel to the CBF measurements with LDF and LASCA. The mean somatosensory evoked potential (SEP) from 20 ms before to 180 ms after the applied rectangle pulse was acquired for each stimulation train (Neuropack 2, Nihon Kohden, Tokyo, Japan). In post-processing analysis, the SEPs were averaged across stimulation blocks and normalized to the maximal SEP amplitude during the experiment. For comparison and to test for correlation, the mean amplitudes of the CBF measurements for each stimulation current intensity in each animal were related to the mean absolute value (positive and negative) of the averaged SEP.#

Regional cortical blood flow was measured with LDF (Blood Perfusion Monitor BPM², Vasamedics St. Paul, USA). The principles and technical details of this widely used method have been described elsewhere (Bonner et al. 1987; Stern et al. 1977). Light from a laser diode (λ=780 nm) is delivered to and detected from a tissue area of approximately 1 mm³ by a glass fiber optic. The light is scattered by both stationary and moving red blood cells. Scattering by a moving cell results in a Doppler frequency shift, while light scattered by stationary cells remains unshifted. The backscattered light yields the frequency of the Doppler shift, which is proportional to the cell velocity. The fraction of the backscattered light that is Doppler shifted is proportional to the total volume of moving blood cells in the sample. The multiplication of the mean Doppler shift by the fraction of light that is Doppler shifted results in a quantitative estimate of blood flow (Dirnagl et al., 1989). The 0.8 mm thick LDF probe of this setup (model P-433-2, Vasamedics, St. Paul, USA) contained one afferent fiber (50 μm diameter) and two afferent fibers (100 μm diameter) separated by 500 μm. The combined laser diode and detector was directed onto the dural surface in the cranial window. By changing the position of the LDF probe with a micromanipulator and testing a forepaw stimulation train, the area with the highest CBF amplitude during stimulation was localized. In this area, CBF was measured continuously while the abovementioned design of electrical forepaw stimulations was performed. The absolute values were digitized with a frequency of 1 Hz and saved by a PC. Since LDF does not provide absolute measurements of CBF but a reliable quantification of relative CBF, the obtained values were converted according to(2)CBFrel(t)= LDFabs(t) LDFabs(0)with LDFabs(t) and LDFabs(0) containing arbitrary units of the LDF values during time t and during pre-stimulus baseline, respectively.#

After LDF measurements, the saline-filled cranial window was sealed with a cover slip to provide a homogenous view on the somatosensory cortex. LASCA was performed by imaging the laser-illuminated cortical surface on a CCD camera. A collimated laser diode (λ=780 nm, driven with Laser Diode Controller LDC 310, Profile, Munich, Germany) was directed at the cranial window with an angle of 45° from vertical and adjusted to provide uniform illumination of the somatosensory cortex. An image of the cortex was projected on a 12-bit CCD camera (CCD 1300D, 1288×1024, pixels of 6.7 μm² size, Vosskühler GmbH, Osnabrück, Germany) via a standard camera lens (f=50 mm, 1:1.2, Nikon Germany) and an extension tube of 68 mm to allow for a closer object distance. To optimize the registration of the visible speckle pattern, the size of a single speckle should be equal to the size of a single pixel in the image (Dunn et al., 2001). The used optical setup has a magnification factor of 1:1.36 which decreases image pixel size to 3.62 μm². The diffraction-limited spot size is equal to 2.44λf/# where λ is the wavelength and f/# is the f number of the system. Therefore, the optimal f/# of this system is 2.8. Images were acquired by a frame grabber (IMAQ PCI-1422, National Instruments, Austin, USA) with a frequency of 12 Hz and an exposure time of 80 ms. Synchronization between imaging and stimulation was achieved by custom-written software.#

The theory behind LASCA blood flow imaging has been described in detail by Briers and Webster (1996). In brief, laser speckle is a random interference pattern which occurs when coherent laser light is scattered from a diffuse medium such as brain tissue. Due to minor differences of the traveled pathlength, the light rays differ in phase and can thus interfere in a constructive or destructive manner. Each image point is subject to intensity fluctuations that depend on these phase differences. The addition of two light rays leads to a light intensity that can be increased (and cause an intensity maximum if there is no phase difference) or decreased (and cause a regional minimum if the resulting amplitude is zero). An image of a laser-illuminated surface thus appears granulated, or speckled, provided that the integration time is sufficiently short. Moving particles within the brain tissue (e.g. blood cells) that change the scattering properties dynamically produce a time-varying speckle pattern in each pixel of the image. In LASCA, the spatial characteristics of this speckle pattern are quantified. During increased blood flow, the intensity variations of the speckle pattern are more rapid. Thus, in the same integration (exposure) time, the speckle pattern loses contrast. The post-processing of laser speckle images was done by quantifying this dynamic blurring of the speckle pattern. Analyzing neighboring squares of 8×8 pixels within the laser-illuminated image, the speckle contrast K within each square was calculated using(3)K=σS〈I〉where σS is the standard deviation and 〈I〉 the arithmetic mean of the pixel intensities (Briers and Webster, 1996). Neighboring squares were only one pixel apart and thus overlapped in the analysis. At the expense of a longer processing time, the resulting image appears smoother even though the spatial resolution is not increased. Each group of 64 pixels delivers one speckle contrast K value that was assigned to the center of the analyzed square. K has a minimal value of 0, when the scattering particles are moving fast enough to average out all of the speckles and a maximal value of 1 when the speckle pattern is fully developed with no blurring. With the following relation, the mean velocity of moving particles can be estimated from the speckle contrast:(4)K=[τc2T{1−exp(−2T/τc)}]1/2where T is the camera exposure time and the correlation time τc is related to the mean velocity v by τc=1/(ak0v) where a is an unknown factor depending on the Lorentzian width of the scattered spectrum and the scattering properties of the medium and k0 is the input light wave number (Bonner and Nossal, 1981). To estimate τc, a look-up table was used. Relative values of CBF were calculated for each speckle square of 8×8 pixels by(5) CBFrel(t)= CBF(t) CBF(0)=τc(0)τc(t)with τc(t) and τc(0) containing the correlation times during time t and during pre-stimulus baseline, respectively.#

For comparison with LDF measured CBF values, the pixels in the 1 mm² area centered around the LDF probe were averaged together. For functional imaging, the blood flow response was mapped with a paired t-test comparing ten consecutive CBF values acquired during electrical forepaw stimulation with baseline values in each pixel for each stimulation block. Based on this functional CBF map and an anatomical image acquired at 570 nm, a compartment analysis was done. Within the activated area, three regions of interest containing (a) a pial arteriole, (b) a pial venule and (c) an area without visible blood vessels were defined. The mean CBF response in these areas was calculated and averaged across animals to estimate the contribution of arterioles, venules and microcirculatory areas to the mapping signal. The amplitudes of microcirculatory areas were compared to the amplitudes of venules and arterioles.#

To estimate possible anesthesia effects due to the fact that LASCA measurements took place after LDF measurements, we performed experiments on additional animals (n=2) in which the order was reversed: LASCA measurements were performed before the LDF measurements.#

All data were analyzed using custom written software based on MATLAB (The Mathworks, Inc, Natick, MA). Functional CBF maps were calculated by applying a paired t-test to LASCA images as described above. The amplitudes of microcirculatory areas within these functional maps were compared to the amplitudes of the feeding arteriole and draining venule using a paired t-test with Bonferroni correction. To estimate the correlation between CBF and underlying neuronal activity, mean magnitudes of LDF and LASCA measurements were plotted against mean SEP magnitudes. The direct comparison of LDF and LASCA was done with a Bland–Altman plot and by determining the correlation coefficient.#