2.1.

System Description

Our ultrasound imaging system is composed of a 1.75-D ultrasound phase array purchased from Tetrad Corporation (now W. L. Gore and Associates, Incorporated, Newark, Delaware) and electronic circuitry developed in our laboratory. The array has ten rows in elevation and 128 elements in each row, totaling 1280 elements (see Fig. 1 ). We have previously demonstrated that it is capable of electronically scanning a volume of 80 and 20 deg in the azimuth and elevation directions, respectively.^{39, 40} The transducer operates at a central frequency of 5 MHz with 60% bandwidth. The element height in the elevation direction is 0.97 mm, and the element width in azimuth is 0.224 mm.⁴¹ An acoustic lens focuses the beam in elevation 40 mm away from the transducer. The 1280-pin outputs are bundled in 40 groups of cables, each containing 32 thin 50-ohm coax cables of 1.75 m in length.

Fig. 1

The 1.75-D ultrasound transducer array. (a) Actual picture with coaxial cable interconnection. (b) Scanning geometry diagram.

The circuitry consists of four identical 320-channel transmit (TX) and receive multiplexer (MUX) unit pairs (Fig. 2 ). Each TX board provides dedicated transmission drivers for 320 channels with transmit/receive isolation. The TX unit can drive each array element independently and simultaneously, with an adjustable time delay and pulse width resolution of 20 ns. The paired MUX units multiplex the 320 channels into ten parallel outputs that are independently amplified by two-stage variable gain (nominal = 60 dB) amplifiers and filtered with 1- to 10-MHz bandpass filters. A common data acquisition unit multiplexes the 4 × 10 = 40 channels from the for TX/MUX pairs for digitization by a 12-bit, 50-MS/s analog-to-digital converter. All modules are controlled with custom C-language software in the host PC through two National Instruments (Austin, Texas) PCI-DIO-32HS high-speed digital I/O cards.

Fig. 2

3-D ultrasound and photoacoustic imaging system. (a) Schematic diagram of the system. (b) Actual photograph of the system with the array.

The system can operate under two modes: conventional ultrasound pulse-echo mode and photoacoustic mode. Both the 3-D ultrasound images and photoacoustic images are reconstructed using a typical delay and sum algorithm based on the transducer array geometry. The algorithm has been implemented in C-language.

In conventional pulse-echo ultrasound imaging mode, users can select a fixed transmission focal depth, number of scan angles, and total range in azimuth and elevation. The transmission circuitry generates high voltage pulses that activate the 1.75-D ultrasound array, and the desired acoustic beams are transmitted into the tissue. The return signals from the array are multiplexed, amplified, and sampled by the data acquisition unit. For each angular location, this transmission sequence is repeated to retrieve the data for all elements. After the data acquisition is finished, the acquired rf data is loaded into the computer memory for filtering, beam forming, and imaging. The rf data from all 1280 channels is used to form the beams. It takes approximately seven seconds to acquire the data and generate the beam for each angular direction. Dynamic focusing in reception has been implemented in software with focusing at every four samples.

In photoacoustic imaging mode, the transmission (TX) circuitry is disabled and only the MUX unit and DAQ unit are enabled. A Ti:Sapphire (Symphotics TII, LS-2134, Symphotics, Camarillo, California) optically pumped with a Q-switched Nd:YAG laser (Symphotics-TII, LS-2122), illuminates the sample orthogonal to the azimuth-axial imaging plane of the transducer. The laser delivers 8- to 12-ns pulses at 15 Hz with tunable wavelengths from 700 to 950 nm. The beam is diverged with a planoconcave lens and homogenized by a circular profile engineered diffuser (ED1-S20, ThorLabs, Newton, New Jersey) to produce uniform illumination on the sample. A scan starts by triggering the circuitry from the laser at the instant when the optical pulse illuminates the tissue. The received electrical signals are amplified, filtered, and digitized. Raw data are transferred into the host PC for further beam forming. It takes approximately one and a half minutes to retrieve the 1280-channel data under this mode. Then, similar to the pulse-echo mode, the acquired rf data is loaded into the computer memory for filtering, beam forming, and imaging. Delay and sum with dynamic focusing at every four samples is used to form the photoacoustic image.

2.2.

System Characterization

In pulse-echo mode, we measured the two-way transmission reception and the one-way transmission-only beam profiles using a 1.0-mm diameter polyvinylidine fluoride (PVDF) hydrophone. By acoustic reciprocity, the one-way transmission beam profile is equivalent to the one-way photoacoustic reception beam profile. The hydrophone, which was used both as a receiver and a target, is broadband with a flat frequency response between 1 and 10 MHz and a sensitivity of 0.1 μV/Pa (Force Institute, Copenhagen). It was connected to an ultrasound pulser-receiver (Panametrics 5072PR) to measure the ultrasound signals transmitted by the 1.75-D array. Both the hydrophone and the 1.75-D transducer array were submerged in castor oil, which has an acoustic absorption of 11.9 dB/cm at 5 MHz and 20 °C,⁴² close to that of soft tissue. The hydrophone, facing the 1.75-D array, was fixed at zero degrees in the azimuth and elevation directions with respect to the center of the array, and 40 mm depth in the axial direction. Sector scans of 40 and 20 deg were performed on the azimuth-axial and elevation-axial imaging planes respectively, while the transmission focus of the system was kept at 40 mm [Fig. 3a ]. For each individual angular scan, the transmitted signals were measured by the hydrophone, while the 1.75-D transducer measured the reflected signals from the hydrophone. The one-way beam profile of the system was obtained by measuring the peak value of the received signal envelope from the hydrophone at each angular position. The two-way beam profile was acquired by measuring the peak value of the envelope of the beam-formed transducer signal at each angular scan.

Fig. 3

System characterization and ovary imaging setups. (a) Experimental setup for one-way versus two-way measurements using a hydrophone as an active target. A similar setup was used for measurements across the field of view, where the hydrophone was replaced by a music string that was positioned at different locations. (b) Experimental setup for ultrasound and photoacoustic imaging of ovaries.

The field of view of the 1.75-D array was evaluated by measuring the two-way beam profiles in azimuth and elevation at different points across the field using a similar setup as discussed before. In this case, a rigid music string with a diameter of 0.5 mm was used as the point-like target instead of the hydrophone. The measurements were grouped in three sets. For the first set, the target was kept at zero degrees in the azimuth and elevation directions of the array, while the separation distance (axial) between the target and the array was varied to 20, 40, and 50 mm. At each axial position, measurements were acquired by focusing the ultrasound array to the location of the target. For the second set of measurements, the target was kept at zero degrees in the elevation plane and it was translated in the azimuth-axial plane while keeping its distance from the center of the array at 40 mm. The angular locations at which the measurements were taken for this set were 0, 20, and 39 deg. The transmission focus was kept at 40 mm for these three locations. For the last set, the target was fixed at zero degrees in the azimuth plane, while it was translated in the elevation-axial plane in an arc. The target locations, at which measurements were taken with a transmission focus fixed at 40 mm, were 0, 5, and 10 deg. For all three sets of measurements, the two-way beam profiles in azimuth and elevation were obtained by performing sector scans and measuring the peak values of the envelope of the formed beams at the corresponding depth of the target.

Resolution measurements for pulse-echo ultrasound and photoacoustic modes were performed using a black nylon thread with 80 μm in diameter. To measure the azimuth resolution, the thread was suspended perpendicular to the azimuth-axial imaging plane at a slight angle (<20 deg) to the incident laser beam. For the elevation resolution measurement, the thread was placed parallel to the azimuth-axial imaging plane and to the array front face. With the transducer and the thread submerged in water with a separation of 40 mm, sector scans were performed in the azimuth and in elevation. The measured −6-dB full-width half-maximum (FWHM) from the beam profiles of the thread was used to obtain the resolutions. Subtraction of the 80-μm thread diameter from the FWHM yields corrected resolution measurements both in azimuth and elevation. The results were compared to theoretical values. Since there is no theory for wideband transducer arrays, the narrowband theory is commonly used to estimate their performance.⁴³ For pulse-echo mode, the narrowband theory indicates that the beam profile follows a ${\mathrm{sinc}}^2(L_{x}x∕\lambda z)$ pattern in the azimuth direction, where $L_x$ is the transducer size in the azimuth direction $x$ , λ is the wavelength, and $z$ is the depth. One can estimate the azimuth resolution as the value of $x$ when the ${\mathrm{sinc}}^2$ function reduces to 0.5. In the elevation direction, the beam profile follows a ${\mathrm{sinc}}^2(L_{y}y∕\lambda z)$ pattern, where $L_y$ is the dimension of the transducer in the elevation direction $y$ . Following the same procedure described for estimating the azimuth resolution, the elevation resolution can be obtained. In photoacoustic mode, the beam profiles follow similar patterns, with the exception that they are given by a sinc function pattern instead of a ${\mathrm{sinc}}^2$ function.

2.3.

Simulations

Further verification of our imaging system was achieved by comparing the system’s measured response to simulations. The simulations were performed using Field II simulation software^{44, 45} and Matlab^®. Field II can simulate pulsed and continuous wave pressure fields generated from ultrasound transducers and arrays of arbitrary shape, apodization, or excitation function. It comprises a set of libraries that can be called from within the Matlab^® programming environment. The libraries model the behavior of ultrasound transducers of arbitrary geometry, including an elevation focused linear multirow transducer that was used in this study.

2.4.

Ovary Imaging

Porcine ovaries freshly excised from local farms were studied with our imaging system. Ovaries were kept in a 0.9% w/v NaCl solution from the time they were removed from the animal and during the imaging process. Sector scans in the azimuth direction were performed both in pulse-echo ultrasound and photoacoustic mode at different elevations without mechanically moving the sample or the transducer. The sample was placed close to the 40-mm elevation focus of the transducer in the axial direction and centered in the azimuth direction [Fig. 3b]. In photoacoustic mode, the wavelength of the laser was adjusted to 740 nm and the incident power was maintained below $8\mathrm{mJ}∕{\mathrm{cm}}^2$ . The obtained ultrasound and photoacoustic images were superimposed to provide structural guidance for interpretation of the photoacoustic results.

After imaging experiments, the ovaries were submerged in 37% formalin at room temperature for 24 h. After fixation, they were rinsed with fresh water, placed in a 0.9% w/v NaCl solution, and kept at 4 °C. For histological evaluation, the ovaries were cut in 5-mm blocks parallel to the imaging plane, dehydrated with graded alcohol, embedded in paraffin, and sectioned to 7-μm thickness using a paraffin microtone. Slides corresponding to the imaged planes were stained using hematoxylin and eosin (HE).

3.1.

System Characterization

The measured one-way and two-way beam profiles using a hydrophone, both in azimuth and elevation, agree with simulation results as shown in Figs. 4a, 4b, 5a, 5b , respectively. In transmission or reception mode, the one-way radiation pattern of an ultrasound array at the focal zone is the Fourier transform of the array’s excitation aperture. In pulse-echo mode, the effective aperture is the autocorrelation of the array’s aperture pattern.⁴⁶ Hence, the two-way beam profile is equivalent to squaring the one-way profile. In Figs. 4c and 5c, the measured one-way and two-way azimuth and elevation beam profiles are shown. Because of the logarithmic scale used, a two-fold difference between the one-way and two-way profiles is obtained as predicted by theory.

Fig. 4

Measured and simulated azimuth beam profiles of a 40-deg sector scan: (a) transmit only, (b) pulse-echo, and (c) transmit only versus pulse-echo.

Fig. 5

Measured and simulated elevation beam profiles of a 20-deg sector scan: (a) transmit only, (b) pulse-echo, and (c) transmit only versus pulse-echo.

From the first group of measurements across the field of view, where the sample was moved axially, it can be observed [Figs. 6a and 6b ] that the beam profile, both in azimuth and elevation directions, remain fairly constant at 40 and 50 mm away from the transducer on the central axis. At 20 mm, the measured beam profiles get wider, which translates into lower angular resolution at this point. This widening is due to the limited individual element directivity, which has a FWHM at −6 dB of 76 deg in azimuth and 18 deg in elevation. Directivity imposes a limit on the number of elements that effectively receive signal from a given point in the field of view of the array. When imaging a point target on the central axis, theory suggests that for our 1.75-D transducer array, the array effective aperture reduces for points within 30 mm from the transducer array, reducing the resolution at these points.

Fig. 6

Two-way beam profile measurements using a music-wire target. First group of measurements at various axial depths: (a) azimuth beam profile and (b) elevation beam profile. Second group of measurements at various angular positions in the azimuth-axial plane: (c) azimuth beam profile and (d) elevation beam profile. Third group of measurements at various angular positions in the elevation-axial plane: (e) azimuth beam profile and (f) elevation beam profile.

Similarly, for the second group of measurements across the field of view, the results show [Figs. 6c and 6d] expected variations in the width of the azimuth and elevation beam profiles due to directivity and steering effects. At 40 deg in the azimuth, the beam profiles are the broadest of all at the three measured locations. At this point the width of the elevation beam profile has almost doubled compared to the no-steering beam profile. On the other hand, for the third group of measurements moving the target in elevation, no significant change of the beam widths is observed due to the small focusing deflections (up to 10 deg) used.

The measured resolution in pulse-echo mode, both in azimuth and elevation, agree very well with theoretical values. From the narrowband theory, the FWHM of the main lobe for the 1.75-D transducer is estimated to be 0.45 deg in azimuth and 1.55 deg in elevation 40 mm away from the transducer. The measured resolutions, after subtracting the diameter of the target thread, are 0.51 deg in azimuth and 1.52 deg in elevation for the pulse-echo mode. Similarly, for photoacoustic mode, the narrowband theory suggests the resolution is 0.60 deg in azimuth and 2.13 deg in elevation. The measured resolutions in photoacoustic mode are 0.76 deg in azimuth and 2.55 deg in elevation, after subtracting the diameter of the target. A plot comparing the resolution measurement results in pulse-echo and photoacoustic modes is shown in Fig. 7 .

Fig. 7

Resolution measurement in pulse-echo and photoacoustic modes performed at the elevation focus: (a) azimuth resolution and (b) elevation resolution.

3.2.

Ovary Imaging

Coregistered images of the ovaries show excellent correlation of the ultrasound and photoacoustic features when compared to histological images. In all three cases, ultrasound reveals the presence of antral follicles. The ovarian follicle consists of the oocyte and the follicular envelope, which is composed of granulosa cells and an outer membrane. Follicles migrate from the innermost part of the cortex to the surface as they mature to become antral follicles, which increase fluid accumulation in their antrum.⁴⁷ Antral follicles are easily imaged with ultrasound and appear hypoechoic. As the follicles are located in the cortex of the ovary, which is comprised mostly of dense cellular connective tissue with hyperechoic appearance on ultrasound, excellent contrast with ultrasound imaging is generally achieved between the follicular fluid and the cortex. Figures 8a, 8c, 8e show a series of pulse-echo ultrasound images of the first excised ovary of a young, but sexually mature pig. These images correspond to different elevation planes obtained by changing the electronic focus in steps of 0.7 deg. One can distinguish many antral follicles across the whole ovary. Figs. 8b, 8d, 8f show the coregistered photoacoustic images superimposed on top of the pulse-echo images. Optical absorption occurs in the central area of the ovary along the long axis, but no absorption is observed around the follicles or the cortex of the ovary itself. The highly absorptive area corresponds to the medulla, where blood vessels, lymphatic vessels, and nerves run from the mesovarium, which attaches the ovary to the uterus via the broad ligament.⁴⁸ Excellent agreement between the coregistered images and HE staining of a corresponding histological section of the ovary is apparent when comparing the previous images to Fig. 8g.

Fig. 8

Coregistered images of ovary 1 at different elevation planes. (a), (c), and (d) show ultrasound-only images. (b), (d), and (f) show photoacoustic images on top of the ultrasound images. Each pair of ultrasound and coregistered images was obtained by changing the electronic focus in elevation in steps of 0.7 deg. (g) HE stained histological slide. The white bar represents 5 mm.

Antral follicles vary in size; they grow from less than a millimeter to a few centimeters. This is more evident in the preovulation stage, where many changes take place in the preovulatory follicle. Shortly before follicular rupture, expansion of the follicle occurs with marked hyperaemia in the theca of the preovulatory follicle due to increased vascularity and increased permeability of its capillaries.⁴⁷ These changes can be observed in the coregistered ultrasound and photoacoustic image of the second ovary, as shown in Fig. 9 . A 1 cm in diameter antral follicle can be observed in the ultrasound image, while optical absorption due to the increase of blood in the theca stands out around the follicle in the photoacoustic image. This corresponds well to histology as indicated by the smaller arrow in Fig. 9c.

Fig. 9

Coregistered images of ovary 2. (a) Ultrasound-only image. (b) Photoacoustic image on top of the ultrasound image. (c) HE stained histological slide. The white bar represents 5 mm.

In primates and pigs, the antrum of the follicle fills with blood and lymph after ovulation, and vessels from the theca grow into the antrum and branch to form a network of capillaries and fibroblasts that develop into the corpus luteum.⁴⁷ This process can be seen in the corpus luteum shown in Fig. 9c larger arrow), which is partially filled with blood and connective tissue. Poor contrast is observed with pulse-echo ultrasound, but good contrast is achieved with photoacoustic imaging due to optical absorption in the surrounding theca and the blood-filled antrum.

Figure 10 shows the results from our third imaged ovary. In the ultrasound image, the only features that stand out are two antral follicles that are just under 5 mm in diameter [smaller arrows in figure 10c]. However, photoacoustic imaging reveals the presence of a highly absorptive area, which represents a hemorrhagic corpus luteum [larger arrow in figure 10c]. In addition, photoacoustic imaging shows absorption around the antral follicles due to the vascularity in the theca of the follicles. HE staining of the corresponding histological slide shows excellent agreement with the follicles seen by ultrasound and the corpus luteum seen by photoacoustic imaging.

Fig. 10

Co-registered image of ovary 3. (a) Ultrasound-only image. (b) Photoacoustic image on top of the ultrasound image. (c) HE stained histological slide. The white bar represents 5 mm.