2.1.

RSTM Surface Modification

Two types of RSTM with different color-shifting modes and transition temperatures ($T_t$) were used an RSTM with black-to-white shifting (Atlanta Chemical Engineering) and an RSTM with red-to-pink shifting (LCR Hallcrest: TP31/031CR). RSTM generally contain a thermosetting polymer shell, such as poly(urea-formaldehyde) (PUF)²⁸ and chromic material such as a mixture of pigment powder, acid, and solvent. The color change of this system relies on the melting/crystallization of the solvent in the dye mixture without the need for an extra color developer. The solvent is in a solid-state below the crystallization temperature, and the acid binds the dye to give the pigment a red or black color.⁴³ Upon heating above the melting temperature, the solvent transforms into a liquid state, and the dye and acid dissociate, resulting in a loss or decrease of color. Thus the RSTM allow safe and reversible color changes [Fig. 1(a)]. However, due to their hydrophobic polymer shell, native RSTM cannot form a homogeneous and stable aqueous suspension and thus are not suitable as contrast agents for PAT.

Fig. 1

RSTM surface modification: (a) RSTM consist of a thermochromic material surrounded by a transparent hydrophobic polymer shell, and undergo reversible thermal-responsive color changes. (b) Surface modification of an RSTM by alginate (Alg) via EDC/NHS coupling to form a water-dispersible ReST. (c) FTIR spectra of Alg, RSTM and ReST with characteristic bands marked. (d) Reversible color change and stability of aqueous suspensions of red-to-pink RSTM and ReST ($2\mathrm{mg}/\mathrm{mL}$) during a heating-cooling cycle. (e) Reversible color change for aqueous suspension of black-to-colorless ReST ($2\mathrm{mg}/\mathrm{mL}$) during a heating-cooling cycle. (f) Bright-field microscopy images of aqueous dispersions of RSTM and ReST. Scale bars: $50\mu \mathrm{m}$.

To address their aqueous insolubility, we modified the surface of RSTM with the biocompatible hydrophilic polymer alginate. The commercially available RSTM consisted of a PUF shell,²⁸ and therefore, the polysaccharide alginate could be used to modify RSTM by a reaction between the carboxylic acid groups on alginate chains and secondary amine groups on the RSTM surface [Fig. 1(b)]. Typically, 100 g of 1 wt. % sodium alginate in 2-($N$-morpholino) ethanesulfonic acid (MES) buffer (pH 5.5) was activated by 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC, $8\times {10}^{-4}\mathrm{mol}{\mathrm{L}}^{-1}$) and $N$-hydroxysuccinimide (NHS, $8\times {10}^{-4}\mathrm{mol}{\mathrm{L}}^{-1}$) for 30 min at room temperature. Then 1.0 g RSTM dispersed in 40 mL dimethylformamide was added slowly into the solution and the reaction was kept at room temperature for 24 h with gentle stirring. The resulting solution was centrifuged at 5000 rpm and the pellet was washed twice with deionized water. The final solution was dialyzed against deionized water for two days, followed by freeze-drying to obtain a dry powder. All other chemicals were purchased from Sigma-Aldrich and were used as received.

2.2.

Fourier Transform Infrared Spectroscopy Characterization of ReST

FTIR spectra were recorded on a Nicolet iS50 FTIR spectrometer (Thermo Scientific) from 550 to $4000{\mathrm{cm}}^{-1}$ by averaging 32 scans of the signal at a resolution of $2{\mathrm{cm}}^{-1}$ in attenuated total reflectance mode. The shell composition and successful surface modification of RSTM were confirmed by FTIR [Fig. 1(c)]. The N─C stretching and N─H bending of secondary amines at $1550{\mathrm{cm}}^{-1}$, O─H stretching vibration at $3460{\mathrm{cm}}^{-1}$, N─H stretching at 3500 to $3100{\mathrm{cm}}^{-1}$, and C═O stretching at $1743{\mathrm{cm}}^{-1}$ are assigned to the PUF shell in RSTM. The C═O stretch at 1601 and $1417{\mathrm{cm}}^{-1}$ are characteristic bands of carboxylic acid in alginate. After surface modification, the emergence of C═O stretching at $1664{\mathrm{cm}}^{-1}$ suggested the formation of a tertiary amide bond, acting as the chemical linkage between RSTM and alginate.

2.3.

Characterization of ReST Color-Changing Behavior and Stability

Macroscopic color changing and ReST morphology were recorded using a digital camera. The unmodified red-to-pink RSTM in suspension rapidly aggregated after the first heating step [Fig. 1(d)]. In contrast, the modified red-to-pink ReST in suspension remained homogeneous during multiple heating-cooling cycles between 25°C and 37°C and exhibited reversible color changes. Similar behavior was also observed for the modified black-to-colorless ReST [Fig. 1(e)]. Thus the chemically anchored alginate chains on the ReST stabilized the microcapsules without interfering with the color transition of the thermochromic materials inside. Fluorescence microscopy (Eclipse, Nikon) under a bright field validated the excellent dispersion and stability of the ReST suspensions. In contrast to prominent particle aggregation in RSTM suspensions, the microcapsules (3 to $6\mu \mathrm{m}$ diameter) in ReST suspensions were evenly distributed in water, as shown by the bright-field microscope and fluorescence microscope images at different magnifications [Fig. 1(f) and Fig. S1 in the Supplementary Material]. Thus ReST exhibited good dispersion stability and reversible color-shifting, making them potentially useful as a chromic contrast agent in PAT.

2.4.

Optical Absorption Measurements

Absorption spectra were measured using a Shimadzu UV-3600 spectrophotometer with a quartz cuvette. ReST were dispersed in DI water at $2.7\mathrm{mg}/\mathrm{mL}$. The absorption spectra were measured at 10°C, 20°C, 30°C, 40°C, and 50°C. The excitation wavelength was changed from 350 to 1300 nm with a step size of 1 nm. The temperature-dependent absorption spectra were further analyzed at 532 and 1064 nm. At each wavelength, the solution was heated from 5°C to 50°C with a temperature increase rate of $2°\mathrm{C}/\min$. The solvent (DI water) background was subtracted.

2.5.

Multiscale PAT

For SG-OR-PAM [Fig. 2(a)],^37,38 excitation was provided by an Nd:YAG fiber laser (IS8II-E, EdgeWave, Würselen, Germany) at 532 nm with a pulse repletion rate (PRR) of 1 kHz. The laser beam was expanded by a beam expander, focused by an objective lens, then delivered to the imaging object. An ultrasound transducer (V214-BB-RM, Olympus-NDT, Massachusetts, USA) detected the resultant photoacoustic signal with a central frequency of 50 MHz. The detected signals were amplified and digitized by the data acquisition card (ATS9350, Alazar Tech, Quebec, Canada) with a sampling frequency of 250 MHz. The optical beam and acoustic wave were confocally aligned by an optical-acoustic combiner, optimizing the imaging sensitivity. A two-dimensional mechanical scanning stage (L-509, PI, Aubum, Massachusetts, USA) provided raster scanning of the sample. The lateral resolution of SG-OR-PAM was $3.7\mu \mathrm{m}$.

Fig. 2

Schematics of the three PAT implementations and the imaging sequence of ReST: (a) SG-OR-PAM, (b) CFT-UT-PAM, and (c) PACT. AMP, amplifier; BE, beam expander; DAQ, data acquisition card; OL, objective lens; UT, ultrasound transducer; and WT, water tank. (d) Time sequences of the thermal switching cycle of ReST.

For CFT-UT-PAM [Fig. 2(b)],³⁹ 532 nm excitation light was provided by an Nd:YAG fiber laser (VPFL-G-20, V-Gen, Tel Aviv, Israel) at a PRR of 250 kHz. The beam expander shaped and expanded the light beam and scanned along the fast-axis direction via a galvanometer scanner (GVS 002, Thorlabs, Newton, New Jersey, USA). The scanning beam was focused by an objective lens with a focal length of 50 mm, transmitted through the CFT-UT, and delivered to the imaging target. The acoustic focal line of the CFT-UT was confocally aligned with the optical focus along the scanning direction (fast axis) of the galvanometer scanner. At the same time, the slow axis was mechanically scanned by a motorized stage (L-509, PI, Aubum, Massachusetts, USA). The resultant photoacoustic signal was detected by the CFT, amplified, and digitized by the data acquisition card (ATS9350, Alazar Tech, Quebec, Canada) with a sampling frequency of 250 MHz.

For PACT [Fig. 2(c)], experiments were performed using a hybrid imaging system for PACT and high-frequency ultrasound imaging.^40–42 The excitation source was a 532-nm pulse laser (Q-smart 850, Quantel Laser, MT, USA), and the PRR was 10 Hz. The excitation light was coupled into a two-branch linear fiber bundle that flanked the linear transducer array and guided to the target surface. The generated photoacoustic signals were received by a 128-element linear array ultrasound transducer (Philips L7-4, Massachusetts, USA; central frequency, 5 MHz,). The system’s spatial resolution is $\sim 300\mu \mathrm{m}$ along the axial and lateral directions and $\sim 1\mathrm{mm}$ along the elevational direction. A high-frequency wobbler ultrasound transducer was used to acquire the B-mode ultrasound images with a central frequency of 35 MHz.

Figure 2(d) shows sequences of ReST thermal switching cycles. In each cycle, the temperature was changed between 20°C and 40°C. For all PAT experiment, 532-nm light illumination was used. PA images were acquired at different temperature points using SG-OR-PAM and PACT imaging, and continuous real-time imaging was performed using CFT-UT-PAM.³⁹

2.6.

In Vitro Multiscale PAT of ReST

For in vitro studies using SG-OR-PAM and CFT-UT-PAM, ReST were dispersed in DI water, and a “D” shape pattern was deposited using the red ReST suspension on a glass slide. After the water evaporated, a thin layer of ReST remained on the slide for imaging. For in vitro studies using PACT, the red or black ReST were mixed with Matrigel and implanted in an agar phantom. One agar phantom with the red ReST contained 7 wt. % agar and 5 vol. % intralipid. Two additional agar phantoms with black and red ReST contained 7 wt. % agar, 5 vol. % intralipid, and 10 vol. % bovine blood to provide the background signals.

2.7.

In Vivo PACT of ReST

Healthy Swiss Webster mice (12 weeks, Hsd: ND4, Envigo.) were used for in vivo studies. The protocol for animal experiments was approved by the Institutional Animal Care and Use Committee of Duke University. Mice were fully anesthetized throughout the imaging experiments using 1.5% isoflurane gas mixed with air. The hair on the left leg or lower flank of the mouse was removed, and $100\mu \mathrm{L}$ of Matrigel-mixed red or black ReST were injected subcutaneously in the mouse leg or flank, mimicking a solid tumor targeted by ReST. The ReST-injected region was subjected to three heating-cooling cycles from 20°C to 40°C performed using cold and warm water and imaged by the hybrid PACT and ultrasound imaging system. Temperature was monitored with a needle thermometer (Fisherbrand, Traceable, 1500714).

3.1.

Optical Absorption Properties of ReST

Figure 3(a) shows absorption spectra of the red-to-pink ReST (“red ReST”) at different temperatures from 15°C to 40°C. The absorption peak at 525 nm is unchanged between 15°C and 20°C and is reduced by 57% upon heating to 25°C. No prominent absorbing peak was observed at 30°C or higher. Figure 3(b) shows absorption spectra of the black-to-colorless ReST (“black ReST”). The absorption peaks at 460 and 590 nm were unchanged between 15°C and 20°C and were reduced by 53% upon heating to 25°C. No prominent absorbing peak was observed at 30°C or higher. Figure 3(c) shows the red and black ReST absorption coefficient versus temperature. For the black ReST, the absorption coefficient was high below 26°C and decreased rapidly (by 85%) between 26°C and 28°C. The red ReST have a 70% lower absorption coefficient than the black ReST below 26°C, and a slower absorption decreases with increasing temperature. Based on the absorption characterization results, both red and black ReST are suitable for thermal reversible photoacoustic imaging with 532 nm excitation. The black ReST are better for deep-penetration PACT due to their higher absorption coefficient.

Fig. 3

Absorption characterization of ReST. Absorption spectra of (a) red-to-pink ReST (“red ReST”) and (b) black-to-colorless ReST (“black ReST”) at 15°C, 20°C, 25°C, 30°C, 35°C, and 40°C. (c) Temperature-dependent absorption of red (right $y$ axis) and black (left $y$ axis) ReST at 532 nm from 20°C to 40°C.

3.2.

High-Resolution PAM Imaging of ReST

We report two different PAM systems to demonstrate the feasibility of high-resolution photoacoustic imaging of ReST. SG-OR-PAM was used to image the red ReST at micrometer level spatial resolution.³⁶ The lateral resolution of SG-OR-PAM was $3.7\mu \mathrm{m}$, which was smaller than the red ReST particle size ($\sim 5\mu \mathrm{m}$). Figure 4(a) shows an image of dispersed red ReST by bright-field microscopy; Fig. 4(b) shows the corresponding maximum amplitude projection (MAP) image by SG-OR-PAM at the single-particle level.

Fig. 4

High-resolution PAM of ReST: (a) bright-field microscopy image of dispersed red ReST at $2\mathrm{mg}/\mathrm{mL}$. Scale bar: $200\mu \mathrm{m}$. Inset: close-up bright-field microscopy image of single red ReST capsules. Scale bar: $5\mu \mathrm{m}$. (b) SG-OR-PAM MAP image of dispersed red ReST. Scale bar: $200\mu \mathrm{m}$. Inset: close-up SG-OR-PAM MAP image of single red ReST capsules. Scale bar: $5\mu \mathrm{m}$. (c) SG-OR-PAM MAP images of the “D” shaped red ReST pattern, imaged at 20°C, 40°C, and 20°C. Inset: photographs of ReST pattern at the corresponding temperature. (d) Differential photoacoustic image between images at 20°C and 40°C. (e) Normalized photoacoustic amplitude in (c). (f) CFT-UT-PAM MAP imaging of red ReST at 20°C, 30°C, 40°C, 30°C, and 20°C. (g) Differential photoacoustic image between images at 20°C and 40°C. (h) Averaged photoacoustic signal amplitude change during temperature changes for the red ReST (Video 1, MOV, 6.477 MB [URL: https://doi.org/10.1117/1.JBO.28.8.082804.s1]).

We first imaged a ‘D’-shaped red ReST phantom. Figure 4(c) shows the photoacoustic images at different temperatures. The baseline image was acquired at 20°C with strong photoacoustic signal detected. When the sample was heated to 40°C, its color changed to light pink, and the photoacoustic signal decreased dramatically. After reducing the phantom temperature back to 20°C, the sample color changed back to red, and the photoacoustic signal amplitude recovered to over 90% of the baseline value. These results demonstrated the excellent reversibility of the ReST for photoacoustic molecular imaging. In this study, the speed of sound in the tissues increased $\sim 2\%$ from 20°C to 40°C since the speed of sound depends on the temperature. The imaging resolution may be degraded when a constant speed of sound was applied in the image formation but should be a small affect compared with the change in the optical absorption of the ReST ($\sim 1000\%$). Further, we generated a differential photoacoustic image between 20°C and 40°C [Fig. 4(d)], which can remove any nonswitching signals and improve the image contrast.

In photoacoustic molecular imaging, it might be necessary to detect the molecular probes in real time. To this end, the thermal switching of red ReST was monitored using a high-speed CFT-UT-PAM at a frame rate of 0.2 Hz (5 s per image) to demonstrate the possibility of real-time photoacoustic molecular imaging as well as the repeatability of the red ReST color changes (shows in Video 1). The D-shaped sample was continuously imaged for 32.5 min, with baseline images acquired at 20°C for the initial 2.5 min, followed by three heating-cooling cycles. In each cycle, the sample was heated from 20°C to 40°C within 2.5 min, kept at 40°C for 2.5 min, and cooled from 40°C to 20°C within 2.5 min. Five representative images at different temperatures within one heating-cooling cycle were shown to demonstrate the photoacoustic dynamics of the red ReST [Fig. 4(f)]. The quantitative results are shown in Fig. 4(h). Strong photoacoustic signals were detected [marked by the black triangle in Fig. 4(h)] when the sample was at 20°C due to the strong absorption of the ReST. Then the photoacoustic signal amplitude dropped significantly at 13.5 min when the sample was heated to $\sim 30°\mathrm{C}$. At 16 min, when the temperature reached 40°C, the ‘D’ shaped phantom was barely visible [Fig. 4(f)]. The photoacoustic signal amplitude at 40°C was $\sim 30\%$ of that at 20°C. Figure 4(g) shows a differential image between 20°C and 40°C. At 21 min, the photoacoustic amplitude recovered to the baseline level following cooling back to 20°C. The photoacoustic dynamics illustrate that the temperature-induced color changes of the ReST could be detected in real time.

3.3.

ReST Reversibility in PACT

We selected the red ReST to demonstrate its color change reversibility and utility in PACT in vitro and in vivo. A phantom was subjected to three heating-cooling cycles from 20°C to 40°C and then back to 20°C. PACT images are shown in Fig. 5(a). Strong photoacoustic signals were detected due to the high absorption at 20°C. After heating the sample to 40°C, the photoacoustic signal amplitude decreased dramatically. When the sample temperature was cooled back to 20°C, the photoacoustic signals recovered. The temperature-dependent photoacoustic signal amplitude of the ReST was quantitatively analyzed [Fig. 5(b)]. The photoacoustic amplitude at 20°C was about four times higher than that at 40°C. After three heating–cooling cycles, the photoacoustic signal amplitude returned to a similar level as the baseline, demonstrating the reversibility and photostability of the ReST.

Fig. 5

PACT of red ReST in vitro and in vivo: (a) in vitro PACT images of red ReST mixed with Matrigel phantom cycling between 20°C and 40°C. (b) Normalized photoacoustic signal amplitude at different temperatures, corresponding to the photoacoustic images in (a). (c) In vivo PACT images of the red ReST-Matrigel implant in the mouse leg at 20°C and 40°C for three cycles, overlayed with ultrasound image of mouse anatomy. (d) Differential photoacoustic image between images at 20°C and 40°C. (e) Normalized photoacoustic amplitude versus temperature. Each data point corresponds to the average photoacoustic amplitude in (c).

Figure 5(c) shows in vivo experimental result of the overlayed ultrasound and photoacoustic images at different temperatures. The ultrasound images provide mouse anatomy, whereas the photoacoustic images monitor the ReST color changes. When the mouse leg was immersed in 20°C water, the ReST (highlighted by the dashed blue box) show a strong photoacoustic signal. The ReST signal was much stronger than the photoacoustic signal of the surrounding blood vessels. After increasing the temperature of the water to 40°C, the photoacoustic signal of ReST injection area decreased to $\sim 25\%$ of the baseline. When the water temperature was reduced back to 20°C, the ReST photoacoustic signal returned to baseline. We also generated a differential PA image between 20°C and 40°C [Fig. 5(d)], which removed any nonswitching signals and improved the image contrast. Photoacoustic images were acquired for another three heating–cooling cycles, and the temperature-dependent photoacoustic signal amplitude was quantified [Fig. 5(e)]. The results demonstrate three cycles of reversible switching between 20°C and 40°C, illustrating the reversibility of ReST in PACT.

3.4.

Deep Tissue PACT of ReST with Improved Image Contrast

We further tested the application of ReST in deep tissue PACT to suppress the background signals and improve the image contrast by taking the differential image at high and low temperatures. First, phantoms with both red and black ReST mixed Matrigel at 5-mm depth [Fig. 6(a)] and 10-mm depth [Fig. 6(b)] were imaged at 20°C and 40°C. Strong photoacoustic amplitude was detected at 20°C for both black and red ReST with 5-mm depth phantom. After heating the phantom to 40°C, the photoacoustic amplitude of both the black and red ReST decreased, more so for the black ReST. In the differential image, the black ReST signal was stronger than the red ReST, and the nonswitching background signal was weaker than that in the original photoacoustic images acquired at 20°C and 40°C. The 10-mm-deep phantom demonstrated a similar temperature-related trend as the 5-mm-deep phantom, but the photoacoustic signal amplitudes were lower due to the light attenuation [Fig. 6(b)].

Fig. 6

Deep-tissue PACT of the red and black ReST. (a), (b) In vitro PACT projection images of red- or black-ReST-Matrigel phantom at 5 and 10 mm depth at 20°C and 40°C, as well as the differential photoacoustic image between 20°C and 40°C. All top-view photoacoustic projection images are shown in log scale. (c), (d) Comparison of CNR quantified from images at different temperatures as well as the differential image at 5 and 10 mm depth. (e)–(g) In vivo PACT projection images of the ReST-Matrigel implant in mouse flank at 5, 8, and 13 mm depth, as well as the corresponding differential photoacoustic images.

The differential images had higher CNR than the original images, and the black ReST had higher CNR than the red ReST, shown in Figs. 6(c) and 6(d). For red ReST in the 5-mm-deep phantom, the improvement in CNR was 16.2% versus the image at 20°C, and 51.6% versus the image at 40°C. For the black ReST, the improvement in CNR was 33.2% versus the image at 20°C and 119.1% versus the image at 40°C. Similar improvements in CNR by the differential images were obtained on the 10-mm-deep phantom.

For the in vivo experiments with the ReST-Matrigel implant in the mouse flank, we overlayed blood-mixed agar with different thicknesses (5, 8, and 13 mm) on top of the mouse skin to mimic different imaging depths, shown in Figs. 6(e)–6(g). The photoacoustic images were presented in log scale to better depict the weak signals, whereas the ultrasound images were shown in linear scale. At 5-mm depth, the background signal amplitude from blood vessels was strong at both 20°C and 40°C [Fig. 6(e)]. With the strong background signals, we were not able to visualize the red ReST implant at 20°C or distinguish black and red ReST implants at 40°C. In contrast, the differential photoacoustic image clearly showed the black and red ReST implant location, with much reduced background blood signals. Additional imaging was performed at 8- and 13-mm depth [Figs. 6(f) and 6(g)]. The differential images at all depths clearly improved the contrast of the ReST implants, even though the photoacoustic signal amplitude decreased with imaging depth due to light attenuation.