2.1.

Animal Preparation

The animal experiments were performed in accordance with the Guidelines for Animal Experimentation from the Ethics Committee of Kyoto University (Permit Number: Kyo Med 15566 and 16202). All surgeries were performed with the rats under anesthesia, and all efforts were made to minimize suffering. Seven male Wistar rats (9 to 12 weeks old) (Japan SLC, Japan) weighing 200 to 270 g were used. The rats were initially anesthetized with isoflurane (2.0% gas) in air through a gas anesthesia mask during a tracheotomy. After the tracheotomy, mechanical ventilation (Independent Dual Output Small Animal Ventilator; NEMI Scientific) was adjusted to maintain physiological parameters within normal ranges under isoflurane anesthesia through the endotracheal tube (0.5% to 1.5% gas). The left femoral artery was cannulated using polyethylene tubing (SP45; Natsume Seisakusho Co., Ltd., Japan) to connect a pressure transducer (KN-211; Natsume Seisakusho Co., Ltd., Japan) for monitoring the systemic blood pressure. Body temperature was maintained at 36.5°C to 37.0°C using a heating pad, and a rectal thermometer (FST-HPS; Fine Science tools, Canada) was used to provide feedback control. Another type-T thermocouple was also inserted 7-cm deep in the rectum to measure body core temperature (RET-2; Physitemp Instruments Inc.). Mannitol ($2\mathrm{g}/\mathrm{kg}$, i.p.) [D-(-)-Mannitol; nacalai tesque, Japan] was administered to prevent brain edema after removing the skull. The rats were fixed to a stereotaxic holder (SR-5R-HT; NARISHIGE, Japan). The skin on top of the skull was removed to expose the location of the bregma. A $3\text{-}\times 3\text{-}\mathrm{mm}$ square cranial window 2- to 5-mm posterior and 2- to 5-mm left lateral to the bregma was drilled in the skull to expose the somatosensory cortex leaving the dura mater intact. During the drilling process, the skull was cooled using a saline solution to prevent heat damage. Cannulation of the caudal vein was performed to administer $\alpha$-chloralose during measurement. Following these procedures, the stereotaxic holder that contained the anesthetized rats was moved to an imaging system (Fig. 1).

Fig. 1

Schematic of the simultaneous imaging system. A near-infrared laser and a CCD camera were used to construct a laser speckle imaging system to evaluate the 2-D distribution of cerebral blood flow changes. An infrared camera measures the temperature distribution of the rat cortex through a cranial window. A computer is controlling and synchronizing image acquisition from the two cameras. The imaging system is covered with 20-mm-thick insulation boards. A part of the cover is a clear acrylic board, which enables observation of the interior. A heater controls and maintains the temperature in the box. A black curtain covers the box to reduce light source noise from outside during measurement.

2.2.

Imaging System

2.3.

Data Collection and Analysis

IR images were captured via an image capture board (NI PCI 1424; National Instruments) and saved on a computer as binary data using in-house software written using LabVIEW (National Instruments). The CCD images for the laser speckle imaging were transferred to the computer using the Gigabit Ethernet communication protocol. These CCD images were converted to the blood flow images using laser speckle imaging developed by Cheng et al.²⁶ In this method, at first, the speckle contrast ($C$) is calculated as

Changes in brain temperature and cerebral blood flow were calculated using previously described processing methods^6,27 implemented in ImageJ (National Institutes of Health). We executed the alignment of the blood flow image ($348\times 260\text{pixels}$) to the brain temperature image ($320\times 256\text{pixels}$) using an affine transformation to evaluate the same region. The affine transformation was computed using the MATLAB Image Processing Toolbox function, affine2d (Mathworks). This transform is obtained from the relation between three distinct points. To use the affine2d function, the $3\times 3$-affine matrix was defined from each of the three coordinates of the aluminum vertex points. The image transformation was applied to the stack of blood flow images. The temperature images and transformed blood flow images were trimmed using a region of interest of $150\times 180\text{pixels}$. Each pixel represents an area of $12\times 12\mu \mathrm{m}$ in the images.

Temporal analyses of cerebral blood flow and temperature were performed by averaging intensity values within the entire image region of interest ($150\times 180\text{pixels}$) on the stack image using the ImageJ, as the image included some misregistration caused by nonlinear deformations such as vessels. In addition, to investigate the relationship between the blood flow and the temperature, we averaged these temporal changes over the entire image region from seven animals.

To visualize the relationship between the cerebral blood flow and the brain temperature, correlation maps between the blood flow and temperature changes were created. Pearson correlation coefficient values were calculated on the basis of one pixel in a time sequence for 5 min. Each pixel $(i,j)$ in a correlation map is represented by a correlation value $r_{ij}$ ($-1\le r_{ij}\le 1$). A two-dimensional (2-D) distribution of the correlation values at each pixel depicted the correlation map of the 5 min changes between the blood flow and the temperature. We created the maps in every 5 min. The original images were processed by a denoising filter in both temporal and spatial directions (Gaussian Blur 3D: ${\sigma }_i=2\text{pixels}$, ${\sigma }_j=2\text{pixels}$, $t=100\text{frames}$) using ImageJ before a calculation. The correlation value was calculated using in-house software written in LabVIEW. The relationship between the cerebral blood flow and the brain temperature was visually inspected on representative vessels.

2.4.

Protocols

Changes in brain temperature were achieved using two different anesthetics (isoflurane and $\alpha$-chloralose). The method of brain temperature control was the same as used in Zhu’s study.²⁵ The time course of this protocol is shown in Fig. 2. The rat was first anesthetized with 2.0% isoflurane. We waited at least 20 min until the physiological parameters stabilized following the 2.0% isoflurane inhalation. Recording of the brain temperature and cerebral blood flow was started under baseline conditions for 10 min after the steady state (isoflurane only). Following the baseline recording, the anesthesia was added to a $5.0\mathrm{mL}/\mathrm{kg}$ intravenous bolus of $\alpha$-chloralose and administered over 1 min with isoflurane followed by continuous intravenous infusion at $5.0\mathrm{mL}/\mathrm{kg}/\mathrm{hr}$ (isoflurane + $\alpha$-chloralose). The isoflurane anesthesia was removed 10 min after the $\alpha$-chloralose infusion ($\alpha$-chloralose only). We maintained the $\alpha$-chloralose condition for 50 min. Image actuations were divided into 14 trials. Each trial took 5 min, including 10-s blanks at the beginning and end, which gave a total recording time of 280 s. The missing data during the blanks were interpolated linearly. Therefore, each trial had 5600 temperature images and 5600 blood flow images, respectively. To investigate responses related to the bolus injection (at 0 min in Fig. 2) and elimination of the inhalation anesthesia (at 10 min in Fig. 2), the middle of the second and fourth trials matched the time points when each event occurred. We applied the Gaussian filtering in both the temporal and spatial directions to the temperature and blood flow images for smoothing using ImageJ.

Fig. 2

Study protocol. The colored horizontal arrow shows how the anesthesia was administered. The black vertical arrows indicate the timing of each event. The middle crossbar shows the sequence of image actuation. The other crossbar of the bottom indicates a detail time sequence of an image recording trial.

2.5.

Statistical Analysis

Experimental data are expressed as $\text{means}\pm \mathrm{SD}$, and error bars in graphs indicate SEs. The statistical significance of the changes was assessed by one-way repeated measures (ANOVA) for temporal changes and the post hoc Dunnett’s test (multiple comparisons versus a control). Data were analyzed using R (Version 2.15.2). The level of statistical significance was set at $P<0.05$.

3.1.

Temperature Controls

The averaged time courses of the rat rectal temperature as core temperature and the ambient temperature in the box covering the imaging systems during experiments are shown in Fig. 3. These temperatures remained extremely consistent (Rectal: $36.74°\mathrm{C}\pm 0.45°\mathrm{C}$, Box: $32.73°\mathrm{C}\pm 1.16°\mathrm{C}$) throughout the experiments.

Fig. 3

Temporal change in rat rectal temperature averaged from seven animals and ambient temperature in the box averaged from seven trials. Values are means (line) ± SE (a fill pattern with similar color of the line). The shade in the graph indicates overlapping usage of isoflurane (IF) and $\alpha$-chloralose ($\alpha \mathrm{C}$).

3.2.

Physiological Changes

The mean arterial blood pressure, the calculated heart rate, and respiratory rate from the femoral artery blood pressure data are shown in Fig. 4. The mean arterial blood pressure [Fig. 4(a)] decreased for 10 min significantly ($P=0.027$) from the initial control value following infusion of $\alpha$-chloralose (at 0 min) with isoflurane anesthesia (from $75.21\pm 14.9$ to $57.13\pm 9.71\mathrm{mmHg}$). However, blood pressure increased immediately after suspension of isoflurane usage (at 10 min) and reached a peak value ($107.51\pm 20.6\mathrm{mmHg}$) at 16 min. After the peak, the blood pressure remained almost stable. The heart rate calculated from the arterial blood pressure [Fig. 4(b)] increased significantly ($P=0.042$) for 10 min after the infusion of $\alpha$-chloralose (from $345.39\pm 18.1$ to $378.56\pm 28.8\text{beats}/\min$) while the blood pressure decreased. Subsequently (10 min after the infusion), the heart rate decreased immediately by about $20\text{beats}/\min$ while the blood pressure increased. From 30 min after the infusion, the heart rate slightly increased. The respiratory rate calculated from the blood pressure data [Fig. 4(c)] remained substantially constant because of the mechanical control.

Fig. 4

Temporal changes in the rat mean arterial blood pressure, heart rate, and respiratory rate averaged from seven animals. Values are means (line) ± SE (a fill pattern with similar color of the line). The shade in the graph indicates overlapping usage of IF and $\alpha \mathrm{C}$.

3.3.

Blood Flow and Brain Temperature Images

Figure 5(b) shows an example of cerebral blood flow images mapping speckle flow indices that describe the blood flow changes in the cranial window. Figure 5(c) also shows the acquired IR thermal images with mapping of the absolute temperatures in the same region of the same rat cerebral cortex during anesthesia while switching from isoflurane to $\alpha$-chloralose in the subject. The percent change in the cerebral blood flow obtained by subtracting and dividing the baseline from each image, and the brain temperature change obtained by subtracting the baseline from each image in Fig. 6. We could not observe the vascular structure on the temperature images, although the blood flow images provided clear 2-D structural information about the cortex vasculature. To understand the spatial information based on the vascular structure on the temperature images, the images showing temperature change were overlaid on a digital camera image of the cranial window at a baseline condition [Figs. 5(c) and 6(b)]. The values of the SFI for the large vein indicated in blue in Fig. 5(a) were higher than those for other vessels throughout the experiment. In addition, we observed that the index values in the center of the vein were higher than the values near the vessel wall from baseline to 12 min after the $\alpha$-chloralose injection. Just after the injection of the $\alpha$-chloralose [from 0 to 2 min in Fig. 5(b)], the value of the SFI increased transiently in the vessels. After dissipation of this overshoot, under both isoflurane and $\alpha$-chloralose anesthesia, the blood flow decreased but did not show dramatic changes. During only the $\alpha$-chloralose anesthesia, the blood flow was significantly decreased, especially in the cerebral vessels. The running of red and blue parts in parallel near the artery resulting from a misregistration appeared in Fig. 6(a) from 15 to 60 min. The artery indicated in red in Fig. 5(a) became thinner and changed shape in Fig. 5(b) accompanied by reduction in the blood flow. In our imaging technique, it is difficult to adjust the misregistration of the artery. As a result, the misalignment of the vessel led to misunderstanding that the percent changes in the blood flow increases that appeared in the artery during blood flow had decreased totally.

Fig. 5

(a) A representative camera image of the cranial window. The image allows a characteristic vein (blue) and an artery (red) to be easily distinguished. (b) Averaged images of speckle flow indices indicate cerebral blood flow changes in the cerebral cortex accompanying the anesthesia for one rat. The figures on each image indicate the elapsed time from the beginning of the $\alpha$-chloralose injection. (c) Averaged images of temperature in the cerebral cortex accompanying the anaesthesia for one rat. The figures on each image indicate the elapsed time from the beginning of the $\alpha$-chloralose injection. The different colors of the framework indicate the status of anesthesia: isoflurane (blue), $\alpha$-chloralose (red), and mix (purple). The camera image taken at baseline is overlaid on each image to add vascular structural information.

Fig. 6

(a) A representative sequence of images showing the percent changes in cerebral blood flow in response to the anesthesia from one rat. (b) sequence of images showing the temperature differences from the baseline image in response to the anesthesia from one rat. The different colors of the framework indicate the status of anesthesia: both isoflurane and $\alpha$-chloralose (purple) and $\alpha$-chloralose (red). The camera image taken at baseline is overlaid on the temperature images to add vascular structural information. The figures on each image indicate the elapsed time from the beginning of the $\alpha$-chloralose injection.

The distribution of the brain temperature varied, and the temperature of the center of the cortex was lower than the temperature of the surrounding area at baseline [Fig. 5(c)]. The temperature of the cortex appeared to decrease while maintaining the temperature distribution during both anesthesia usage (from 0 to 9 min), irrespective of the vascular arrangement. During only the $\alpha$-chloralose anesthesia, the temperature also decreased $\sim 0.5°\mathrm{C}$ until 30 min. From 40 to 60 min, an $\sim 0.1°\mathrm{C}$ increase in the brain temperature was observed. The temperature differences from the baseline [5 and 10 min in Fig. 6(b)] indicate a reduction in the temperature. The decreases in brain surface temperature were observed in the center of the region [from Fig. 5(c)]. In fact, however, the temperature decreased not only in the center of the cortex, but also in the surrounding area until 30 min [from Fig. 6(b)]. During the $\alpha$-chloralose anesthesia, there were small areas in which the temperature differences near the center of the cranial window were smaller than those in the surrounding areas [Fig. 6(b)]. However, the areas with less temperature variation did not match the vascular distribution. Following the temperature reduction until 30 min, the brain temperature increased again. The pial vein region in the rat brain showed a clear temperature decrease relative to the region surrounding the vein from 25 to 60 min.

3.4.

Relationship between Changes of Blood Flow and Temperature

Figure 7(a) shows the interactive relationship between the cerebral blood flow changes and the brain temperature changes over time. Each data point indicates the ratio of the change from the baseline value. In our study, it is difficult to indicate the averaged images from seven animals as vessel structures on the cortical surface are quite different among them. Instead of showing the averaged images, Figs. 7(b) and 7(c) show that temporal changes in the brain temperature and the percent changes in cerebral blood flow averaged over the entire brain region from seven animals, respectively. Both the blood flow and temperature increased transiently from the initial control value (about 13% and 0.08°C increase, respectively) following the $\alpha$-chloralose infusion. The initial increases reached a peak in 1 min. The mean temperature decreased during the isoflurane and $\alpha$-chloralose anesthesias after the peak, and the value at 10 min was significantly ($P=0.042$) lower than the control value. The mean blood flow change also decreased by about 10% between the value at the peak and at 6 min after the $\alpha$-chloralose injection under the isoflurane and $\alpha$-chloralose anesthesias. The change in the blood flow remained flat from 6 to 10 min while the temperature continued to decrease. After the isoflurane anesthesia was removed, the blood flow indicated significant exponential ($P<0.001$) reduction from 10 to 60 min. We also observed similar temperature reduction until $\sim 30\min$. However, following this reduction, the brain temperature started to increase gradually to 60 min. The relationship between the cerebral blood flow and brain temperature was not correlated partly. During the isoflurane and $\alpha$-chloralose anesthesia, at first, both blood flow and temperature increased. Following these increases, we observed a discrepancy between the decrease in the cerebral blood flow change and the increase in brain temperature change, and the direction of change was inclined upward left [arrow (1) in Fig. 7(a)]. After this decoupling, we confirmed that both the blood flow and temperature were decreasing. Following these decreases, the steady state of the blood flow and the decrease in temperature were appeared, and the direction of change was downward [arrow (2) in Fig. 7(a)]. During the $\alpha$-chloralose anesthesia, until 20 min after the $\alpha$-chloralose injection, both the blood flow and the temperature were decreasing. However, from 20 to 60 min after the injection, we also observed another discrepancy between the decrease in cerebral blood flow change and the increase in brain temperature change, and the change direction was also inclined upward left [arrow (3) in Fig. 7(a)].

Fig. 7

(a) Relationship between temporal change in cerebral blood flow and brain temperature averaged over the entire image region from seven animals. Three black arrows [(1), (2), and (3)] indicate directions of travel and ranges of decoupling between the temperature and the blood flow. (b) Temporal change in temperature averaged over the entire image region from seven animals. (c) Graph illustrating the percent change in cerebral blood flow averaged over the entire image region from seven animals. Values are means (line) ± SE (a fill pattern with similar color of the line). The shade in the graph indicates overlapping usage of IF and $\alpha \mathrm{C}$.

3.5.

Correlation of Brain Temperature Changes with Cerebral Blood Flow Changes

Figure 8 shows maps indicating the spatial distribution in the correlation coefficients between the cerebral blood flow and brain temperature every 5 min. The images were overlaid on a digital camera image of the cranial window at a baseline condition. As a result of a visual inspection for the relationship between the blood flow and the temperature, correlations were observed along the vessels indicated in red and blue in Fig. 8(a). During the mixed anesthesia (5 to 10 min), the correlation coefficients were mostly low because the blood flow was almost constant and the temperature decreased. However, the correlation coefficients for the areas surrounding the vessels tended to be higher than those of the other areas. After removing the isoflurane (10 to 15 min, 15 to 20 min), there were remarkable positive correlations between cerebral blood flow and brain temperature changes, but we found that the correlation coefficients near the vessels became smaller from 15 to 20 min. From 20 to 45 min, there were no strong correlations between the temperature and blood flow. Moreover, the correlation coefficients in the vessels were negative from 20 to 25 and 25 to 30 min. On the other hand, we found positive correlations for the vessels from 30 to 35, 35 to 40, and 40 to 45 min. From 45 to 50 and 50 to 55 min, only weak negative correlations were observed. We observed that there were remarkable negative correlations overall from 55 to 60 min. These negative correlations indicated decoupling between the changes in brain temperature and cerebral blood flow.

Fig. 8

(a) A representative camera image of the cranial window. Colored parts on the representative vessels (red—artery, blue—vein) indicates positions where some characteristic correlations are observed. (b) Images mapping the Pearson correlation coefficients between brain temperature changes and cerebral blood flow changes for every 5 min. The camera image taken at baseline is overlaid on the temperature images to add vascular structural information. The figures on each image indicate the time interval for calculating the Pearson correlation coefficient.