2.1.

Photoacoustic Imaging System

A deep reflection mode PA imaging system was used.⁹ Figure 1 shows the schematic of the system. Upon laser excitation, the tissue generates ultrasonic waves, known as PA waves. The PA waves were first received by an ultrasonic transducer, amplified by a pulser/receiver (5072PR, Panametrics-NDT, Watham, Maryland), digitized by an oscilloscope (Tektronix TDS 5054, Natham, Massachusetts) and stored in a computer, which also controlled an XY-linear translation stage (XY-6060, Danaher Motion, Washington, DC) for raster scanning. A photodiode (SM05PD1A, Thorlabs, Newton, New Jersey) was used to compensate for the energy instability of laser pulses.

Fig. 1

Schematic of the deep-penetrating reflection-mode photoacoustic imaging system. A Cartesian coordinate is shown.

To achieve deep penetration of light, a tunable near-infrared Ti:sapphire laser (LT-2211A, LOTIS TII, Minsk, Belarus) pumped by a Q-switched Nd:YAG laser (LS-2137/2, LOTIS TII, Minsk, Belarus) was used for PA excitation. The laser pulse duration was <15 ns, and the repetition rate was 10 Hz. The incident laser beam on the tissue surface was controlled to be less than the ANSI standard for maximum permissible exposure (31 mJ/cm²).¹⁰

To receive deep PA signals with minimal ultrasonic attenuation, a 5-MHz central frequency ultrasonic transducer (V308, Panametrics-NDT, Watham, Maryland) was used. This transducer had a spherical focus with a 2.54-cm focal length, a 1.91-cm active diameter, and a 72% nominal bandwidth based on the full width at half maximum amplitudes. The spatial resolutions were 144 μm in the axial direction and 560 μm in the transverse direction. The scanning time depends on the laser pulse repetition rate, the scanning step size, and the field of view. Typical values are a 0.1-mm scanning step size for a 1-D scan and a 10-Hz laser pulse repetition rate. The acquisition time is ∼6 s for a B-scan. Note that the signal was not averaged for any of the images. The transducer was located inside a water container with an opening of 5×5 cm at the bottom, sealed with a thin, clear polyethylene membrane. The object was placed under the membrane, and ultrasonic gel was used for US coupling. To reduce the surface PA wave generation, a concave lens, a spherical conical lens, and an optical condenser were used to form the dark-field illumination.¹¹

2.2.

Ultrasound Imaging System

With some slight modifications to the PA imaging system, we can get coregistered pulse-echo US images with the PA images. A function generator (33250A, Agilent, Santa Clara, California) instead of the laser was used as the trigger source. The pulser/receiver (5072PR, Panametrics-NDT, Watham, Maryland) was set to transmit/receive mode for pulse echo transmission and detection. For this study, another clinical US imaging system (iU22; Philips Healthcare, Andover, Massachusetts) with a linear-array (L8-4, 128 elements, 4–8 MHz) was used for comparison to PA images as well.

2.3.

Sample Preparation

Chicken breast tissue was used as the background medium mimicking the human soft tissue.¹² Pieces of foreign bodies with arbitrary shapes were embedded in the chicken breast tissue at depths ranging from 3 to 10 mm. A metal blade, a piece of plastic, a piece of wood, a piece of cloth, and a piece of glass were used to test the ability of PA imaging to detect foreign bodies. Also, two pieces of wood, two pieces of cloth, and a piece of plastic were used for a comparison between PA imaging and US imaging. Then, a piece of wood and a piece of cloth were used for spectroscopic PA imaging.

3.1.

Photoacoustic Ability to Detect Foreign Bodies

To show the feasibility of PA imaging of foreign bodies, we imaged various foreign bodies embedded in chicken tissue using an optical wavelength of 766 nm. We could clearly see the objects in the PA images. Figures 2a, 2g show the PA maximum amplitude projection (MAP) and B-scan without any objects embedded in the tissue, respectively. We do not see anything. Figures 2b, 2c, 2d show the PA MAP images of a piece of wood, a piece of glass, and a piece of cloth. Figure 2f shows the PA MAP images of a piece of metal and a piece of plastic. MAP was performed by projecting the maximum signal amplitude from each A-line onto the XY plane. The shapes of the objects agree well with those of the actual samples shown in Figs. 2e, 2i. Figure 2h shows the B-scan image of the piece of metal and piece of plastic. Imaging depth can reach >1 cm. The results indicate that PA is excellent for exactly locating the foreign bodies of various debris materials.

Fig. 2

PA images and digital photographs for various foreign bodies. Objects were embedded in chicken breast tissue: (a) PA MAP image without objects embedded in the tissue, (b–d,f) PA MAP images, (e,i) digital photographs, (g) PA B-scan image without objects embedded in the tissue, and (h) PA B-scan images. TS: tissue surface.

3.2.

Comparison to Ultrasound Imaging

US is not sensitive to tiny objects (a few millimeters) because of their low acoustic contrasts or because of operator variability, especially in the presence of ubiquitous US speckles.¹³ By contrast, PA imaging provides excellent optical absorption contrast and US spatial resolution, enabling detection of tiny objects. To show this advantage of PA imaging, PA imaging and US imaging were compared. A piece of wood and a piece of cloth were embedded in chicken breast tissue and were imaged by both the PA system and the clinical US system. Then, they were cut to smaller sizes and were imaged again by both imaging modalities. Figures 3a, 3f show digital photographs for the samples. Figures 3b, 3c show the PA B-scan images of the larger wood and larger cloth, respectively. Figures 3d, 3e show the US B-scan images of the larger wood and larger cloth, respectively. Figures 3g, 3h show the PA B-scan images of the smaller wood and smaller cloth, respectively. Figures 3i, 3j show the US B-scan images of the smaller wood and the smaller cloth, respectively. Although both modalities successfully imaged the objects, PA images had much higher contrasts than US images. The contrast is defined as (I – I _b)/I _b with I and I _b representing the average signal intensity of the foreign body and that of the background (chicken breast tissue), respectively. The PA contrasts of the larger wood and larger cloth are ∼11 and ∼13, respectively, whereas the corresponding US contrasts are only ∼1.2 and ∼1.5, respectively. Moreover, after the targets were cut, the PA contrasts of the wood and cloth remained almost the same, but the corresponding US contrasts were reduced to only ∼0.6 and ∼0.7, respectively. The poor US contrasts made it difficult to identify the smaller foreign bodies in US images.

Fig. 3

PA and US B-scan images of foreign bodies of two sizes. Objects were embedded in chicken breast tissue: (a,f) Digital photographs, (b,c) PA B-scan images of the larger foreign bodies, (d,e) US B-scan images of the larger foreign bodies, (g,h) PA B-scan images of the smaller foreign bodies and (i,j) US B-scan images of the smaller foreign bodies. TS: tissue surface; ST: surrounding tissue.

US specificity is limited by the overlapping acoustic characteristics of foreign bodies and the surrounding tissue.¹⁴ If the foreign bodies have low optical absorption contrast but sufficient acoustic contrast, they may show up in US images but not in PA images. Therefore, PA imaging and US imaging are complementary and will not supplant each other. However, PA images and US images can be acquired from the same cross sections of the sample.^{15, 16, 17, 18} The two types of images can be coregistered naturally to combine the advantages of PA and US imaging. As shown in Fig. 4, a piece of dark soft polyethylene plastic (thickness, ∼150 μm) buried in chicken breast tissue was detected by coregistered images. In the PA images shown in Figs. 4a, 4d, the location and size of the target were accurately specified. It is not showing up in the coregistered US MAP and B-scan images [Figs. 4b, 4e] because the piece of plastic has a similar acoustic property with the tissue. As shown in Fig. 4f, the clinical US system cannot detect the piece of plastic either.

Fig. 4

Coregistered PA and US images for a piece of plastic. The object was embedded in chicken breast tissue: (a) Photoacoustic MAP image (signals from the tissue surface were removed), (b) coregistered US MAP image (signals from the tissue surface and from the polyethylene membrane were removed), (c) digital photograph, (d) B-scan image of (a) at the dashed line, (e) B-scan image of (b) at the dash line, and (f) B-scan image acquired by the clinical US system. TS: tissue surface; PM: polyethylene membrane.

3.3.

Spectroscopic PA Imaging

For spectroscopic PA study, we used the same Ti:sapphire laser with the wavelength ranging from 750 to 800 nm, and the other parameters are the same as those used in the previous PA experiments. Figure 5 shows the PA spectra of a piece of green cloth and a piece of brown wood. To validate the results, the optical absorbance of the corresponding color ink was acquired by a spectrophotometer (Cary 50 Bio, Varian, Walnut Creek, California). Each curve was normalized by its own minimum value. The correlation coefficients between the PA curve and the absorbance curve in Figs. 5a, 5b are 0.82 and 0.92, respectively. Spectral analysis would be helpful in choosing the optimal light wavelengths for imaging specific objects to produce high-contrast images. In addition, viable and nonviable tissue can potentially be distinguished spectrally.¹⁹

Fig. 5

Optical absorption spectra of foreign bodies measured by PA imaging. (a) PA amplitude of the brown wood and optical absorbance of brown ink versus the optical wavelength. (b) PA amplitude of the green cloth and optical absorbance of green ink versus the optical wavelength.