INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE ORIGINAL PHOTOELECTRIC METHOD developed by Tomita et al. (15) permitted repeated measurements of the cortical mean transit time (MTT) by use of a nondiffusible intravascular tracer, which necessitated minor assumptions. The observed region was, however, relatively large (3 × 5 mm); flow values in arteries, capillaries, and veins were averaged all together. For a modification of the original technique, we attempted to combine it with the two-dimensional (2-D) method described by Eke (3). The main difference between the method of Eke and the present one used by us is the method of illumination; namely, Eke used reflected light, whereas our method employed transmitted light. This difference is important in that it improves spatial resolution and quantification of dye concentration and is free from contamination by artifactual aberrant surface-light reflection.

The most significant problem in any system using optical signals for monitoring cerebral blood flow or neural function is light scattering by the brain parenchyma. Light scattering causes the optical path length between the light source and the detector to become longer than the physical distance. This difference was estimated by use of the Monte Carlo simulation and was determined in vivo by time-of-flight measurements by Delpy et al. (2). Their results indicated that the optical path length was ~4.5 times longer than the interoptode spacing. This ratio is known as the differential path-length factor (DPF), which is a function of the tissue absorption coefficient, the tissue scattering coefficient, and the tissue geometry. The demarcation of the cortical MTT requires one to measure the relative dye concentrations in the brain parenchyma. This is possible only if the DPF is constant during dye transport. As mentioned above, the DPF changes, albeit slightly, simultaneously with changes in the absorption coefficient. In this paper, we will examine whether the relative dye concentration can be measured with the use of the Lambert-Beer approach. If the answer is in the affirmative, the microregional MTT values can be calculated, and the 2-D map of the reciprocal MTT values will represent the microregional flow distribution, because the blood volume in infinitesimally small cortical areas is relatively constant. The new evidence emphasizes the importance of 2-D cortical blood flow (CBF) measurements: studies using intravital microscopy have shown spontaneous fluctuations in red blood cell (RBC) velocity in capillaries (7), which cannot be explained by a simple "stopcock" theory (10). Functional activation was found to decrease the heterogeneity of capillary plasma perfusion in the rat barrel cortex (17). Hudetz et al. (6), also using intravital microscopy, reported the maintenance of capillary RBC flow velocity during hypotension. The significance of 2-D flow mapping was pointed out by Heimann et al. (5), who employed the scanning laser-Doppler flow (LDF) method. Intravital microscopy and confocal laser-scanning microscopy are very versatile tools for analyzing individual capillaries but are inadequate if one chooses to observe the interrelation of microflows or flow heterogeneity in a wider area of the cerebral cortex. The technique described herein may solve this problem.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Preparations. Experiments were carried out on male Sprague-Dawley rats (n = 20) weighing 300-350 g. The animals were anesthetized by intraperitoneal administration of 50 mg/kg of pentobarbital. Additional doses of the anesthetic were used intravenously if needed, according to tail pinch and cornea reflex. During the experiment, the rectal temperature was kept constant by use of a thermostat-controlled heated holder. Both femoral arteries and one femoral vein were catheterized [polyethylene (PE)-50 catheter] to measure blood pressure, blood withdrawal, and drug administration. The most critical procedure was the insertion of a catheter into the carotid artery for the dye injection without disturbing the internal carotid flow. Under additional local anesthesia (lidocaine), a midline skin incision was made at the neck. A tracheal cannula was inserted first to secure respiration. Taking care to avoid excessive bleeding and nerve injuries, we moved the muscle layers to the sides and carefully explored the branches of the external carotid artery. Approximately 1.5 cm from the bifurcation, an incision to one-third of the diameter of the external carotid artery was made, and a PE-10 catheter filled with heparinized saline was led proximally into the carotid artery until its tip reached the bifurcation. The external carotid artery was tied to the catheter just above the bifurcation to avoid dye escaping to the external carotid system. The skin was sutured to avoid fluid evaporation. The animals were placed on a head holder with ear bars. A 5-mm-diameter cranial window was trephined above the right temporoparietal cortex, leaving the dura intact. To transilluminate the region of interest (ROI), we inserted a 200-µm-outer diameter (OD) glass-fiber light source from a xenon lamp into the brain parenchyma through a hole drilled 5 mm to the back of the cranial window. An ultraviolet (UV) filter was employed to reduce UV light. The glass fiber was fixed to the edge of the skull with dental cement, so that the light-emitting tip was located beneath the gray matter of the ROI. A laser-Doppler probe (Advance laser flowmeter, 21R, 1-mm probe) was applied above the thinned bone on the contralateral side of the skull. During the drilling process, the skull was cooled with a continuous superfusion of saline. We employed a silicon intensifier target (SIT) camera with a Nikon lens to monitor the transmitted light intensity focused on the pertinent brain surface. When we used saline to measure the MTT, we employed a green filter between the lens and the detector ( = 548 ± 15 nm, one of the isosbestic points of hemoglobin). The camera was linked to an image processor (Argus 10, Hamamatsu Photonics), a personal computer (NEC), a video cassette recorder (ED Beta, Sony), and a Macintosh computer through an 8-bit frame-grabber card (Scion LG-3).

Experiments. In the first group (n = 6), rats were used for basic experiments. In three rats, concentrated carbon black (CB) dye was injected intravenously, and the light intensity was measured continuously in the cranial window by a silicon photodiode (SPD). Blood samples (0.2 ml) were repeatedly taken from the arterial catheter and were anticoagulated with 0.05 ml of 3.1% citrate. The RBCs were hemolyzed to reduce light scattering by addition of 0.5 ml of distilled water. The light extinction of the samples was measured in a transilluminated glass chamber with the use of the same SPD. The in vivo and in vitro CB concentrations were calculated according to the method described in Data analysis. These were not absolute concentrations, however; the maximum was defined as 1, and the relative values were calculated. The corresponding in vivo and in vitro data were compared by use of linear regression analysis. The neurological consequences of the surgical operation and brain damage were estimated in three other rats. Short-cut glass fibers were implanted into the brains of the animals, and, after they regained consciousness, their behaviors were monitored for 8 h.

Data analysis. To analyze dye-dilution curves, we examined whether the relationship between the dye concentration and the output amplitude of the SIT camera coupled to the A/D board adhered to the principle of superposition. We filled a glass chamber with saline containing various concentrations of CB and measured the light intensity on the surface, transilluminating it with the same light source. When the recorded outputs were plotted against the corresponding concentrations in a semilogarithmic chart, the relationship was found to be linear. The linearity in blood as a scattering medium has been previously tested in cats by Tomita et al. (15). To determine the regional MTT, 200 consecutive frames of CB dye transport through a 2 × 2-mm ROI were captured by the Power Macintosh at a rate of 15 frames/s. The captured images were stored in a magnetooptical disk. Sequential frame analysis was performed by MATLAB by employing a Gateway 2000 desktop computer. The ROI was initially divided into 250 × 250 matrices. However, we found that the unit pixels exhibited a low signal-to-noise ratio and inevitably required averaging. Therefore, we averaged each element of the matrix, which represented either a 5 × 5-pixel area (20 × 20 µm) of the ROI (ROI_H) or a 10 × 10-pixel area (~40 × 40 µm; Fig. 1) of the ROI (ROI_L). For the shorter calculation time, 2-D flow maps were calculated on the basis of the latter (ROI_L) system, yielding 2,500 dilution curves. One of the merits of this method was that the flow maps could be recalculated for finer spatial resolution from the stored data. The pixel resolution, however, is not the same as the actual resolution because of the effects of sideward light scatter.

View larger version (190K): [in this window] [in a new window]   Fig. 1.   The first step of the analysis. The region of interest (ROI) is divided into a matrix, with each of its elements representing a 10 × 10-pixel area of the video image or the equivalent of a 40 × 40-µm area of the ROI. The video pixels were averaged to reduce noise level.

(1)

(2)

t

(3)

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

General observations. Small arterioles and veins with diameters ranging from 300-20 µm were well focused with clear contours, suggesting that the light transmitted through the complicated deep-tissue structures illuminated homogeneously only the surface layer, where the vessels were located. Figure 2 shows typical 2-D flow maps. It should be noted that the picture was quite similar to the photograph shown in Fig. 1, although the 2-D flow map was constructed from an entirely different element. The rats that had received implantation of short-cut glass fibers, as described in MATERIALS AND METHODS, awakened within 1.5 h, ultimately becoming alert and attentive and responsive to noise. Two hours later, they recovered completely, showing no signs of neural disturbance. The insertion of the glass fiber seemed to have no affect on their behavior.

View larger version (102K): [in this window] [in a new window]   Fig. 2.   A and B: typical series of two-dimensional (2-D) maps of the reciprocal mean transit time (MTT) values in the ROI at different perfusion pressure levels and after reinfusion. The maps were generated with the use of interpolated colors between the neighboring pixels. The units on both axes represent 40 µm. The arrows indicate pial arterioles. The cerebral blood volume (CBV) did not change during the experiments, so that the reciprocal MTT represented the relative cortical blood flow (CBF). A: flow was maintained at a 75-mmHg perfusion pressure, and arteriolar dilation was noticeable. B: flow was already deteriorated in some parts of the ROI but was maintained in other parts. In both cases, the flow decreased at 50 mmHg and showed a marked increase after reinfusion. Note the heterogeneous flow during the reinfusion phase during the experiment. C: 2-D maps during control and at 10 s and 2 min of 5% CO2 inhalation. The map generated at 10 s reveals a heterogeneous relative flow increase along the vessels. The relative flow further increased and became homogeneous after 2 min of inhalation. Note the intensified mixing in the vein indicated by the disappearance of the dark central area.

In vivo tissue and in vitro blood dye concentration. We observed a linear relationship between the in vitro and in vivo measured concentrations. Figure 3 shows the scatter plot of the corresponding values and the best-fitting line (y = 0.93x + 0.12, r² = 0.92, P < 001).

View larger version (24K): [in this window] [in a new window]   Fig. 3.   In three rats, concentrated carbon black (CB) dye was injected in a stepwise manner intravenously. Tissue and blood CB concentrations were calculated according to the routine described in MATERIALS AND METHODS. A linear relationship was found between the in vitro and in vivo measured CB concentrations. The scatter plot of the corresponding values and the best-fitting line are shown (y = 0.93x + 0.12, r2 = 0.92, P < 0.001). Both axes represent relative dye concentrations.

Exsanguination and reinfusion. The arterial blood gas values were within the normal range during the experiments: arterial partial pressure of O₂ (Pa_O2), 97 ± 7 mmHg (means ± SD); arterial partial pressure of CO₂ (Pa_CO2), 39.7 ± 1.7; and pH, 7.37 ± 0.04. In the control group, the average mean arterial blood pressure was 109 ± 11 mmHg. The average MTT of CB in the total ROI was 2.11 ± 0.67 s in the control; increased to 2.61 ± 0.79 and 3.26 ± 1.18 s at mean blood pressures of 75 and 50 mmHg, respectively; and decreased to 1.67 ± 0.28 s after reinfusion. The respective values of LDF were 21.4, 17.4, 15.9, and 27.9. Neither the transmitted-light intensity nor the mass signal of the laser-Doppler flowmetry changed throughout the experiment. Figure 4 shows the relative changes in the measured parameters. Figure 2, A and B, shows the obtained 2-D flow maps in the ROI in two control experiments at BPs of 75 and 50 mmHg and after reinfusion. In Fig. 2A, two pial arterioles are visible in the flow maps, descending from the top. They appear to be dilated at a BP of 75 mmHg, as the two yellow bands are wider. During the reperfusion phase, the arterioles were still dilated, with the bright color indicating reactive hyperemia. Between the two arterioles, a pial vein was clearly discernible, especially during hypotension (BP 50 mmHg), when the flow decreased in the vessel, and during the reinfusion phase, when the reactive hyperemia was visible in the vein as well. The tissue perfusion remained constant when the BP decreased to 75 mmHg but decreased markedly in the tissue flow at a BP of 50 mmHg. During the reinfusion phase, an increase in the tissue perfusion was visible, but its degree was quite heterogeneous throughout the ROI. In Fig. 2B, two tortuous arterioles are seen emerging from the left (arrows) as a pial vein exits the ROI on the right. In this case, the tissue perfusion became quite heterogeneous at a BP of 75 mmHg: some regions showed good autoregulation, maintaining the regional blood flow, but the perfusion decreased in other areas of the ROI. The tissue perfusion remained heterogeneous even during the reinfusion phase. It should be noted that the "laminar" flow was visualized in the vein by use of this technique during reinfusion as two distinct layers of different flow velocities. In Fig. 5, the regional differences in the CBF autoregulation are more visible. Figure 5 was created by calculation of the relative changes in the local CBF values. The light tones represent those areas of the ROI where the CBF_rel remained unchanged at a perfusion pressure of 75 mmHg, during which time the flow decreased in the dark regions. Figure 6 shows representative arterial, venous, and tissue dilution curves obtained during the experiment. It should be noted that the distance between the peaks of the arterial and venous curves correlated well with the measured CBF. The analysis of MTT values in avascular areas, arteries, and veins from four experiments is presented in Fig. 7.

View larger version (33K): [in this window] [in a new window]   Fig. 4.   Relative changes (means ± SE, calculated as measured value/control) in the laser-Doppler flow (LDF), 1/MTT, and "red blood cell mass" values during exsanguination at 75 and 50 mmHg and after reinfusion. The CBV was derived from the light intensity, which did not change during the exsanguination-reinfusion procedure.

View larger version (103K): [in this window] [in a new window]   Fig. 5.   Relative changes in the CBF during hypotension (75 mmHg) compared with the control state. The light tones represent well-autoregulating areas, and the dark grades indicate nonautoregulating areas. The relative change was calculated in each pixel as control MTT value/MTT value at 75 mmHg.

View larger version (29K): [in this window] [in a new window]   Fig. 6.   Representative dilution curves from arteries, avascular areas, and veins. The curves represent the light intensity in the given pixels as a function of time. The curves were "well behaved," with a very low noise level even when they were obtained from a small area of 40 × 40 µm2. The y-axis indicates the measured light intensity in arbitrary units, and the x-axis is the time in seconds. A: control. B: hypotension, 75 mmHg. C: hypotension, 50 mmHg. D: reinfusion.

View larger version (40K): [in this window] [in a new window]   Fig. 7.   MTT values measured in arteries, avascular areas (tissue), and veins. The error bars indicate SE (n = 4).

CO₂ inhalation. In this group (n = 6), the control MTT ranged from 0.98 to 1.49 s, with an average of 1.41 ± 0.14 s (mean ± SD). At 2 min of CO₂ inhalation, the MTT decreased to 0.76 + 0.22 s. This indicates an 85% increase in the CBF. Figure 2C shows flow maps from three consecutive states of CO₂ inhalation. The first map represents the control state. In the central area of the map, a big cortical vein is visible, and the flow in the surrounding tissue is relatively homogeneous. The second map reflects the flow alteration 10 s after the initiation of the CO₂ inhalation. The tissue flow became very heterogeneous during this phase, increasing initially in the area surrounding the feeding arteries. The map reveals different flow layers in the pial vein. The third map shows the highly increased CBF after 2 min of CO₂ inhalation. During this phase, the pial arteriole, which was not visible in the previous maps, is clearly discernible on the right side of the map. The mixing in the vein intensified, as indicated by the disappearance of the dark central area.

Comparison between laser-Doppler flowmetry and the new optical method. During the exsanguination-reinfusion procedure, we performed 55 MTT measurements in eight rats. In each case, except for the controls, we calculated the relative change as CBF_measured/CBF_control. The results were paired with the corresponding LDF_measured/LDF_control values. The statistical analysis yielded a significant correlation between the LDF and the optical flow measurement (Pearson coefficient = 0.847, P < 0.001). Figure 8 shows the scatter plot of the data set and the best-fitting curve, obtained by linear regression analysis (y = 0.93x + 0.05). The distribution of the calculated differences (CBF_rel  LDF_rel) was virtually normal, and the average mean difference was 0.03 ± 0.05% (means ± 95% confidence interval), indicating a 3% overestimation of the CBF alteration as measured by the optical technique compared with the LDF. An analysis of the discrepancy in the function of relative CBF change revealed a significant divergence between the two methods when the measured CBF was below 80% of the control (Fig. 9). In this case, the mean difference was 14% (P < 0.001); our method resulted in lower CBF values. When the CBF decrease was <20% and when it increased during the reinfusion phase, the differences were not significant. The limits of agreement were 27 and 21%.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Linear regression analysis between the corresponding relative (r) flow values, calculated from the reciprocal MTT and the LDF. The best curve was described by y = 0.91x + 0.12 (r2 = 0.84, P < 0.001).

View larger version (22K): [in this window] [in a new window]   Fig. 9.   Mean difference ± SE between the two techniques at various relative flow values. The discrepancy was significant when the flow was <80% of the control (P < 0.001, n = 8), with our method measuring lower CBF values.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Tissue damage. One possible criticism of our method might be that it is invasive. The present method requires insertion of a glass fiber of 200-µm OD into brain tissue, which inevitably causes brain tissue damage. The diameter is less than that of a hydrogen electrode used for CBF measurements or as a probe during microdialysis and is almost equivalent to that in electrodes used for measuring deep action potentials, a technique also requiring insertion of the instrument into brain tissue. However, it should be noted that the latter three methods measure the flow at the sensitive surface of the probes, i.e., the most damaged portion facing the sensor. On the other hand, our method measures the undamaged layer of the cortex sandwiched between the light source and the detector. One other factor that must be taken into consideration is the spreading depression (SDE). We confirmed that SDE after insertion of a microglass pipette spread in a wavering manner at a speed of 1.6 mm/min, and the cortex was stabilized within 30 min (unpublished observation). Therefore, measurements performed 60 min after the insertion were considered free from the artifacts generated by SDE waves. It is also known that the SDE never spread to the contralateral side where the laser-Doppler probe was employed. As confirmed by the results, paired values of CBF obtained by the laser-Doppler technique and the present optical method correlated well, ruling out the influence of SDEs on the measurements. Autoregulation was partially maintained, and vascular responses to CO₂ were comparable with those reported in the literature. Despite the damage, it is concluded that the present method can measure CBF within the limits of these experimental conditions.

Light scattering. Because the tissue consists of numerous particles and structures in the deep layer, transmitted light from the glass fiber is subjected to light-pass change by reflection and refraction and, therefore, scattering. Tomita et al. (16) noticed a complete mixing of transmitted light beams when they characterized the three-dimensional light scattering in blood in a thin transparent tube. Even though a parallel beam was projected through a specific point on the tube, the light emerged from the surface of the blood at all angles from the injected plane (detected by a small photodetector sliding on the surface). The blood in the tube was as luminous as a glow lamp, resulting in scattered light throughout the cerebral cortex on the brain surface.

In vivo tissue and in vitro dye concentration. We investigated whether the intravascular dye concentration could be determined by use of the Lambert-Beer approach. Because we believed that the brain CB filling was a function of the arterial CB concentration, we compared the in vivo data, which were measured in the highly scattering medium, with those obtained in vitro from the arterial blood samples. To reduce light scattering, we decided to hemolyze the RBCs. Because plasma has an even lower scattering coefficient, however, the repeated blood samplings might have altered the systemic hematocrit, and thus we chose to measure the total light absorption of the blood. The linear relationship shown in Fig. 3 confirms the validity of the calculated MTT values in the brain tissue. However, the in vivo measured concentrations were higher than the corresponding in vitro values, with the best-fitting curve intercepting the y-axis at 0.12. This might be the result of higher scattering losses in tissue and CB pooling in the veins or CB adhesion on the endothelial surface. Cerebral vasodilatation due to slight exsanguination could have contributed to the discrepancy.

Autoregulation. We found that the mean CBF was rather maintained when compared with decreases in the mean BP. However, we found heterogeneous distribution of areas well autoregulated and nonautoregulated, as shown in Fig. 5. This visualization of autoregulation in such a small area would be the first occurrence in the literature, although Leninger-Follert and Lübbers (9) reported heterogeneous perfusion in capillaries, according to the metabolic requirement. Our observation that the cerebral blood volume (CBV) did not change during the exsanguination-reinfusion procedure is in accordance with the data reported in the literature (12, 14). Hudetz et al. (6) reported that the flow autoregulation in the cerebral microcirculation was effectuated through the preservation of RBC capillary transit time instead of the recruitment of new capillaries. It must be kept in mind that CB is a plasma indicator; thus the maintenance of the RBC velocity could remain unrecognized if the plasma flow decreased selectively. This could explain the slightly higher discrepancy between the two methods when the CBF was <80% of the control value (Fig. 9). This point should be studied in the future. The present method may provide a versatile tool for that purpose.

CO₂ effects. For the analysis of the CO₂ inhalation, we used saline as a nondiffusible tracer. The blood-brain barrier is relatively impermeable to sodium under physiological conditions. The difference between saline and CB, as we emphasized in MATERIALS AND METHODS, is that saline is an RBC indicator and CB is a plasma indicator. Thus we can expect lower MTT values measured with saline as a tracer, because RBC moves at a more rapid rate in the microcirculation (13). Our results are consistent with this, as the average MTT was 1.41 ± 0.06 s with saline as a tracer and 2.11 ± 0.24 s with CB as a tracer, indicating that the RBC rate was ~50% faster than that of plasma. The systemic inhalation of CO₂ caused an 85% increase in the CBF. The flow maps presented in Fig. 2C show that the flow alteration was diffuse throughout the ROI but was heterogeneous in the early phase. The average rate of increase of microflow was again in accordance with previous reports (8, 14).

Comparison with an accepted method. A comparison of our technique with laser-Doppler flowmetry reveals a very low mean difference, but with relatively wide limits of agreement. This may be due to practical and theoretical dissimilarities between the two techniques: the CB is a plasma dye, and its transport refers to plasma flow; meanwhile, the LDF measures RBC velocity. Any shift in the plasma flow-to-RBC flow ratio can cause a bias in the comparison. The surface areas over which the measurements were performed were different: at the optical method, the ROI was 2 × 2 mm; on the other hand, we employed a 1-mm-diameter LDF probe. MTT values from larger veins and arteries contributed to the optical CBF measurement; however, the LDF probe was positioned above the "avascular" area. A more important difference is the contribution of deeper cortical layers to the measurement. Because the light source was positioned below the cortex in the case of the optical technique, the signal provided information about flow in layers IV-VI. The LDF, nevertheless, focuses on the top layers. Taking into account these factors, we think that results of the comparison are convincing enough to validate our method.