2.1.

Imaging System Description

The PA imaging system adapted from a clinical US scanner (iU22, Philips Healthcare, Andover, MA) is shown in Fig. 1. The channel board architecture was modified to allow acquisition of the raw per-channel PA and US data. The dual-modality image data were transferred to a custom-built data acquisition (DAQ) system for image reconstruction and display. For 3-D PA and US imaging, we used the 2-D matrix array US probe (X7-2, Philips Healthcare) which has 2500 elements and a nominal bandwidth of 2 to 7 MHz. Each frame of the PA volumetric image requires 36 laser shots and takes 20 sec, limited by the current DAQ system performance. The DAQ system was synchronized with laser firings by an FPGA-based electronic board and digital delay generator (Stanford Research Systems, Sunnyvale, CA). A 3-D back-projection algorithm was implemented using Matlab (MathWorks Inc., Natick, MA) to reconstruct PA images.¹⁹ The image reconstruction for one $2\times 2\times 2{\mathrm{cm}}^3$ volumetric image consisting of $100\times 100\times 400\text{voxels}$ took approximately 3 h on a 4-core central processing unit (CPU). Because the X7-2 array probe performs partial beamforming within the transducer to reduce the number of DAQ system channels needed for the large number of transducer elements, a novel proprietary algorithm reverses the partial beamforming process to recover raw PA channel data.

Fig. 1

Schematic of the integrated 3-D PA and US imaging system using a 2-D matrix array ultrasound probe.

A wavelength tunable dye laser (PrecisionScan-P, Sirah, Kaarst, Germany), pumped by a Q-switched Nd:YAG laser (PRO-350-10, Spectra-Physics, Santa Clara, CA) serves as the light source for the PA imaging. The laser pulse is coupled to a fused end, bifurcated fiber bundle (LightGuideOptics, Rheinbach, Germany) integrated with the matrix array US probe. The laser emits pulses at a repetition rate of 10 Hz and pulse duration of 6.5 ns. An optical wavelength of 650 nm, which is close to the peak absorption wavelength of methylene blue dye, was utilized for all the experiments. The light fluence on the skin was less than $10\mathrm{mJ}/{\mathrm{cm}}^2$, which is within the American National Standards Institute (ANSI) safety limits.²⁰

2.2.

Gelatin Phantom Preparation

To study the spatial resolution of the 3-D PA imaging system, an optically and acoustically transparent gelatin phantom (10% by weight) was prepared with a human hair embedded, which was placed parallel to the probe surface at a depth of $\sim 2\mathrm{cm}$. A second gelatin phantom consisting of one white hair, one black hair, and a gelatin slab dyed with black ink was imaged to illustrate the complementary optical absorption and US scattering contrasts of PA and US imaging. A third phantom consisted of three hairs placed 10, 15, and 20 mm deep in a clear gelatin phantom. This phantom was designed to demonstrate 3-D volumetric PA imaging.

2.3.

In Vivo 3-D PA Imaging of Rat SLNs

Sprague Dawley rats ($\sim 200\mathrm{g}$) were anesthetized using a mixture of ketamine ($85\mathrm{mg}/\mathrm{kg}$) and xylazine ($15\mathrm{mg}/\mathrm{kg}$). After hair depilation in the left axillary region, a series of PA and US images were collected before and after a 100 μL intradermal injection of 1% methylene blue dye ($10\mathrm{mg}/\mathrm{mL}$, American Reget, Inc., Shirley, NY) into the left forepaw pad. The axillary region of the rat was imaged for 30 min following methylene blue injection. For US coupling, a water tank was applied between the US probe and the tissue surface during in vivo experiments. Axillary lymph nodes in the rat are approximately 2 mm below the skin surface. After imaging, animals were euthanized with an overdose of pentobarbital, and SLNs were dissected to confirm methylene blue dye uptake. Animal handling was performed according to the guidelines on the care and use of laboratory animals at Washington University in St. Louis.

3.1.

Gelatin Phantom Imaging

The resolution of the matrix array US probe was estimated using PA images of a human hair embedded within the gelatin. The hair is first imaged when placed along the azimuth direction of the probe and then imaged with the hair placed along the elevation direction. The PA maximum amplitude projection (MAP) images are formed by projecting the 3-D PA image onto the surface plane of the probe. Figure 2(a) and 2(b) shows the PA MAP images of the hair positioned along the azimuth and elevation directions, respectively. Figure 1(c) displays the PA B-mode image of the hair positioned along the elevation direction. Figure 2(d)–2(f) displays the 1-D PA amplitude profiles crossing the hair along the elevation, azimuth, and axial directions as indicated by the dashed lines in Fig. 2(a)–2(c). We used the full width at half maximum of each 1-D PA amplitude profile to quantify the corresponding spatial resolution. At a depth of $\sim 2\mathrm{cm}$, the spatial resolution was found to be $0.69\pm 0.05$ ($\text{mean}\pm \text{standard deviation}$) mm in the azimuth direction, $0.90\pm 0.03\mathrm{mm}$ in the elevation direction, and $0.84\pm 0.04\mathrm{mm}$ in the axial direction.

Fig. 2

PA MAP images of the hair positioned along the (a) azimuth and (b) elevation direction. (c) PA B-mode image of the hair positioned along the elevation direction. PA amplitude profiles crossing the human hair along the (d) elevation, (e) azimuth, and (f) axial directions, respectively. PA, photoacoustic; US, ultrasound; and DAQ, data acquisition.

Figure 3(a) shows a photograph of the second gelatin phantom. The PA MAP image [Fig. 3(b)] only shows the black hair and the ink-dyed gelatin slab because of the PA wave generation by the light absorbing chromophores. In the pulse-echo US MAP image [Fig. 3(c)], two hairs are seen, but not the ink-dyed gelatin, because the acoustic impedance mismatch between the hair and the gelatin produces a strong US reflection, while the acoustic impedance mismatch between the ink-dyed gelatin and surrounding gelatin is too small. The $1\times 1\times 1{\mathrm{cm}}^3$ volumetric PA and US images of the phantom are shown in Fig. 3(d) and 3(e). Figure 4(a) shows a photograph of the third gelatin phantom from top view. In the PA 3-D volumetric images [Fig. 4(b) and 4(c)], all three hairs at different depths were seen.

Fig. 3

(a) Photograph of the gelatin phantom consisting of one white hair, one black hair, and a gelatin slab dyed with black ink. (b) PA MAP image showing the black hair and the ink dyed gelatin slab. (c) US MAP image showing the white hair and the black hair. (d) PA 3-D image. (e) US 3-D image. PA, photoacoustic; and US, ultrasound.

Fig. 4

(a) Photograph from top view of the gelatin phantom consisting of three black hairs placed 10, 15, and 20 mm deep. (b) 3-D PA image showing the black hairs from top view. (c) 3-D PA image showing the black hairs from side view. PA, photoacoustic.

Fig. 5

(a) Control PA MAP image of rat axillary region acquired before the methylene blue dye injection. (b) PA MAP image taken 5 min post-injection showing the uptake of the blue dye in the SLN. (c) PA MAP image taken 30 min post-injection showing the PA signal enhancement in the SLN. (d) Post-mortem photograph of the rat taken after PA imaging and skin removal. (e) PA 3-D image taken 30 min after injection. (f) Plot of PA signal amplitudes from the SLNs (immediately after injection, at 5 min, 10 min, and 30 min post injection) fitted to a single compartment model $C=16.1(1-e^{-0.142t})+4.03$ with an $R^2$ value of 0.785 ($C$ is the methylene blue concentration in the SLNs, and $t$ is the post-injection time). PA, photoacoustic; and SLN, sentinel lymph node.

3.2.

Rat SLN Mapping

To explore 3-D PA mapping of SLNs with methylene blue dye, we used the matrix array US probe to image the axillary region in a rat. A control PA MAP image was acquired before injection of methylene blue dye [Fig. 5(a)]. After injection, the blue dye accumulated in the SLNs by lymphatic drainage. Figure 5(b) and 5(c) shows the PA MAP images acquired at 5 and 30 min post-injection, respectively. The uptake of methylene blue in the rat SLN was clearly observed in the PA MAP images, which are all displayed with the same dynamic range. After dissection, the methylene blue-dyed SLN was visually identified as shown in the post mortem photograph [Fig. 5(d)].

The 3-D photoacoustic image from the rat SLN acquired at 30 min post-injection is rendered volumetrically in Fig. 5(e). The dynamics of the methylene blue accumulation in the SLN following injection for three rats is plotted in Fig. 5(f). A single-compartment model^21–24 was used to fit the methylene blue dye accumulation in the SLN.