Recently, optical-resolution photoacoustic microscopy (OR-PAM) has emerged as a viable tool for in vivo microvascular imaging [7]. OR-PAM provides high optical absorption contrast—from either intrinsic or exogenous absorbers—and axial ultrasonic resolution at depths up to the optical transport mean free path (~1 mm in the skin) [8]. If a single-element ultrasonic transducer is used, the imaging speed is limited by the 2-D mechanical scanning for 3-D imaging. One solution to improving the speed is to use an ultrasound array. Previously, an ultrasound array was used to accelerate the imaging rate of acoustic-resolution photoacoustic microscopy (AR-PAM) by ~100-fold from the single-element implementation—reaching a B-scan imaging rate of 166 Hz, a 3-D imaging rate of ~0.1 Hz, and a single 3-D imaging time of only 1–2 s [9, 10]. However, the poor acoustic elevational focus (>200 μm)—common to all linear ultrasound arrays—became a bottleneck for spatial resolution [11, 12], which limited its application in microcirculation studies (most microvessels have diameters less than 100 μm).

In this work, we developed section-illumination photoacoustic microscopy (SI-PAM) to improve the elevational resolution. Moreover, by optimizing data acquisition and transfer, we improved the imaging speed to 249 Hz for B-scans and 0.5 Hz for continuous 3-D scans. SI-PAM was used to image in vivo microcirculation dynamics in mouse ears noninvasively. To our knowledge, this is the first report of dynamic 3-D in vivo photoacoustic imaging with both high temporal and spatial resolutions.

The principles of SI-PAM are shown in Fig. 1. In order to achieve section illumination, the laser beam was first expanded, and then cylindrically focused into the sample. The numerical aperture of the focus was 0.015, which in theory would result in an elevational resolution of 24 μm—10-fold better than the one defined acoustically; the depth of focus in air was ~2.7 mm, greater than the targeted 1-mm imaging depth. To detect the photoacoustic waves from the sample, a custom-built ultrasound array of 30-MHz center frequency was used [9], positioned opposite the laser illumination. Photoacoustically exciting the entire B-scan imaging region with each laser pulse, the section illumination was able to take full advantage of the ultrasound array for high-speed imaging. While 2-D B-scan imaging required no mechanical scanning, 3-D imaging necessitated linearly translating the sample in the elevational (y) direction. By storing photoacoustic signals in the data acquisition (DAQ) card, we achieved 3-D image acquisition at 0.5 Hz, corresponding to a 2-D (B-scan) image acquisition rate of 249 Hz. Currently, in order to stream data from the 48-channel ultrasound array to the 8-channel DAQ card, 6 laser pulses were needed for one B-scan. Thus the laser repetition rate corresponding to the 249-Hz B-scan rate was ~1.5 kHz, approximately the highest rate at which our laser could operate. Further details about image acquisition and reconstruction were presented in our previous publications [9, 10].

Figures 2(a)–(c) show that the elevational resolution (y) was improved ~10 fold—from 200–400 μm to 28 μm—by the section illumination, while the in-plane lateral (x) resolution (~70 μm) was unaffected. Figure 2(d) shows that SI-PAM can penetrate ~1.6 mm through biological tissue. Figure 2(e) is an in vivo photoacoustic image of a mouse ear microvasculature acquired by SI-PAM at 584 nm, sensing the intrinsic absorption contrast of hemoglobin. The image is shown in the form of maximum amplitude projection (MAP)—the maximum photoacoustic amplitudes projected along a direction to its orthogonal plane—along the z axis with depth encoded by color. Unless otherwise mentioned, all MAPs are along the z axis. Microvessels in diameters down to 30 microns were clearly imaged, and a predominantly two-layered structure of blood vessels was observed, consistent with the previous results from OR-PAM [13]. Figure 2(f) is a snapshot of a 3-D animation (Video 1) showing the mouse ear microvasculature from various perspectives. All animal experiments were carried out complying with Washington University approved protocols.

The dynamic 3-D in vivo imaging capability of SI-PAM was demonstrated by real-time monitoring of the wash-in dynamics of Evans Blue (EB) dye in mouse ear microcirculation. Swiss Webster mice (Harlan, Inc., USA) weighing ~25 g were used. Upon injection of ~0.05 ml of 3% EB through the tail vein, the mouse ear was continuously imaged by SI-PAM at 600 nm for up to 2 min at 5-s intervals. At this wavelength, EB has much stronger absorption (which peaks at 620 nm) than hemoglobin, and thus its signal dominates the contrast. Video 2 shows the entire EB wash-in process recorded by SI-PAM—with representative frames shown in Fig. 3. It is clearly seen that the dye progressively reaches different levels of vessel branches—from the root to the edge of the ear—at different time points. Yet the overall wash-in process is as short as 15–20 s. After 1–2 min, the photoacoustic signal decreases (Figs. 3 (g) and (h)), indicating the beginning of the wash-out of EB. However, the entire wash-out process, which was not monitored in this study, could take up to a few days.

Although the spatial resolution of the SI-PAM was insufficient to resolve closely located arteriole-venule pairs even if the oxygen saturation of hemoglobin were measured spectrally, we found that the dynamics enabled us to distinguish arterioles from venules in the microcirculation. In fact, four distinct stages of the wash-in process can be observed in Video 2 (or Fig. 3):

In the end, the entire microcirculation of the mouse ear was perfused with EB dye. Figure 4 is a pseudo-colored composite image showing the separated arterioles (red) and venules (green). Furthermore, Video 3 shows the wash-in dynamics of EB in both grayscale and pseudo-color.

With 50 Hz B-scan imaging rate, the entire EB uptake process was quantitatively imaged by SI-PAM. An MAP image and a representative B-scan image of the mouse ear microvasculature are shown in Figs. 5(a) and (b), respectively. Video 4 is a real-time B-scan movie showing the EB uptake in a single vessel (vessel 1). The photoacoustic amplitude representing the dye concentration was quantified as a function of time (Fig. 5 (d)). The EB injection started at t = 0 s and took ~2 s to complete. At ~14 s, the photoacoustic signal stabilized, suggesting that the dye concentration had reached a steady state in the blood circulation. This stabilization time agreed well with the circulation time needed to fully mix the dye in blood, which was ~15 s as estimated by using a stroke volume of 20 μl, a heart-beat rate of 400 beats/min (mice under anesthesia), and a total blood volume of 2 ml.