Photoacoustic imaging provides optical contrast with improved tissue penetration and spatial resolution compared to pure optical techniques. Three-dimensional photoacoustic imaging is particularly advantageous for visualizing non-planar light absorbing structures, such as blood vessels, internal organs or tumours. We have developed a fast 3-D photoacoustic imaging system for small animal research based on a sparse array of ultrasonic detectors and iterative image reconstruction. The system can acquire 3-D images with a single laser-shot at a frame rate of 10 Hz. To demonstrate the imaging capabilities we have constructed phantoms made of a scanning point source and a rotating line object and imaged them at a rate of 10 frames per second. The resulting 4-D photoacoustic images depicted well the motion of each target. Comparison of the perceived motion in the images with the known velocity of the target showed good agreement. To our knowledge, this is the first report of single-shot high frame-rate 3-D photoacoustic imaging system. With further developments, this system could bring to bear its inherent speed for applications in small animal research, such as motion tracking of tumour outline during respiration, and rapid monitoring of contrast agent kinetics.
© 2008 Optical Society of America
Photoacoustic Imaging (PAI) is a promising hybrid modality capable of providing image contrast by optical absorption, but with tissue penetration and spatial resolution at depth that is superior to pure optical imaging methods . PAI has been employed in-vivo to visualize blood-vessel structures [2–4] and tumour angiogenesis [5–7], detect breast tumors [8–10], estimate oxygenation levels [11–13], perform functional imaging [2, 14], and track molecular probes .
The technique is based on volume irradiation with a short laser pulse followed by preferential energy deposition at the optically absorbing structures within the volume. These structures heat up slightly, but so rapidly that the condition of thermal confinement is met and thermo-elastic expansion leads to a transient bipolar pressure wave that propagates outwards toward the surface . The frequency composition of the pressure transient (i.e. the “photoacoustic wave”) is typically in the ultrasonic range , and detectable by wide-band ultrasound detector(s) placed in acoustic contact with the surface of the volume. Based on the time-domain surface measurements, the distribution of the photoacoustic (PA) sources (structures) inside the volume can be inferred using an image reconstruction algorithm, several of which have been suggested and adapted from medical imaging [18–22].
The photoacoustic process is intrinsically three-dimensional. This is a product of the optical radiation scattered in tissue, which diffusely illuminates a volume, rather than a single point, line, or plane. Furthermore, the PA waves are of a spherical nature and propagate in all directions from their point of generation. The advantages of 3-D imaging for in-vivo studies are obvious: visualization of non-planar structures, and easier orientation and interpretation. Several approaches have been suggested for 3-D PAI and are differentiated by the detection scheme. These can be divided into scanning methods, where a single detector is scanned along the detection surface in two dimensions [2–4, 23, 24], staring methods, where a 2-D array of detectors is used and no scanning is necessary [10, 25, 26] and a combined scanning-staring approach, where a linear array or combination of single detectors is mechanically scanned [8, 27–31].
High frame-rate PA imaging has been mostly reported in 2-D, by the use of a linear transducer array. Zemp et al.  have achieved real-time acquisition and display at 50 frames-per-second (fps) by using a multiplexed 48-channel linear array and a laser with 1-KHz repetition rate. Niederhauser et al.  have demonstrated real-time acquisition and display at 7.5 fps with a 64-channel array and a single laser shot (no multiplexing). Dean et al. and Liao et al. [34, 35] presented real-time acquisition with off-line image reconstruction and display at 10 fps and 15 fps, respectively.
The extension of the above mentioned real-time PAI method to 3-D would require scanning the linear array perpendicular to its length, imaging the 3-D volume slice by slice. Such a system has been recently realized by Song et al.  using a 1-KHz repetition rate laser and yielded a frame-rate of 1 Hz. A total of 996 laser pulses were fired for each 3-D frame, resulting in significant heat deposition in the sample, hence limiting the allowable laser pulse energy and the number of frames that could be acquired.
An alternative approach to fast 3-D PAI would be to use a two-dimensional staring array. However, with current approaches of using dense arrays counting hundreds of elements [10, 36], the electronic system required for parallel data acquisition of all channels at high frame-rate becomes prohibitively expensive. We have previously reported on a method for 3-D PAI based on sparse-array detection and iterative image reconstruction . The method provides fast image acquisition, since no scanning of the detectors is required. It is also simple and inexpensive, thanks to the low number of detectors and corresponding data acquisition channels. We have reported results where phantoms consisting of arrays of point sources and of a line target were imaged in 3-D, using single laser shot illumination. This suggested that consecutive single-shot frames could be imaged in similar manner to obtain a 3-D PA movie (i.e. 4-D photoacoustic imaging) at a frame-rate limited by the laser repetition rate. We have therefore developed an imaging protocol that concentrated on high frame-rate image acquisition, leaving the iterative image reconstruction to be performed off-line. For each laser pulse, the PA signals were captured in parallel from all elements of the sparse array, and stored in sequence for consecutive pulses. After the imaging sequence was completed the data was transferred to the PC and reconstructed. Our method therefore offered real-time acquisition at 10 frames-per-second with off-line display.
In this report, we present 3-D PA movies of moving light absorbing targets, taken in order to demonstrate and validate the fast image acquisition capabilities of the system. To our knowledge, this is the first report of high frame-rate 3-D PA imaging using a single laser shot per frame.
2.1 Photoacoustic imager
The imaging system, shown in Fig. 1, used sparse-array detection, whose principles were described in a prior publication . However, while previously we used an annular array of 15 custom-made PVDF detectors, the imager for this report consisted of 15 commercial ultrasound detectors (“V304”, 1-inch diameter, 2.25 MHz immersion transducer, Olympus- NDT, Waltham, Massachusetts, USA). The detectors were mounted on 5 custom-built curved holders, each holder supporting three detectors at elevation angles of 22.5°, 45° and 67.5°. The five holders were arranged along a circle with azimuthal separation of 72°. A transparent container was built around the detectors and was filled with filtered water for acoustic coupling. The rationale behind the choice of the new detector configuration was: (i) Commercial transducers are more reliable, reproducible and more stable over time than the hand-made detectors previously used. (ii) The curved geometry provided a more uniform coverage of the imaging volume in the elevation angles. This was expected to result in a more isotropic spatial resolution, whereas before, resolution was significantly better in z than in the x-y plane. (iii) The center frequency of the detectors was chosen to match the target resolution (1 mm) and the previous PVDF detectors (estimated at 1.5 MHz, see ). (iv) Detectors were chosen with an aperture of 1 inch in order to achieve a collimated sensitivity field that attenuated contributions from photoacoustic sources outside the volume. As explained in the Discussion section, this was part of the design to avoid boundary artefacts.
The laser illumination was directed from below the tank, through an optical window at the bottom between the detector holders. For experiments in water, an OPO-coupled Nd:YAG laser was used with an energy of 20 mJ per pulse (“Vibrant”, Opotek, Carlsbad, California, USA). For experiments in optically-scattering medium, a more energetic OPO-coupled Nd:YAG laser was employed with an energy of 35 mJ per pulse (“Surelite III”+“SLOPO”, Continuum, Santa Clara, California, USA). The wavelength for both sets of experiments was 700 nm. The object to be imaged was suspended from above at the desired position relative to the detectors. For each laser pulse, the analog photoacoustic signals from the detectors were filtered and amplified by a dedicated electronics box and then transmitted to an A/D board for digitization. The digital data was transmitted to a PC and reconstructed into an image by an iterative reconstruction algorithm. A detailed description of the laser, signal transfer and image reconstruction is available in our previous publication .
2.2 Calibration scan
Prior to any attempts to image objects, a calibration scan was carried out where a photoacoustic source was raster scanned through the volume of interest, and sensitivity and coverage maps were generated, in a similar fashion to the method originally proposed for the annular sparse array [37, 38]. However, the source used in previous publications was inadequate since its emission profile was limited in the elevation angular range. Instead, a different point source was used that had a more omni-directional emission profile . The source consisted of a 400 µm-core optical fibre exposed at one end, polished to a near-hemispherical shape and coated with black paint. The resulting emission profile was more omni-directional than the coated flat-ended fibre. The centre of the calibration-scan volume was chosen to align with the intersection of the detectors’ line-of-sight and the volume size was 25×25×25 mm3.
We used three phantoms - a scanning point source in water, a rotating graphite rod in water and a rotating light-scattering gel embedded with a graphite rod. The point source was the coated hemispherical fibre used for the calibration scan, illuminated by the “Vibrant” laser through the distal end of the fibre. It was mounted on a motorized translation stage and moved at a constant speed along either the x or the y axis. Since the same gantry was used for the calibration scan, which defined the imaging volume, the source scanning axes coincided with the principal axes of the image. The scanning speed of the source for both directions was measured during the experiment and was determined to be 4.5±0.2 mm/sec. The rotation experiment in water was carried out with a 0.9 mm diameter graphite rod, commonly used in mechanical pencils. The rod was held horizontally (parallel to the bottom of the tank) and was illuminated from below (Fig. 1(a)). The rod was attached to a DC motor whose rotation speed was controlled by a regulated DC power supply. During rotation the rod remained horizontal, i.e. in a single z plane. The rotation speed was measured during the experiment and was estimated by timing and averaging 10 full rotations. The angular velocity was determined to be 120±4 °/s. The light-scattering phantom consisted of a cylindrical gel made of agar at 1% w/v, mixed with IntralipidTM to produce tissue mimicking light-scattering. The concentration of the IntralipidTM was 1% v/v, a generally accepted value to represent tissue scatter [23, 40]. The dimensions of the cylindrical gel were 28 mm in diameter and 40 mm in length. The gel was mounted with its long axis aligned with the motor shaft, and the graphite rod was inserted horizontally ~10 mm above the bottom of the gel (Fig. 1(b)). The rod was completely hidden from view once inside the gel, as can be appreciated from the picture on the right of Fig. 1(b). The mounted phantom was rotated inside the water tank, and its angular velocity was measured to be 100±3 °/s. The absorbing rod inside the phantom was intended to rotate at a plane parallel to the bottom of the tank, but we had no indication if this was actually the case.
2.4 Movie creation (4-D photoacoustic imaging)
In order to create a 3-D movie, a sequence of 50 laser pulses was fired for a duration of 5 seconds while the target was moving. For each laser pulse, PA data was acquired and stored. After all frames were acquired, data was transferred to the PC and fed to the image reconstruction algorithm. The 4-D images were then loaded as a multi-volume entity into an image rendering software (“Analyze”, ver. 8.1, Mayo Clinic, Rochester, MN, USA) . A movie was created by rendering consecutive 3-D time frames, using maximum intensity projection. At the same time, a slice was selected from each frame that best represented the object in 2-D. The same slice was collected from all 50 frames and merged to create a 2-D movie. Both 2-D and 3-D movies are presented.
3.1 Translating target
Figure 2 shows results from the scanning point source experiment. PA movies of the target scanned in the positive x direction are presented both in 2-D (Fig. 2(a)) and in 3-D (Fig. 2(b)). Similarly, 2-D and 3-D movies of the target scanned in the positive y direction are presented in Fig. 2(c) and Fig. 2(d), respectively.
The reconstructed shape of the point source was used to compute the Point Spread Function (PSF) of the imager. (The actual size of the source was ~0.4 mm3). We estimated the PSF to be 2.0–2.5 mm in each direction. However, when the source location was close to the edge of the volume, the reconstructed shape became flattened, as can be seen in the left pane of Fig. 2(a) and Fig. 2(c). This effect of the finite reconstruction volume will be discussed further in the Discussion section. It is also worth noting the change in the shape and relative intensity of the source even when it was well within the reconstruction volume. Although this change was not as drastic as that observed near the edge of the volume, it indicated variability in the imaging quality at different locations in the reconstructed volume. It is not clear whether this effect arose from an inhomogeneous spatial resolution or from temporal variability, and needs further exploration.
3.2 Rotating target in water
Figure 3 shows results from the rotating rod experiment. PA movies of the 0.9 mm rod rotated counter clockwise are presented both in 2-D (Fig. 3(a)) and in 3-D (Fig. 3(b)). The 2-D frames in Fig. 3(a) represented well the linear shape of the rod, with significant contrast to background. It was apparent from the sequence that the center of rotation shifted slightly between frames, which correlated to an actual wobble of the rod holder. The perceived diameter of the rod was ~3 mm [estimated as the cross-sectional width indicated by the overlaid rectangle in the left pane of Fig. 3(a)]. This corresponded to an increase of 2 mm relative to the actual diameter; in agreement with the estimated PSF. Frame #10 in Fig. 3(a) showed an example of an image artefact that appeared throughout the sequence. It presented as a ghost image of part of the rod (indicated by the arrow), overlaid on the true image. The ghost image was probably a consequence of the piece-wise reconstruction of boundaries by the sparse array, which is further discussed in section 4.5. The intensity and quality of the reconstructed target varied significantly between frames. This was a combined result of the directionality of the acoustic waves emitted from the rod and the piece-wise detection of the sparse array, as discussed later in sections 4.3 and 4.5.
In order to test the quality of image representation, the 2-D movie of the rotating rod (Movie 5) was analyzed to determine the perceived rod movement through the sequence of images. This was achieved by visually fitting a line to the image of the rod in each frame and recording its angle relative to the x-axis. The graph in Fig. 4 shows the estimated angle as a function of frame number. As the rod rotated counter clockwise, the angle increased between frames. We expected the angle to increase at a constant rate, so a linear fit was applied to the data, and the slope was determined to be 12.8 °/frame, with R2=0.999. The deduced angular velocity was 128 °/s, in good agreement with the measured angular velocity of 120±4 °/s, indicating that the movie provided a reliable representation of the movement of the phantom. However, even though the overall motion was well represented, it was evident that on frame by frame basis the perceived angle of the rod did not increase smoothly. Rather, clusters of two-three frames of similar angle were noted followed by a “jump” to higher angle. This can also be observed in Fig. 4 for frames 17–19 or 37–39. We will offer an explanation to this phenomenon in section 4.5.
3.3 Rotating target in scattering medium
The imaging results of the rotating 0.9 mm rod embedded in the light-scattering gel are presented in Fig. 5. As the rod was inserted deep into the opaque gel cylinder, it was not possible to visually determine its orientation or to position it parallel to the floor of the imaging tank. Hence, we inadvertently obtained a PA object that could not be presented on any primary plane of the imaging volume and could only be represented fully in 3-D. Fig. 5(a) shows one x-y slice through the object at several time points. It was apparent that the rod did not lie in a single x-y plane. Rather, the oblong shape and its precession in the plane indicated that the rod was sectioned obliquely, and outside its center of rotation. The 3-D movie in Fig. 5(b) revealed an even more complex motion. In addition to the precession, the center of rotation shifted back and forth in the y direction, due to a wobble of the gel holder.
It was also observed that the shape of the rod in Fig. 5(b) changed with its orientation, in a similar way to the experiment in water. For some orientations, the rod retained its linear shape. There were five such orientations in a full 360° rotation. For intermediate orientations, the shape became broadened and distorted. As mentioned earlier, this was the result of the sparse detection scheme, and will be further discussed in section 4.5.
The period of the rotation was estimated by observing the number of frames it took for images to repeat themselves. As can be seen in Fig. 5(a), where frames 5 and 41 are almost identical, the period was determined to be 36 frames, or 3.6 seconds. This corresponded to a rotation speed of 100 °/s, in agreement with the measured phantom rotation of 100±3 °/s.
4.1 Frame rate
We have demonstrated the capability of high frame-rate PA imaging in 3-D. The staring, sparse detection array permitted acquisition of full 3-D images with a single laser shot and acquisition of 3-D movies with frame rate limited only by the laser repetition rate. The delay imposed by the data acquisition electronics was well below 1 ms, meaning that the frame rate could be increased, in principle, to as high as 1,000 frames/s if a 1 KHz repetition laser was used. In reality, the on-board memory for each channel was limited to storing 100 frames in the current configuration, so a trade-off between frame-rate and overall acquisition time needs to be considered.
4.2 Boundary effects
The translation experiments demonstrated an effect of the size of imaging volume on the reconstruction. As was apparent for both translated targets (Fig. 2(a) and Fig. 2(b)), when the object was located outside the imaging volume, artefacts in the form of ghost shells appeared in the volume originating from the object’s deduced position. As the object entered into the volume, it formed a smeared line, until it was far enough into the volume that its entirety could be reconstructed. This suggested that PA sources outside the reconstructed image volume manifest as bright objects on the volume boundary closest to their actual location. To reduce those boundary artefacts, it will be necessary to keep all the detectable PA sources within the imaging volume. One way would be to spatially constrain the laser illumination, albeit this solution may not be practical in tissue due to light scatter. Another, more realistic way may be to restrict the detection volume. Our detection scheme achieved this to some degree by the use of directional detectors (which limited the lateral extent of the sensitive region) combined with time-windowing (which limited the axial extent). Another method may be to increase the reconstruction volume. However, care should be taken to concurrently increase the voxel size in order to avoid scaling-up of the number of voxels in the volume, leading to instability in the iterative reconstruction algorithm.
The reconstruction of the rod in section 3.2 highlighted an important property of the sparse array. Due to the directional nature of the PA waves emitted from the rod, as opposed to the omni-directional emission of the point source, only detectors that viewed the rod perpendicular to its length could detect significant PA signals. As the rod rotated, the count of such detectors changed, and so did the quality and brightness of the reconstructed image. For some rod orientations, as little as five detectors could sense significant signals. This property of non-isotropic reconstruction of line sources is a direct result of the sparse-array arrangement and the piece-wise viewing angles associated with it, and it will be further discussed in sub-section 4.5. Some inhomogeneity was also observed for the scanned point source, which manifested itself in slight changes in the brightness and shape as it moved through the volume. However, these variations were small compared to those for the rotating rod. This suggested that the current sparse hemispherical arrangement of detectors is more suitable for imaging small or spherical objects which generate PA waves propagating in all directions, rather than extended linear sources whose reconstruction depends on their angular orientation relative to the detectors. It should be possible to improve the isotropy of line source images by adding detectors to the array, thereby reducing the effect of piece-wise coverage.
4.4 Spatial resolution
In our previous publication using a planar annular array of fifteen 3-mm diameter detectors , we estimated the PSF to be 1.5 mm in the z direction, and 2.5 mm in the x and y directions. The curved modular array used in this study yielded a more uniform PSF of 2.0–2.5 mm in each coordinate direction. This was plainly a result of the detector arrangement, since the annular array had all detectors pointing at the volume with the same elevation angle (~30°), which resulted in better resolution of boundaries in the z direction than in the x-y plane. On the other hand, the current curved array had detectors distributed evenly over multiple elevation angles of 22.5°, 45° and 67.5°, which led to greater symmetry in the spatial resolution across all three coordinate directions.
4.5 Temporal resolution
As was noted in the Results section 3.2, the reconstruction of the rotating rod did produce a reliable representation of the overall movement, but had a difficult time depicting small angular changes. Rather, it stuck at certain angles for two to three frames, and then skipped to the higher angle. This discontinuous representation of the rod rotation was directly related to the piece-wise coverage of the volume by the sparse array. It has been shown [42, 43] that boundaries in PA imaging can be accurately reconstructed only when the normal to the boundary intersects the detection surface. For the rod experiment this implied that good reconstruction was possible only for certain rod angles, where the azimuthal orientation of the rod was perpendicular to one of the five groups of detectors in the sparse array. As the rod rotated into intermediate positions, its boundaries became less defined. When the rod entered a new region of optimal detection, the boundaries re-formed and an apparent jump in orientation occurred. Similar effect was observed with the rod in the scattering gel: five orientations produced good reconstruction of the rod shape, while intermediate ones resulted in poorly defined boundaries. This explanation was further supported by the translating point source experiments, where the motion between frames appeared continuous and the observable change in position was limited only by the spatial resolution of the system. This smooth motion was due to the fact that a point source does not have a preferred boundary. As the point source was translated, the boundaries that were well reconstructed may have changed slightly, but the overall shape was not greatly affected. These results suggested that the sparse array may be more suitable for imaging translational movements, since the same boundary of the object will face the detector and be reconstructed over time, which will result in a continuous transition between frames.
4.6 Further developments
The results in this work show preliminary 4-D imaging capabilities at laser repetition rate. However, the system in its current state has some limitations in the form of low resolution and image artefacts. We plan the following improvements to address and mitigate the limitations:
i) Our results suggest that the sensitivity to orientation of line objects can be significantly reduced by doubling the number of detectors. This will bring the detector count to 30, still relatively low and readily implemented by expanding our modular data-acquisition system.
ii) For the current work, we calibrated the system at a 5 mm step size, and used interpolation for voxels in between the calibration points. We plan to implement a calibration scan at a 1mm step size, which will resemble the density of the voxels in the imaging volume. This is expected to improve the accuracy of the calculations and the quality of the reconstructed images.
iii) We intend to incorporate broader bandwidth detectors to the array, thereby increasing the higher-frequency content of the signals, which in turn could improve the spatial resolution.
iv) We are working to improve the reconstruction algorithm through theoretical simulations that may shed light on the limitations imposed by the sparse nature of the detector array.
4.7 Potential applications
The existing system appears to be well suited for imaging strongly absorbing objects that are moving rapidly or that are sensitive to translational motion artefacts.
Recent work by Lao et al.  demonstrated that 2-D PAI could be used to track neovascularisation during tumour growth. Our technique does not have the power to resolve individual blood vessels, but the increased blood content in the tumour should provide enough contrast to allow imaging of the contour of the tumour, at least at the surface closer to the skin (back surface may not be visible due to strong absorption of light). Taking 3-D snapshot images of the tumour contour, unaffected by breathing artefacts, or even tracking the motion of the tumour contour during breathing should be possible. Another application could be monitoring the kinetics of exogenous contrast agent as it moves into and out of the tumour. Again, this will not result in metrics on an individual blood vessel, due to the limited resolution, but rather in measures averaged over the tumour volume.
We have shown the capability to capture fast 3-D images at a frame rate limited only by the laser pulse repetition. 3-D movies of moving targets were presented and analyzed in terms of accuracy, resolution and artefacts. It was noted that the overall movement of the targets was reconstructed accurately, with difficulty tracking small rotational changes. Although the system is still in development and requires further improvement, the capabilities demonstrated so far are by themselves of significance and might bring about unexpected applications in small animal research where rapidly-moving high-contrast targets are concerned.
The authors would like to acknowledge Lynn Keenliside and John Patrick for their valuable technical help. PE was supported by a CGS-D Scholarship from the Natural Sciences and Engineering Research Council of Canada (NSERC) and by the University of Western Ontario (UWO). The research was supported by a Discovery grant from NSERC (JJLC), the UWO Academic Development Fund (JJLC), the Canadian Foundation for Innovation Leading Edge Fund (FSP and JJLC), the Ontario Research and Development Challenge Fund (ORDCF) OCCI and BRAIN (FSP and Dr. Terry Thompson) and through research support from MultiMagnetics Inc.
References and links
1. L. V. Wang, “Ultrasound-mediated biophotonic imaging: a review of acousto-optical tomography and photoacoustic tomography,” Disease Markers 19, 123–138 (2003).
2. H. F. Zhang, K. Maslov, G. Stoica, and L. V. Wang, “Functional photoacoustic microscopy for highresolution and noninvasive in vivo imaging,” Nature Biotech. 24, 848–851 (2006). [CrossRef]
4. E. Z. Zhang, J. Laufer, and P. Beard, “Three-dimensional photoacoustic imaging of vascular anatomy in small animals using an optical detection system,” in Photons Plus Ultrasound: Imaging and Sensing 2007: The Eighth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics(SPIE, San Jose, CA, USA, 2007), 6437, pp. 64370S–64378.
5. G. F. Lungu, M.-L. Li, X. Xie, L. V. Wang, and G. Stoica, “In vivo imaging and characterization of hypoxia-induced neovascularization and tumor invasion,” Inter. J. Oncol. 30, 45–54 (2007).
6. L. Yeqi, X. Da, Y. Sihua, and X. Liangzhong, “Noninvasive photoacoustic imaging of the developing vasculature during early tumor growth,” Phys. Med. Biol. 53, 4203–4212 (2008). [CrossRef]
7. E. Z. Zhang, J. Laufer, R. B. Pedley, and P. Beard, “3D photoacoustic imaging system for in vivo studies of small animal models,” in Photons Plus Ultrasound: Imaging and Sensing 2008: The Ninth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics(SPIE, San Jose, CA, USA, 2008), 6856, pp. 68560P–68568.
8. R. A. Kruger, K. D. Miller, H. E. Reynolds, W. L. Kiser, D. R. Reinecke, and G. A. Kruger, “Breast Cancer in Vivo: Contrast Enhancement with Thermoacoustic CT at 434 MHz-Feasibility Study,” Radiology 216, 279–283 (2000). [PubMed]
9. S. Ermilov, A. Stein, A. Conjusteau, R. Gharieb, R. Lacewell, T. Miller, S. Thompson, P. Otto, B. McCorvey, T. Khamapirad, M. Leonard, and A. Oraevsky, “Detection and noninvasive diagnostics of breast cancer with 2-color laser optoacoustic imaging system,” in Photons Plus Ultrasound: Imaging and Sensing 2007: The Eighth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics(SPIE, San Jose, CA, USA, 2007), 6437, pp. 643703–643711.
10. S. Manohar, S. E. Vaartjes, J. G. C. van Hespen, J. M. Klaase, F. M. van den Engh, A. K. H. The, W. Steenbergen, and T. G. van Leeuwen, “Region-of-interest breast images with the Twente Photoacoustic Mammoscope (PAM),” in Photons Plus Ultrasound: Imaging and Sensing 2007: The Eighth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics (SPIE, San Jose, CA, USA, 2007), 6437, pp. 643702–643709.
11. W. Xueding, X. Xueyi, K. Geng, L. V. Wang, and G. Stoica, “Noninvasive imaging of hemoglobin concentration and oxygenation in the rat brain using high-resolution photoacoustic tomography,” J. Biomed. Opt. 11, 24015–24011 (2006). [CrossRef]
12. J. Laufer, E. Zhang, and P. Beard, “Quantitative in-vivo measurements of blood oxygen saturation using multiwavelength photoacoustic imaging,” in Photons Plus Ultrasound: Imaging and Sensing 2007: The Eighth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics(SPIE, San Jose, CA, USA, 2007), 6437, pp. 64371Z–64379.
13. H. F. Zhang, K. Maslov, M. Sivaramakrishnan, G. Stoica, and L. V. Wang, “Imaging of hemoglobin oxygen saturation variations in single vessels in vivo using photoacoustic microscopy,” Appl. Phys. Lett. 90, 53901–53901 (2007). [CrossRef]
14. X. Wang, Y. Pang, G. Ku, X. Xie, G. Stoica, and L. V. Wang, “Noninvasive laser-induced photoacoustic tomography for structural and functional in vivo imaging of the brain,” Nature Biotech. 21, 803–806 (2003). [CrossRef]
16. G. J. Diebold and T. Sun, “Properties of photoacoustic waves in one, two, and three dimensions,” Acustica 80, 339–351 (1994).
17. A. A. Oraevsky and A. A. Karabutov, “Ultimate sensitivity of time-resolved optoacoustic detection,” in Biomedical Optoacoustics (SPIE, San Jose, CA, USA, 2000), 3916, pp. 228–239.
18. P. Liu, “Image reconstruction from photoacoustic pressure signals,” in Laser-Tissue Interaction VII(SPIE, San Jose, CA, USA, 1996), 2681, pp. 285–296.
19. C. G. A. Hoelen and F. F. M. de Mul “Image Reconstruction for Photoacoustic Scanning of Tissue Structures,” Appl. Opt. 39, 5872–5883 (2000). [CrossRef]
20. K. P. Kostli, D. Frauchiger, J. J. Niederhauser, G. Paltauf, H. P. Weber, and M. Frenz, “Optoacoustic imaging using a three-dimensional reconstruction algorithm,” IEEE J. Sel. Top. Quantum Electron. 7, 918–923 (2001). [CrossRef]
22. X. Minghua and L. V. Wang, “Universal back-projection algorithm for photoacoustic computed tomography,” Phys. Rev. E (Statistical, Nonlinear, and Soft Matter Physics) 71, 16706–16701 (2005). [CrossRef]
23. C. G. A. Hoelen, F. F. M. de Mul, R. Pongers, and A. Dekker, “Three-dimensional photoacoustic imaging of blood vessels in tissue,” Opt. Lett. 23, 648–650 (1998). [CrossRef]
24. J.-T. Oh, M.-L. Li, H. F. Zhang, K. Maslov, G. Stoica, and L. V. Wang, “Three-dimensional imaging of skin melanoma in vivo by dual-wavelength photoacoustic microscopy,” J. Biomed. Opt. 11, 34032 (2006). [CrossRef] [PubMed]
25. S. Vaithilingam, I. O. Wygant, P. S. Kuo, X. Zhuang, O. Oralkan, P. D. Olcott, and B. T. Khuri-Yakub, “Capacitive micromachined ultrasonic transducers (CMUTs) for photoacoustic imaging,” in Photons Plus Ultrasound: Imaging and Sensing 2006: The Seventh Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics (SPIE, San Jose, CA, USA, 2006), 6086, pp. 608603–608611.
26. J. J. Niederhauser, M. Jaeger, and M. Frenz, “Real-time three-dimensional optoacoustic imaging using an acoustic lens system,” Appl. Phys. Lett. 85, 846–848 (2004). [CrossRef]
27. R. A. Kruger, W. L. Kiser, D. R. Reinecke, G. A. Kruger, and K. D. Miller, “Thermoacoustic molecular imaging of small animals,” Molec. Imag. 2, 113–123 (2003). [CrossRef]
28. P. Guo, J. Gamelin, S. Yan, A. Aguirre, and Q. Zhu, “Co-registered 3-D ultrasound and photoacoustic imaging using a 1.75D 1280-channel ultrasound system,” in Photons Plus Ultrasound: Imaging and Sensing 2007: The Eighth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics(SPIE, San Jose, CA, USA, 2007), 6437, pp. 643713–643711.
29. G. Paltauf, R. Nuster, P. Burgholzer, and M. Haltmeier, “Three-dimensional photoacoustic tomography using acoustic line detectors,” in Photons Plus Ultrasound: Imaging and Sensing 2007: The Eighth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics(SPIE, San Jose, CA, USA, 2007), 6437, pp. 64370N-64310.
30. R. J. Zemp, R. Bitton, M.-L. Li, K. K. Shung, G. Stoica, and L. V. Wang, “Photoacoustic imaging of the microvasculature with a high-frequency ultrasound array transducer,” J. Biomed. Opt. 12, 010501 (2007). [CrossRef] [PubMed]
31. L. Song, K. Maslov, R. Bitton, K. K. Shung, and L. V. Wang, “Fast 3-D dark-field reflection-mode photoacoustic microscopy in vivo with a 30-MHz ultrasound linear array,” J. Biomed. Opt. 13, 054028–054025 (2008). [CrossRef] [PubMed]
32. R. J. Zemp, L. Song, R. Bitton, K. K. Shung, and L. V. Wang, “Realtime photoacoustic microscopy of murine cardiovascular and respiratory dynamics in vivo,” in Photons Plus Ultrasound: Imaging and Sensing 2008: The Ninth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics(SPIE, San Jose, CA, USA, 2008), 6856, pp. 68560G–68568.
33. J. J. Niederhauser, M. Jaeger, R. Lemor, P. Weber, and M. Frenz, “Combined ultrasound and optoacoustic system for real-time high-contrast vascular imaging in vivo,” IEEE Trans. Med. Imaging , 24, 436–440 (2005). [CrossRef] [PubMed]
34. J. Dean, V. Gornstein, M. Burcher, and L. Jankovic, “Real-time photoacoustic data acquisition with Philips iU22 ultrasound scanner,” in Photons Plus Ultrasound: Imaging and Sensing 2008: The Ninth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics (SPIE, San Jose, CA, USA, 2008), 6856, pp. 685622–685611.
35. C.-K. Liao, S.-W. Huang, C.-W. Wei, and P.-C. Li, “Nanorod-based flow estimation using a high-frame-rate photoacoustic imaging system,” J. Biomed. Opt. 12, 064006–064009 (2007). [CrossRef]
36. S. Manohar, A. Kharine, J. C. G. van Hespen, W. Steenbergen, and T. G. van Leeuwen, “The twente photoacoustic mammoscope: system overview and performance,” Phys. Med. Biol. 50, 2543–2557 (2005). [CrossRef] [PubMed]
37. P. Ephrat, L. Keenliside, A. Seabrook, F. S. Prato, and J. J. L. Carson, “Three-dimensional photoacoustic imaging by sparse-array detection and iterative image reconstruction,” J. Biomed. Opt. 13, 054052–054012 (2008). [CrossRef] [PubMed]
38. P. Ephrat and J. J. L. Carson, “Measurement of photoacoustic detector sensitivity distribution by robotic source placement,” in Photons Plus Ultrasound: Imaging and Sensing 2008: The Ninth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics (SPIE, San Jose, CA, USA, 2008), 6856, pp. 685610–685610.
39. M. Roumeliotis, P. Ephrat, and J. J. L. Carson, “Development of an omni-directional photoacoustic source for the characterization of a hemispherical sparse detector array,” in Photons Plus Ultrasound: Imaging and Sensing 2009: The Tenth Conference on Biomedical Thermoacoustics, Optoacoustics and Acousto-optics (SPIE, San Jose, CA, USA, 2009), 7177, pp. 7177–7188.
40. B. C. Wilson, M. S. Patterson, and D. M. Burns, “Effect of photosensitizer concentration in tissue on the penetration depth of photoactivating light,” Lasers Med. Sci. 1, 235–244 (1986). [CrossRef]
41. R. A. Robb and D. P. Hanson, “The ANALYZE software system for visualization and analysis in surgery simulation,” in Computer integrated surgery, T. R. Lavalle S, G Burdea, and R Mosges, ed. (MIT Press, Cambridge, MA, 1995).
43. M. A. Anastasio and J. Zhang, “Image reconstruction in photoacoustic tomography with truncated cylindrical measurement apertures,” in Photons Plus Ultrasound: Imaging and Sensing 2006: The Seventh Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics (SPIE, San Jose, CA, USA, 2006), 6086, pp. 608610–608617.