3102

OPTICS LETTERS / Vol. 39, No. 11 / June 1, 2014

Molecular specificity in photoacoustic microscopy by time-resolved transient absorption Ryan L. Shelton, Scott P. Mattison, and Brian E. Applegate* Department of Biomedical Engineering, 5045 Emerging Technologies Building, Texas A&M University, College Station, Texas 77843, USA *Corresponding author: [email protected] Received January 22, 2014; revised April 21, 2014; accepted April 21, 2014; posted April 22, 2014 (Doc. ID 205063); published May 19, 2014 We have recently harnessed transient absorption, a resonant two-photon process, for ultrahigh resolution photoacoustic microscopy, achieving nearly an order of magnitude improvement in axial resolution. The axial resolution is optically constrained due to the two-photon process unlike traditional photoacoustic microscopy where the axial resolution is inversely proportional to the frequency bandwidth of the detector. As a resonant process, the arrival time of the two photons need not be instantaneous. Systematically recording the signal as a function of the delay between two pulses will result in the measurement of an exponential decay whose time constant is related to the molecular dynamics. This time constant, analogous to the fluorescence lifetime, but encompassing nonradiative decay as well, can be used to differentiate between molecular systems with overlapping absorption spectra. This is frequently the situation for closely related yet distinct molecules such as redox pairs. In order to enable the measure of the exponential decay, we have reconfigured our transient absorption ultrasonic microscopy (TAUM) system to incorporate two laser sources with precisely controlled pulse trains. The system was tested by measuring Rhodamine 6G, an efficient laser dye where the molecular dynamics are dominated by the fluorescence pathway. As expected, the measured exponential time constant or ground state recovery time, 3.3  0.7 ns, was similar to the well-known fluorescence lifetime, 4.11  0.05 ns. Oxy- and deoxy-hemoglobin are the quintessential pair whose relative concentration is related to the local blood oxygen saturation. We have measured the ground state recovery times of these two species in fully oxygenated and deoxygenated bovine whole blood to be 3.7  0.8 ns and 7.9  1.0 ns, respectively. Hence, even very closely related pairs of molecules may be differentiated with this technique. © 2014 Optical Society of America OCIS codes: (110.5120) Photoacoustic imaging; (180.4315) Nonlinear microscopy; (170.3650) Lifetime-based sensing; (170.6920) Time-resolved imaging. http://dx.doi.org/10.1364/OL.39.003102

Photoacoustic imaging is a hybrid modality that combines the absorption contrast of optical imaging with the depth penetration of ultrasound imaging. The combination provides many of the benefits of optical molecular imaging but with the advantage of detecting sound waves instead of light waves. Typical frequencies used for ultrasonic imaging are attenuated 1000× less in tissue than visible light, resulting in a significant advantage by converting the incident optical energy to acoustic energy. This has enabled the development of a group of techniques which span a wide range in resolution and imaging depth [1]. Even subcellular resolution suitable for imaging light absorption in single organelles is possible in the transverse dimension. Unfortunately, the axial resolution remains limited to ∼10 μm due to the properties of the ultrasound transducer [2]. We have recently introduced a new molecular imaging technique, transient absorption ultrasonic microscopy (TAUM) [3], which combines photoacoustic microscopy with pump–probe spectroscopy. The marriage of these techniques enables strong absorption contrast imaging with subcellular resolution in all dimensions, equivalent to that of multiphoton and confocal microscopy. Recent work has improved the speed of this technique 1000-fold and demonstrated volumetric imaging of erythrocytes [4]. The reliance on transient absorption as measured by a two-photon pump–probe technique provides TAUM with both optically limited spatial resolution and access to unique molecular properties that cannot be obtained through other imaging techniques. The transient absorption spectrum can be measured by recording the TAUM 0146-9592/14/113102-04$15.00/0

signal while varying either the pump or probe wavelength. The ground state recovery time can be measured by recording the TAUM signal as a function of the interpulse delay (i.e., the time delay between the pump and probe pulses) and then estimating the decay constant from the exponential decay. The ground state recovery time is related to fluorescence lifetime in that the fluorescence lifetime is a measure of the depopulation rate of the excited state, while the ground state recovery time is a measure of the repopulation rate of the ground state following photoexcitation. Both properties could be used to differentiate chemically distinct molecular species. The ground state recovery time is dependent on both the chromophore and its environment. This property has previously been used employing photoacoustic methods to selectively image chromophores in a proof-of-concept study [5] and to directly measure ground state recovery for evaluating the local oxygen concentration via quenching of the methylene blue triplet state [6–8]. TAUM extracts the pump–probe signal in the frequency domain, separating the contributions of the pump and probe pulses by amplitude modulating each beam at different frequencies. This results in a temporal resolution equal to the pulse duration of the laser source. In principle, ground state recovery times could be measured with TAUM even down in the femtosecond pulse regime given the appropriate laser source. This is important for many biomolecules, such as melanin [9], which has a recovery time in the subnanosecond range. The dependence of the TAUM signal on interpulse delay has been derived previously for a two-state molecular system [3], with the final result shown below in Eq. (1) © 2014 Optical Society of America

June 1, 2014 / Vol. 39, No. 11 / OPTICS LETTERS

Δpr; t 

  σλpu F pu −td σN 01 F pr κr; tΣi exp ; hc τi

3103

(1)

where Δpr; t is the change in the spatially and temporally dependent photoacoustic pressure wave in the presence and absence of the pump illumination, σ is the absorption cross section of the molecule of interest, λpu is the wavelength of the pump illumination, F pu and F pr are the fluences of the pump and probe illumination, respectively, h is Planck’s constant, c is the speed of light, N 01 is the population of the ground state before photoexcitation, and κr; t is a simplifying term describing the space and time dependence of the pressure profile. There is rarely a single pathway by which an excited molecule relaxes, so the signal is best represented by a sum of exponential decays, where i represents a unique relaxation path back to the ground state. td is then the interpulse delay and τi is the ground state recovery time for each relaxation path that exists in the molecule under investigation. These paths vary depending on the molecule. Some molecules will have radiant relaxation paths, such as fluorescence and phosphorescence, each of which will have different relaxation times, leading to different exponential decays. All molecules have nonradiant relaxation paths, such as internal conversion and vibrational relaxation, which will contribute still different exponential decays. The measured ground state recovery time of the molecule will be dependent on the branching ratio into each of the possible relaxation pathways. Two important observations can be gleaned from examination of Eq. (1). First, the TAUM signal, or Δpr; t, is proportional to the product of F pu and F pr . This square-dependence on fluence provides the increased sectioning capabilities of the TAUM technique. Second, the interpulse delay in the numerator of the negated exponential term results in maximum TAUM signal at interpulse delays much less than the ground state recovery time and minimum TAUM signal at interpulse delays much greater than the ground state recovery time. In the following experiments, the interpulse delay will be controlled and varied in order to measure the ground state recovery of various molecules. The system used to measure the ground state recovery times is modified from the original TAUM design. In the modified design, two Nd:YVO4 frequency-doubled, 532 nm lasers (Advanced Optical Technologies, Inc.) make up the pump and probe sources. These lasers have a 0.9 ns pulse duration and a maximum repetition rate of 100 kHz. The schematic of the TR-TAUM system can be seen in Fig. 1. The two lasers are controlled externally using triggers sent from a computer through a digital pulse delay generator (P-400, Hiland Technologies). The pulse delay generator receives the master trigger (TTL pulse) from the custom LabVIEW program, and then generates two more TTL pulses, each with a controllable delay from the master pulse. These two delayed TTL pulses are then used to trigger the lasers. In this configuration, the interpulse delay can be easily adjusted through a serial connection to the pulse delay generator. Jitter is not a problem at the time scales we are investigating, as the jitter from the lasers is ∼200 ps and the jitter from the pulse delay generator is ∼2 ps.

Fig. 1. Schematic of the TAUM system modified for ground state recovery measurements. A, Isometric view of the modified TAUM microscope. B, Top-down view of the microscope with components labeled. BS, 50∕50 beam splitter; DFC, dualfrequency optical chopper; G, galvanometer scanning mirror pair; SL, scan lens, TL, tube lens; MH, microscope head.

In order to extract the TAUM signal, the pulses emitted from each laser are chopped at separate frequencies using a dual-frequency optical chopper. This process is described in detail in [3]. After optical chopping, the beams are combined using a beam splitter, then passed through a beam expander to slightly overfill the aperture of a 0.8 NA water immersion objective (CFI Apo 40XW NIR, Nikon, Inc.) and achieve diffraction-limited performance. After excitation, the photoacoustic emission is collected using a 6 MHz transducer (V-305, Olympus NDT) and amplified using two RF in-line amplifiers. Each TAUM pixel is collected and processed in real time using the field-programmable gate array (FPGA) configuration described in [4]. This second-generation TAUM system is capable of pixel rates up to 500 Hz, a 1000-fold improvement over the speed of the first-generation system. The Fourier domain processing and filtering is performed on the FPGA in real time and a custom LabVIEW program synchronizes the delay generator sweep with the TAUM collection and processing. In order to validate the system, we measured the timeresolved TAUM signal from Rhodamine 6G (R6G). R6G has a fluorescence quantum yield of 0.95 [10], which means that 95% of the molecules pumped into the excited state return back to the ground state via fluorescence. The other 5% are responsible for the photoacoustic signal. Since the fluorescence channel is the dominant pathway back to the ground state, the ground state recovery time will be very close to the fluorescence lifetime. The fluorescence temporal decay is well characterized by a single exponential decay with a time constant (fluorescence lifetime) of 4.11  0.05 ns [11]. However, the TAUM temporal decay should be at least bi-exponential with 5% of the molecules decaying nonradiatively with a time constant likely no more than a few picoseconds and 95% of the molecules decaying with a time constant equal to the fluorescence lifetime. For this reason, the expected ground state recovery time should be close to the fluorescence lifetime but not exactly the same. A 500 μM concentration of R6G was placed in a 100 μm ID capillary tube, and the ground state recovery time was measured using the TAUM system. The average 1∕e recovery time over 44 datasets was measured to be 3.3  0.7 ns (μ  σ), which agrees well with the known

3104

OPTICS LETTERS / Vol. 39, No. 11 / June 1, 2014

fluorescence lifetime, taking into consideration the experimental error and previously stated 200 ps laser jitter. Figure 2A shows a plot of the ground state recovery time of R6G as measured by TAUM (dots), along with a single exponential decay (line) with a decay constant of 1∕3.5 ns−1 . The plotted data is an average of 10 measured decay curves. The data has been normalized and the mean of the noise floor has been subtracted. It is important to note that in addition to the expected noise floor in these plots, a small background signal existed that was due to sidelobes of the single photon signals bleeding over into the TAUM (multiphoton) frequency bins. This background is consistent across all interpulse delay times and thus easy to remove.

Fig. 2. Ground state recovery time plots measured using TAUM. A, Rhodamine 6G recovery time data with a single exponential decay plotted with a decay constant of 1∕3.5 ns−1 . B, Oxygenated and deoxygenated blood recovery time data with single exponential decays plotted with decay constants of 1∕3.5 and 1∕6.5 ns−1 , respectively. C, Histogram showing the distribution of measured ground state recovery times in R6G (N  44), oxygenated blood (N  50), and deoxygenated blood (N  44). The histogram shows a clear distinction between oxygenated and deoxygenated states of whole blood.

After verifying the ability of TAUM to measure the ground state recovery time of a well-documented chromophore, the system was used to investigate whole blood. Clotted human whole blood has been shown in previous 532 nm degenerate pump–probe studies to have a ground state recovery time of ∼9 ns [12]. However, these previous studies did not investigate the effects of oxygenation on the recovery time of blood. Whole bovine blood samples with an anticoagulant were prepared in a similar manner to the R6G samples described above and the samples were measured immediately after preparation. Sodium dithionite was used to deoxygenate the blood as described in [13]. Figure 2B shows representative recovery time plots of oxygenated and deoxygenated whole blood as measured by TAUM, along with single exponential decays plotted for each dataset. The data was averaged 10 times, normalized, and the mean of the noise floor was subtracted. The average 1∕e lifetime over 50 fully oxygenated whole blood samples was measured to be 3.7  0.8 ns (μ  σ). The average 1∕e lifetime over 44 fully deoxygenated whole blood samples was measured to be 7.9  1.0 ns (μ  σ). The data from oxygenated and deoxygenated blood shows a distinct difference in their ground state recovery times, further illustrated in the histogram in Fig. 2C. The pulse energy used in each of these studies was 10 nJ for each pulse, resulting in a total of 20 nJ incident on the sample during each measurement. The potential for two-dimensional imaging of ground state recovery time was investigated by imaging a capillary tube while varying the interpulse delay. The 100 μm ID acrylic tube was filled with 500 μM R6G. The tube was imaged repeatedly using the modified TAUM system and the interpulse delay was changed between each image. Figure 3 shows a film strip of successive TAUM images depicting en face scans of the tube. Beginning at the temporal overlap of the two pulses, (interpulse delay  0 seconds, the first image in the film strip) at which the TAUM signal should be strongest, the interpulse delay was then changed by 2 ns between each image, resulting in a total change in interpulse delay of 14 ns. Beneath the en face film strip is a plot of the image intensity through the center of each panel. The signal strength of the R6G solution drops below the 1∕e intensity point between panels 2 and 3, which corresponds to a ground state recovery time between 2 and 4 ns, which is consistent with the measurement of 3.3 ns in the previous experiment. In future studies, chromophores that exhibit significant

Fig. 3. Film strip panel of TAUM images of a 100 μm capillary tube filled with R6G and imaged at varying interpulse delay values. The drop in signal crosses the 1∕e value between panels 2 and 3, which is consistent with the measured recovery time for R6G reported above. Scalebar  40 μm.

June 1, 2014 / Vol. 39, No. 11 / OPTICS LETTERS

differences in ground state recovery time could be differentiated with as few as two time points, which only requires two consecutive 2D TAUM images. Of course, as the ground state recovery time of the chromophores of interest become more similar, more time points will be required in order to differentiate them. It will be important to improve the speed of TAUM in future experiments designed to fully evaluate the capability of this technique to compete with the exogenous contrast imaging of other high-resolution microscopy techniques. Lowering the number of pulses used for each measurement can increase imaging speed. In the current configuration, 128 pulses were used per measurement, however, 64 pulses have also been used with success. Fewer pulses results in lower signal-to-noise ratio, but higher speed. A large speed advantage may be gained by reducing the number of time points taken to 2 or 3, which should be sufficient to differentiate many biological chromophores. The TAUM system could be simplified in this case by building in optical delay paths for each time point using beam splitters, rather than relying on additional circuitry to provide these delays. This modification would also allow the system to function with a single laser, rather than require two lasers. Each separate delay path could be modulated at a different frequency, effectively multiplexing the signals from multiple interpulse delays. One potentially useful application for this technology is the differentiation of oxygenated and deoxygenated blood. Using the difference in ground state recovery time between oxidized and reduced forms of hemoglobin, oxygen saturation could be measured using a single wavelength. For this study, a statistical difference was found in the ground state recovery times of oxygenated and deoxygenated blood; however, differentiation of the oxidized and reduced forms and measurement of oxygen saturation will be the focus of future work. Since the characteristic ground state recovery times of oxygenated and deoxygenated blood is known for this excitation wavelength, a simplified configuration designed for the measurement of oxygen saturation could be built using only a single laser source with multiple optical delay paths corresponding to, for instance, 0.5, 3, and 6 ns in order to quickly differentiate these chromophores in a single image. Ultimately, the goal is to refine the technique to the point that a femtosecond Ti:sapphire laser may be used. This would serve to increase the imaging speed dramatically due to the highly increased repetition rate of mode-locked lasers. Using a femtosecond laser would

3105

also provide much better temporal resolution for the transient absorption process, allowing greater control over the interpulse delay for measurements of ground state recovery time in molecules that would be inaccessible to a nanosecond excitation source. Additional chromophores that would be of interest for future studies include melanin, cytochrome C, myoglobin, DNA, and RNA. In conclusion, we have developed and demonstrated a modified TAUM setup capable of measuring ground state recovery times in the nanosecond range. The system was verified by measuring R6G and was then used to measure ground state recovery times for oxygenated and deoxygenated blood. It was found that the recovery times for blood were highly dependent on oxygen concentration, with more than a factor-of-two difference in the characteristic ground state recovery times of each. The results of this initial work and unique capability of TAUM to take these measurements at submicrometer spatial resolution suggest the possibility of imaging the oxygen gradient that exists within a single erythrocyte in future TAUM research. We would like to thank Dr. Javier Jo for allowing us to borrow a laser. We also gratefully acknowledge financial support for this work via a grant from the National Science Foundation (CAREER, CBET-1055359). References 1. L. V. Wang and S. Hu, Science 335, 1458 (2012). 2. K. Maslov, H. F. Zhang, S. Hu, and L. V. Wang, Opt. Lett. 33, 929 (2008). 3. R. L. Shelton and B. E. Applegate, Biomed. Opt. Express 1, 676 (2010). 4. R. L. Shelton, S. P. Mattison, and B. E. Applegate, “Volumetric imaging of erythrocytes using label-free multiphoton photoacoustic microscopy,” J. Biophoton., doi:10.1002/ jbio.201300059 (2013). 5. S. W. Huang, J. F. Eary, C. X. Jia, L. Y. Huang, S. Ashkenazi, and M. O’Donnell, Opt. Lett. 34, 2393 (2009). 6. S. Ashkenazi, S. W. Huang, T. Horvath, Y. E. L. Koo, and R. Kopelman, J. Biomed. Opt. 13, 034023 (2008). 7. S. Ashkenazi, J. Biomed. Opt. 15, 040501 (2010). 8. Q. Shao, E. Morgounova, C. Jiang, J. Choi, J. Bischof, and S. Ashkenazi, J. Biomed. Opt. 18, 076019 (2013). 9. D. Fu, T. Ye, T. E. Matthews, G. Yurtsever, and W. S. Warren, J. Biomed. Opt. 12, 054004 (2007). 10. R. F. Kubin and A. N. Fletcher, J. Lumin. 27, 455 (1982). 11. Q. S. Hanley, V. Subramaniam, D. J. Arndt-Jovin, and T. M. Jovin, Cytometry 43, 248 (2001). 12. B. E. Applegate and J. A. Izatt, Opt. Express 14, 9142 (2006). 13. K. Dalziel and J. R. P. O’brien, Biochem. J. 78, 236 (1961).

Molecular specificity in photoacoustic microscopy by time-resolved transient absorption.

We have recently harnessed transient absorption, a resonant two-photon process, for ultrahigh resolution photoacoustic microscopy, achieving nearly an...
339KB Sizes 2 Downloads 3 Views