Noninvasive visualization of electrical conductivity in tissues at the micrometer scale

Principle of RThAM

Figure 1A illustrates a schematic of RThAM (see details in Materials and Methods, fig. S1, and notes S1 and S2). RThAM requires nanosecond pulses and a substantial energy density in the form of an electrical field to generate detectable thermoacoustic signals (30). On the basis of the principle of reactive near-field coupling (35), substantial amounts of pulsed RF energy in nanoseconds could be deposited locally to stimulate sufficient Joule heating in conductive tissue for microscopic thermoacoustic imaging. Accordingly, we engineered an open transmission line (open TxLine; similar to twin leads) that is made of parallel plates. Compared to previous efforts using near-field coupling to generate high-strength thermoacoustic signals (32, 35–37), the open TxLine of RThAM “concentrates” the nanosecond pulsed RF energy homogeneously in the form of an electric field between the plates, like in parallel-plate capacitors (note S1). Exposed to an electric field that is stronger than the Johnson noise-equivalent field strength, a charged ion is accelerated along or against the electric field lines depending on the charge types. The added kinetic energy of the ion dissipates as heat due to collisions with other particles in the medium, causing a temperature rise and the subsequent thermal expansion, thus eventually emanating thermoacoustic waves (dashed red circles in Fig. 1A). The excitation source we used is a commercialized RF pulser built with a drift step recovery diode (FPG 10-10NK1, FID GmbH), delivering a near-Gaussian pulse with a full width at half maximum (FWHM) of 1.5 ns and an excitation frequency up to 300 MHz at −6 dB. To optimally couple the RF energy to tissue, we used silicone oil with a low dielectric constant (ε_r = 2.45 at 20°C; PSF-2cSt, Clearco Products), which avoids field strength decrease because of high permittivity dielectrics like water (ε_r = 80.1 at 20°C). An RF pulse energy of up to 2.91 mJ was thus coupled in the cross-sectional area between the TxLine (note S1), allowing high-strength thermoacoustic signal generation. However, thermoacoustic wave is dissipated severely in silicone oil (note S2). Because water attenuates US waves less than silicone oil (5 dB/mm in silicone oil, against 0.97 dB/mm in water at 50 MHz), we therefore attached a deionized water bag in front of a transducer (central frequency, 50 MHz; 90.66% bandwidth; focus, 6 mm; diameter, 6.35 mm; V3330, Olympus NDT) to couple the thermoacoustic waves. Thermoacoustic signals were amplified at the US transducer using a 63-dB low-noise amplifier (AU-1291, MITEQ) and then digitized by a 500 MSamples/s PCI (peripheral component interconnect) data acquisition card (CompuScope 12502-128M, Gage Applied). Raster scanning was performed by controlling two linear stages to collect two-dimensional (2D) grid data points. The 3D signals were reconstructed into images using a beam-forming algorithm (30). From the reconstructed 3D image, we can analyze either a single slice or a maximum intensity projection (MIP) of a stack of the consecutive slices, representative of the conductivity distribution of the sample.

To image conductivity only (excluding permittivity), we chose an excitation frequency below 300 MHz. In this range, dielectric absorption by polar molecules is minimal (32, 38), as shown by both theory (39) and experiment (36, 38). For example, the dielectric-to-conductivity absorption ratio is about 1% at frequencies between 13.67 and 36.335 MHz (<5% up to 300 MHz), corresponding to the current RThAM US detection bandwidth of 27.34 to 72.67 MHz according to the second harmonic generation principle (36). We have confirmed the linear relation between electrical conductivity and thermoacoustic responses at low frequencies [low compared to the picosecond relaxation time of polar molecules (40)] using blood (36) and aqueous NaCl solutions (36, 38). RThAM is thus designed to show predominantly contrast from conductivity, especially for tissues with a high ionic content, such as blood, body fluids, cerebellum fluids, gray matter in the brain, retinas, corneas, and the aqueous humor in the eye (39).

RThAM imaging performance

To characterize the RThAM setup, we raster-scanned the transducer in 10-μm steps over a 2.2 × 2.2 mm² field of view (FOV) with a pair of copper wires (Ø50 μm; electrical conductivity σ = 5.8 × 10⁷ S/m) placed in the TxLine made of water as dielectrics (note S2). Figure 1B shows the resulting thermoacoustic mesoscopic image that was formed by an MIP along the depth/z axis (see Materials and Methods for signal processing and image formation). Figure 1C shows a cross-sectional image in the yz plane, and Fig. 1D shows the profiles of the copper wire in the lateral and axial directions. Inspection of the copper wire profiles reveals an FWHM of 23 μm in the z direction (axial) and 33 μm in the xy plane (lateral). The apparent axial FWHM is a result of the 13-μm skin depth (at which depth the RF field strength decreases to 1/e and the energy to 1/e²) of the 25-MHz RF field in copper (fig. S1G, and note S3), while the apparent lateral FWHM is a result of the cylindrical shape of the copper wire and the limited acceptance angle of the US transducer. Figure 1E shows a temporal sequence of the maximum RThAM amplitude. A 34-dB signal-to-noise ratio (SNR) was achieved due to both the excellent RF energy coupling to the sample and the high electrical conductivity/RF absorption of copper, which affords high sensitivity and a broad dynamic range for conductivity measurements. Fourier analysis (blue line) shows that the signal contains frequency components centered at ~50 MHz, extending to 100 MHz, which reflects the detection bandwidth of the transducer. Because of the second harmonic generation of the thermoacoustic signal (36), the induced RThAM wave oscillates at the second harmonic of the excitation RF. Therefore, the detected 50-MHz RThAM signal corresponds to the absorption of RF at 25 MHz, with a wavelength of 12 m. The achieved RThAM resolution is about 0.5 × 10⁶ times shorter than the excitation wavelength. Even though the super-resolution ratio remains the same, the resolution is scalable to microscopic regime by applying higher frequencies of excitation and detection, e.g., using 100-MHz US detection and the same phantom we managed to image 7 and 16 μm in axial and lateral resolution, respectively (fig. S2 and note S4). Note that a long oil path in the hybrid oil/water coupling medium of RThAM could deteriorate the acoustic resolution (fig. S3, A and B and note S5). To minimize this deterioration, we used the sample bed to position the sample close to the water bag so that the oil path was less than 1 mm. Because RThAM resolution is acoustically limited, similar to conventional optoacoustic mesoscopy (30, 41), the polarization of the excitation field does not change the resolution, but higher contrast can result from longitudinal objects that are aligned with field polarization compared to those that are perpendicular to the field (fig. S3, C to E, and note S6).

Imaging single-cell stage embryos

To test our hypothesis that RF absorption could be exploited to resolve microscale spatial variation of electrical conductivity, we imaged zebrafish (Danio rerio) embryos using our hybrid oil/water coupling RThAM system. The single-cell embryo offers a simple anatomical structure with marked conductivity differences (42); since the egg yolk consists primarily of lipids, the egg cell comprises a blastodisc where genetic materials are contained and duplicated and the albumen area containing proteins for development. We chose this compartmental structure as an imaging model to evaluate the RThAM ability to resolve conductivity contrast in vivo. Another advantageous feature of this imaging model is that the embryo is a self-contained system protected by chorion, uncontaminated by extrinsic ions. Figure 2A shows an optical bright-field microscopic image of a one-cell stage (zygote period) zebrafish embryo at ¹/₂ hour post-fertilization (hpf). The localized blastodisc in the animal pole and a large yolk are clearly visible. For subsequent RThAM imaging, the egg cell was immersed in silicone oil intact (Fig. 1A). Figure 2B shows the RThAM image of the embryo formed by MIP and Fig. 2C shows an overlay of the RThAM image on the bright-field microscopic photo. Movie S1 shows the 3D distribution of RThAM contrast of the embryo. Visually, the brightest region in the RThAM image corresponds to the developing blastodisc. Strong RThAM response suggests enhanced RF absorption, high electrical conductivity, and therefore high ion concentration in the developing cell region. We show in Fig. 2D a profile across the blastodisc and yolk. The RThAM signal from the blastodisc area was found to be on average five times higher than from the yolk area. We would expect the blastodisc-to-yolk signal ratio to be even higher. However, there is a background signal (~20 mV in Fig. 2D), which might be a result of an ionic content in the surrounding albumen of the embryo (movie S1). Besides the apparently rich contrast, we also noticed some fine features of size <38 μm in the yolk region. They might be a result of the propagation of duct-type structures or angiogenesis originated from the animal pole that is developing into the embryonic fish body. Further studies are needed to confirm correlations between the RThAM images and known fine structures of within a zebrafish egg.

Multicellular stage structures

To visualize more of these fine and complex multicellular structures in three dimensions, we imaged an embryo in a later developmental stage with RThAM (512 cells). An embryo at the 512-cell stage is interesting for 3D imaging because finer structures form as the mid-blastula transition starts (42)—the deep cells in between the yolk syncytial layer and the outermost enveloping cells split asynchronously, becoming motile and streaming over the yolk toward periphery. Figure 2E shows a bright-field microscope image of the zebrafish embryo 4 hpf, in which the blastomeres were at the 512-cell stage. We would again expect a stronger signal from the more ion-rich region containing the blastomeres. Figure 2 (F and G) shows the RThAM MIP image of the embryo and the corresponding overlay with the bright-field image. Movie S2 showed the 3D distribution of the RThAM contrast. While it appears that there are strong signals from both the blastomere and yolk regions, a cross-sectional image in the xz plane (Fig. 2H and movie S2) indicates that the apparent signals from the yolk actually originate from the embryo surface. This again supports the idea that the yolk consists of mostly non-RF absorbing material (primarily lipids). The distribution of RThAM signals in 3D that were surrounding the yolk probably reflected the spread of cells from animal pole in the developing embryo. It is further apparent from Fig. 2H that the most intense signals emanate from the area corresponding to the blastodisc center, where there is a high density of splitting deep cells. Figure S4 (A and B) shows a 200-μm slice of the RThAM image in the xy plane at the depth of the blastomeres’ center and an overlay of this slice on the bright-field image. Within this image slice, a maximum SNR of 29 dB (fig. S4C) and features as small as 45 μm (fig. S4D) were recorded.

Contrast in zebrafish larvae

We further probed the limits of RThAM by imaging zebrafish larvae at more advanced stages [<5 days post-fertilization (dpf)], at which larvae contain discrete internal organs. Furthermore, the optical transparency of zebrafish larvae at this stage facilitates the comparison of RThAM and optical contrast. We expected to see regions with high ionic content (43), and thus high electrical conductivity compared to the background tissues, such as muscle (0.65 S/m at 25 MHz) or fat (0.06 S/m). Examples of these regions are blood (1.15 S/m) in the circulatory system, intestines (1.47 S/m) in the digestive system, urine (1.75 S/m) in the bladder, vitreous (1.50 S/m) and aqueous humor (2.01 S/m) in the eyes, or cerebrospinal fluid (2.01 S/m) in the brain (0.22 to 0.55 S/m). Figure 2I shows a transparent Casper-type zebrafish larva at 3 dpf in low-power optical reflection-mode microscopy, in which most internal organs and tissues of interested are clearly visible. Figure 2J shows a 150-μm-thick RThAM xy slice at the depth of the spinal cord, overlaid on a negative image of the optical contrast. Figure 2K shows the contrast distributions along depths obtained by projecting the 3D data cube in the x direction and forming a B-scan–like yz MIP image. High-RF energy absorption can be observed in areas that correlate with the eyes, blood/heart, forebrain/midbrain, and ventral areas containing the intestines, bladder, and pancreas. We also note some discernable RThAM contrast from the trunk, which emanates from muscle or dorsal/ventral blood vessels. We observed little contrast in the yolk sac region, which was expected because the yolk mainly consists of hydrophobic lipids. Similar contrasts were observed from 4-dpf wild-type larvae in both the sagittal (fig. S5, A to C) and the coronal positions (fig. S5, D to F, and movie S3). However, a precise delineation of the RThAM contrasts in terms of the optical contrast was not possible.