2.1.

LSI and fPAM System

The schematic of our fPAM system is shown in Fig. 1(a). In brief, the system consists of a tunable dye laser (Cobra-Stretch, Sirah), a computer controlled scanner integrated with dark-field illumination and high-frequency ultrasonic detection, and a digital oscilloscope.²⁹ The photoacoustic signals from the ultrasonic transducer were sampled by the digital oscilloscope, which was triggered by synchronized pulses produced by a fast photodiode. Simultaneously, the synchronous pulses from the digital oscilloscope were sent to the motion controller to transmit the command to the scanner, the raster scan across the $x\text{-}y$ plane was performed. To realize multi-wavelength fPAM, the wavelength of laser should be synchronously tuned in scanning. Another photodiode was used to monitor fluctuation of the deposition of laser energy, and the incident laser energy density on the sample surface is about $8\mathrm{mJ}/{\mathrm{cm}}^2$, which is well within the ANSI safety standard ($20\mathrm{mJ}/{\mathrm{cm}}^2$) in the spectral regions used here. The measured lateral and axial resolution are 45 and 15 μm, respectively. The maximum imaging depth in chicken breast tissue is 3.1 mm.

In our previous works, we designed a LSI system and obtained high imaging quality by improving the laser speckle contrast analysis method.^20,38 As shown in Fig. 1(b), the system is made up of laser source, beam collimation and adjustment, photoelectric imaging, data collection, and image processing. The beam from the He–Ne laser illuminated the cranial window with an incident angle of about 30 deg after magnification and collimation. A series of speckle patterns were captured by a 12 bit charge coupled device (CCD) camera (Pixelfly qe, PCO; $\text{pixel size}=6.45\times 6.45\mu \text{m}$) for subsequent data processing. The CCD camera was attached to a stereomicroscope (Z16 APO, Leica; working distance 97 mm) with a magnification of $1.25\times$, consequently, the spatial resolution was estimated to be about 20 μm. To guarantee accurate velocity measurement, the aperture diaphragm is well adjusted to ensure that the average speckle size is about two pixels, and the variable attenuator is also well adjusted to ensure the light intensity matches the dynamic range of the CCD camera.

Fig. 1

Schematic of the experimental system and photographs of the animal model. (a) Schematic of the fPAM system and (b) the LSI system; and (c) the photograph of a rat with a cranial window and (d) the UCCA legation of a rat.

2.2.

General Surgical Preparations

Male wistar rats ($n=10$) weighting 200 to 250 g, obtained from Hubei center of disease control and prevention, were used in the experiments. All experimental procedures were performed according to the institutional animal care and use committee of Hubei province. The rats were intraperitoneally anesthetized with a mixture of a-chloralose and urethane, and an additional one-fifth of the initial dose was administrated per hour to maintain a state unresponsive to toe pinching. Body temperature of the rats was kept constant at $37\pm 0.5°\mathrm{C}$ by utilizing a feedback-controlled heating pad. Left femoral artery was cannulated for continuous measurement of arterial pressure and blood gases. The anesthetized rat was placed in a stereotaxic apparatus. A midline scalp incision was made, and then an about $4\times 5{\mathrm{mm}}^2$ cranial window at 3.5 mm lateral to and 3.5 mm posterior to the bregma was made with intact dura using a high speed dental drill. Normal saline was used to bath the skull for cooling during drilling. The cranial window was bathed with artificial cerebrospinal fluid to maintain physiological environment of brain and prevent cortex from drying. The photograph of a rat with a cranial window is shown in Fig. 1(c).

2.3.

CCA Ligation

After the craniotomy, the rat was allowed to rest for 30 min. Then, the ipsilateral CCA ligation was performed. Since a fast occlusion and recanalization of CCA in situ is needed in the experiments, the common ligation method is not adaptive. Therefore, we utilized a special method to realize the ligation quickly and recovery simply. After a midline cervical incision, the UCCA was exposed and freed from its sheath and vagus nerve, which was carefully separated and maintained. A 4-0 silk suture was placed around the exposed UCCA, and the two ends of the suture were both pulled through a transparent silicone tube with an inner diameter of 0.8 mm and a length of 2 cm. When performing the ligation, draw the suture until the CCA is pulled into the tube for about 2 mm. Thus, the blood flow in the UCCA to ipsilateral brain was interrupted due to the stress of the tube and the pulled suture. When the CCA was completely occluded, we find that the part of CCA in the tube turned white, as shown in Fig. 1(d). After the occlusion, the tube was pulled back from the CCA for recanalization. To conduct occlusion and recanalization quickly and to avoid damage to the vessels, the tube and the suture were lubricated.

2.4.

Experimental Protocol

In each trial, the craniotomy and the ipsilateral CCA isolation in a rat was randomly arranged. Then, after a stabilization period of 30 min, the LSI and fPAM were performed, respectively. The rat was placed in a stereotaxic apparatus, LSI started at 5 min before the UCCA ligation, as a base level, and continued for 2 to –3 min after the ligation. When LSI was completed, the blood flow was restored immediately by release the ligation. After a recovery period of about 2 h, fPAM was performed. Firstly, structural images of the cortex blood vessels were obtained by a two dimension (2-D) single wavelength (584 nm) scan. Then, several blood vessels were selected to perform dual wavelength continuous b-scans for functional monitoring, the observed time points was set as same as the experiments with LSI. In our studies, the scanning step was 30 μm. Although statistical error is reduced by arranging the LSI and fPAM observation of the same rat in a random order, the time interval of the observations should be the same to guarantee the validity and reproducibility of changes of the parameters.

2.5.

Image Analysis

Photoacoustic imaging, which images the distribution of optical energy deposition (or optical absorption distribution) in tissue based on the detection of ultrasonic wave induced by absorption of pulsed laser energy, presents intrinsic physiologic specific absorption of tissues.^28,39 In this paper, the maximum-amplitude projection (MAP) along the $z$ axis was used for 2-D image reconstruction. For functional parameter monitoring, a series of continuous images were acquired by sequential dual-wavelength line scanning of the same selected blood vessels. Before calculating the ${\mathrm{SO}}_2$ value, the threshold method was applied to the MAP image to eliminate the pixels with low signal-to-noise ratio. The threshold was selected as the intensity of the half-maximum of the chosen vessels, and then the pointwise ${\mathrm{SO}}_2$ value was calculated with the multi-wavelength method. Consequently, the ${\mathrm{SO}}_2$ value of each vessel was defined as the average value of ${\mathrm{SO}}_2$ located in this vessel. Here, we describe the multi-wavelength method for ${\mathrm{SO}}_2$ calculation simply. In brief, when oxyhemoglobin (${\mathrm{HbO}}_2$) and deoxyhemoglobin ($\mathrm{HbR}$) are treated as the dominant absorbing compounds in blood, the ${\mathrm{SO}}_2$ image could be calculated using least-squares fitting as:³⁴

LSI is an optical technique for the imaging of blood flow velocity. The relative flow velocity is calculated from laser speckle contrast, which is defined as a ratio of standard error and mean amplitude of optical intensity of the speckle pattern. The speckle pattern is formed by interference of diffused laser and is modulated by moving particles.^20,38,40 The laser speckle temporal contrast analysis method was used to achieve high spatial resolution. A velocity image was derived from a speckle contrast image, which was calculated from 50 continuously captured frames of original laser speckle patterns. To achieve fast blood flow visualization, the graphics processing unit was used to accelerate the calculations.⁴¹ To maintain high temporal resolution, a sliding window along the time sequences of the original speckle patterns was applied to achieve the velocity images. Each original speckle pattern with index $i$ had its corresponding velocity image, which was acquired from a series of speckle patterns with the indices $i-24,\cdots ,i,\cdots ,i+25$ form $N$ speckle patterns. The velocity images were displayed in real time, and the frame rate of velocity image was 20 fps. In practical measurements, we applied a time interval of 2 s to save and display the data of the velocity image. Thus, the temporal resolution of the velocity image was 2 s in our measurements. The CBF value of each vessel was calculated by averaging the velocity in the selected area of the chosen vessel, and the CBF value of each rat was achieved by averaging the CBF value of the chosen vessels.

In functional parameter measurements, about three or four micro veins with diameters ranging from about 20 to 100 μm in the partial cranial window were selected for observation according to the anatomic structure and blood flow velocity of brain blood vessels. Before calculating the time profiles of the parameters, the MAP image from PAM has been spatially co-registered with the LSI image using control point registration method based on the global affine transformation model, and the corresponding bifurcations of the same vessel in two images are usually selected as control point pairs. All the average base value of each parameter of a rat before occlusion was set to 100%, and the values of each parameter of a rat after occlusion were normalized by its base value in percent. The final dynamic profile of each parameter was averaged from all the rats in the experiments over the time. All the changed values were presented as $\text{mean}\pm \mathrm{SEM}$. The paired $t$-test was used to determine significant differences of each parameter at a given second compared to its basal level.

3.1.

LSI and PAM of Rat Vasculature in the Cerebral Cortex

Figure 2 shows LSI image and 2D PAM image of blood vessels in the same region of a rat partial brain cortex. Although the contrast mechanism is different, LSI and PAM both can exhibit the structure of blood vessels in brain cortex explicitly with high spatial resolution. Figure 2(a) is a whitelight picture of the rat brain cortex, Fig. 2(b) shows the velocity image of the brain cortex with LSI. Because LSI is based on scattering properties of the moving scatterers in brain cortex, the structural image of the blood vessels in cortex can be obtained with good contrast by utilizing the velocity discrepancy of the scattered particles in blood vessels and surrounding brain tissue. Figure 2(c) shows the 2D PAM image of the cortex with laser wavelength of 584 nm. The structural imaging of cortex vasculature is achieved with high contrast due to the high optical absorption contrast between blood vessels and the surrounding brain tissues. As shown in Fig. 2(a)–2(c), the structural images of the brain cortex vasculature in LSI and fPAM agree well with the whitelight picture. Therefore, LSI and fPAM both have the ability to image the brain cortex vasculature with good quality, and it is the base of functional imaging by combing these two imaging techniques.

As shown in Fig. 2(b), the cortex arteries with relatively high CBF velocity were imaged with high contrast, while the veins were imaged with relatively low contrast. It demonstrates that LSI has an advantage on imaging the arteries even if their calibers are much smaller than the veins. According to the difference of blood flow velocity of the vessels and their directions and calibers, one can differentiate the arterial branches from the venous ones for further functional imaging detection. In Fig. 2(b), we can identify the vessel with the arrow label of “1” to be an artery branch and the vessel with the arrow label of “2” to be a vein branch. Compared with LSI, the artery labeled with “1” and the vein labeled with “2” both have high contrast in PAM in Fig. 2(c), and both have a high signal-to-noise ratio (SNR) of about 30 dB. The small veins in the region with the label of dotted circle in Fig. 2(b) were imaged with low contrast, the detailed branches of the vessels could be hardly identified; while the corresponding PAM image in Fig. 2(c) shows the branches clearly.

Fig. 2

LSI and PAM image of rat partial brain cortex and the registered image of the two images. (a) Whitelight picture of the cortex, (b) LSI image of the cortex, (c) PAM image of the cortex, (d) photoacoustic section image at the location indicated by solid line in (c), and (e) the superimposition of the spatial registered PAM image on the LSI image (A, anterior; M, medial; H, high; L, low).

Comparing Fig. 2(b) with Fig. 2(c), there is a lack of depth information in the LSI image, while partial depth information of the cortex vasculature can be identified from the 2D PAM image. For example, we may make a mistake of considering that the artery with the label of “1” is located above the vein with the label of “2” in LSI image, but in fact, the artery goes through beneath the vein in PAM image. The depth locations of the vessels could be well identified in photoacoustic section image in Fig. 2(d), which is a b-scan image at the location indicated by the solid line in Fig. 2(c). As shown in Fig. 2(d), the blood vessels with the label of “a,” “c,” and “e” are the branches of the vein labeled with “2,” and the vessels with the label of “b” and “d” are the branches of the artery labeled with “1,” it can be clearly seen that the vein is above the artery at the location imaged.

As shown in Fig. 2(e), the PAM image of the cortex vasculature after image registration using the LSI image as the reference is superimposed on the LSI image. From Fig. 2(e), the following are indicated: Firstly, images of the cerebral cortex vasculature using the two imaging techniques agree well. Secondly, fPAM and LSI complementarily provide complete 2D structural information about cortex vasculature, the superimposed image exhibits more structural information than the single image modality. Especially at the locations where the vessels seems to be superposed in the 2D images, such as the “superpositions” of the artery “1” and the vein “2,” the depth locations are clearly described.

3.2.

Dynamic Monitoring of CBF, ${\mathrm{SO}}_2$ and HbT in Rat Transient Cerebral Hypoperfusion

The exemplar blood vessels selected in LSI and PAM image of a rat brain cortex for monitoring of the functional parameters in transient cerebral hypoperfusion are shown in Fig. 3. The LSI image of the cortex is shown in Fig. 3(a), the 2D PAM image of the corresponding region indicated by dashed rectangle in Fig. 3(a) is shown in Fig. 3(b). The continuous photoacoustic b-scan imaging was performed at the location indicated by a dotted line in Fig. 3(b), the corresponding section image was shown in Fig. 3(c). The venous blood vessels with the label of “V1,” “V2,” and “V3” in Fig. 3(b) were selected for continuous dual wavelength fPAM according to the method mentioned above, in order to monitor the changes of ${\mathrm{SO}}_2$ and HbT in these vessels before and after UCCA ligation. The change of CBF was calculated from these blood vessels in the corresponding LSI image. It is worth noting that we conduct image registration of PAM and LSI before the calculation. The section images of the selected vessels labeled in Fig. 3(b) are shown in Fig. 3(c) with the same labels. Here, the different depth locations of the vessels are clearly described, and several deeper blood vessels labeled with dotted arrow are also shown in Fig. 3(c), while they could not be found in LSI image. Moreover, these deeper blood vessels are also almost invisible in 2D PAM, also suggesting that the section imaging in depth direction is very essential.

Fig. 3

The exemplar blood vessels selected in LSI and PAM image of a rat brain cortex for dynamic imaging monitoring. (a) LSI image of the brain cortex, (b) 2D PAM image of the corresponding region in cortex indicated by dashed rectangle in (a), and (c) b-scan PAM image at the location indicated by dotted line in (b). The blood vessels with the label of “V1,” “V2,” and “V3” are the selected vessels for functional monitoring.

Figure 4 presents the dynamic changes of CBF, HbT, and ${\mathrm{SO}}_2$ in micro veins in ipsilateral cortex after UCCA ligation from ten rats. The parameters of each rat after ligation were normalized by their baseline values. As shown in Fig. 4(a), the rapid changes of all the parameters occurred in about 30 s after UCCA ligation, which were indicated by a rectangle, and the expanded profiles were shown in Fig. 4(b). As shown in Fig. 4(b), CBF decreased abruptly to $37\pm 3\%$ ($p<0.001$) of base level in 6 s after ligation, then recovered rapidly to $71\pm 3\%$ in the next 30 s; ${\mathrm{SO}}_2$ dropped to $72\pm 7\%$ ($p<0.005$) of base level in 6 s after ligation, then fluctuated around this level in the next 30 s; HbT also presented a little decrease to $93\pm 2\%$ ($p<0.001$) of base level in 6 s after ligation, then recovered slowly. From Fig. 4(a), we can see that the parameters all came back over time after a transient decrease. CBF went up slowly after a rapid recovery to $71\pm 3\%$ of baseline; an apparent recovery tendency in ${\mathrm{SO}}_2$ was found at about 40 s after the decrease; HbT slowly recovered to $95\pm 2\%$ of baseline in about 60 s after the decline, and then fluctuated to baseline level. In short, right after UCCA ligation, CBF decreased very much, ${\mathrm{SO}}_2$ decreased less and HbT decreased the least. Furthermore, the responses to UCCA occlusion in the parameters were different: there was an immediate recovery after an abrupt drop in CBF right after the occlusion; while there was a period of fluctuation after decrease in ${\mathrm{SO}}_2$ after the occlusion; HbT presented a little decrease right after UCCA ligation and an immediate increase with small magnitude after the occlusion. In summary, the recovery of ${\mathrm{SO}}_2$ was lagged behind the recovery of CBF after decrease, and the temporal profile of HbT was similar to that of CBF, except that the magnitude of decrease of CBF was greater than that of HbT.

Fig. 4

Dynamic monitoring of CBF, ${\mathrm{SO}}_2$, and HbT in rat brain transient hypoperfusion. (a) Average time courses of the three parameters before and after UCCA occlusion, and (b) the expand profiles indicated by rectangle in (a). “0” is the ligation time point.

The time profiles in Fig. 4 represent hemodynamic responses and changes in blood oxygenation level in ipsilateral cerebral cortex induced by UCCA occlusion in rats. As seen in Fig. 4, the changes of CBF were much greater than the others. The abrupt decrease and rapid recovery made it a transient hypoperfusion status in the brain cortex. Figure 5 shows an exemplar response of CBF in a rat before and after the UCCA ligation. The time serial velocity images reflect the temporal and spatial cerebral cortex blood velocity responses to transient cerebral hypoperfusion. As shown in Fig. 5, the blood velocity in the ipsilateral partial cerebral cortex decreased to the minimum level at 6 to 8 s after ligation, then recovered to a relative high level in several seconds.

Fig. 5

Exemplar time serial 2D laser speckle velocity images in a rat before and after UCCA ligation.