Open Access
Add to BibSonomyAbstract
A strategy is presented to enable optical-sectioning microscopy with improved contrast and imaging depth using low-power (0.5 - 1 mW) diode laser illumination. This technology combines the inherent strengths of focal-modulation microscopy and dual-axis confocal (DAC) microscopy for rejecting out-of-focus and multiply scattered background light in tissues. The DAC architecture is unique in that it utilizes an intersecting pair of illumination and collection beams to improve the spatial-filtering and optical-sectioning performance of confocal microscopy while focal modulation selectively ‘labels’ in-focus signals via amplitude modulation. Simulations indicate that modulating the spatial alignment of dual-axis beams at a frequency f generates signals from the focal volume of the microscope that are modulated at 2f with minimal modulation of background signals, thus providing nearly an order-of-magnitude improvement in optical-sectioning contrast compared to DAC microscopy alone. Experiments show that 2f lock-in detection enhances contrast and imaging depth within scattering phantoms and fresh tissues.
© 2014 Optical Society of America
1. Introduction
To enable noninvasive real-time pathology (optical biopsy) of fresh tissues [1–3], improving the depth of optical sectioning is of high priority and would provide a broad clinical impact. For example, 90% of cancers originate in epithelial tissues and metastasize by invading into the underlying stroma [4]. The ability to visualize these tissues clearly and to investigate these processes of cancer invasion would benefit from a technology capable of imaging intact tissues to a depth of 500 μm or more. The need for high spatial resolution (super-resolution microscopy [5–8]) in clinical diagnoses is less immediate since the “gold standard” of histopathology relies on the observation of tissue sections under low magnification (e.g., 2x - 10x is typical for Mohs surgical pathology) [9]. Rather, for applications such as early cancer detection, guided biopsy (e.g., in oral cancer), or surgical guidance (e.g., in dermatology), the ability to visualize large fields of view in three dimensions with cellular resolution is valuable, especially if there is sufficient depth to enable assessment of the local invasion of cancer.
It is a generally accepted constraint in the field of biomedical optics that a tradeoff between spatial resolution and imaging depth [Fig. 1] governs modalities including selective plane illumination microscopy (SPIM) [10], two-photon microscopy (2P) [11], optical coherence tomography (OCT) [12], single- and dual-axis confocal microscopy (SAC, DAC) [13–16], photoacoustic tomography (PAT) [17] and diffuse optical tomography (DOT) [18]. We are interested in molecular imaging with exogenous agents or fluorescent proteins. While photoacoustic methods have attractive properties in terms of resolution and penetration depth, their reliance on absorption as a contrast mechanism limits their ability to take advantage of the large arsenal of fluorescent contrast agents that have been developed in recent decades, especially for multiplexed imaging. OCT provides deep optical sectioning with coherent reflected light but cannot be easily adapted to detect fluorescence [19]. Regarding DAC microscopy, our previous Monte-Carlo simulations and experiments with DAC microscopy have demonstrated the ability to image with sub-cellular resolution to a depth of about 3 - 4 mean free paths (MFPs) in tissue (approximately 250 - 400 μm deep in epithelial tissues) [15, 16, 20] where a mean free path (MFP = 1/μs) is generally 10 times shorter than the transport mean free path (TMFP = 1/μs’) in tissues (where μs’ = μs (1 – g) and g ~0.9). Our goal is now to push DAC microscopy to image even deeper, approaching an imaging depth of nearly one TMFP, or ~500 μm in typical epithelial tissues. To the best of our knowledge, the modulated-alignment dual-axis (MAD) confocal microscopy approach has never been explored, and is potentially an inexpensive and scalable method for achieving tissue microscopy with sufficient depth to interrogate epithelial tissues over the full thickness of their basement membranes.
Fig. 1 Resolution and depth tradeoffs in common optical imaging modalities. In general, deeper imaging requires sacrificing spatial resolution. Approximate lower and upper bounds on these parameters are mapped for single-axis confocal microscopy (SAC), selective plane illumination microscopy (SPIM), optical coherence tomography (OCT), two-photon microscopy (2P), dual-axis confocal microscopy (DAC), photoacoustic tomography (PAT) and diffuse optical tomography (DOT). Note that modulated-alignment dual-axis (MAD) confocal microscopy seeks to extend DAC’s penetration depth at minimal cost to resolution. Depth of imaging is expressed in terms of mean free path (MFP = 1/μs) and transport mean free path (TMFP = 1/μs’), where μs’ = μs(1 – g). Since the anisotropy factor of tissue, g, is typically ~0.9, the transport mean free path is typically one order of magnitude larger than the mean free path (TMFP ~10MFP).
The MAD microscopy concept was inspired by focal modulation microscopy [21] and a “spatial overlap modulation” technique for nonlinear optical processes such as 2-color 2-photon microscopy, sum-frequency generation, etc. [22]. In the context of DAC microscopy, focal modulation microscopy is not ideal in its original form [21] since the Rayleigh range of the low-NA beams utilized in DAC microscopy is large and would therefore result in a modulated point spread function that extends over a large area, rather than being confined to the focal volume defined by the intersection of the DAC illumination and collection beams. Alternatively, the spatial overlap modulation technique reported by Isobe et al. [22, 23] relies on a slight spatial modulation in the lateral direction between two coaxial beams, which generates a strong modulation in the signal generated at the focus of a nonlinear microscope but negligible modulation of the out-of-focus background signals.
In this paper, we describe an architecture where the DAC microscope provides an elegant and optimized platform for spatial modulation with low-power linear excitation since it utilizes illumination and collection beams that only overlap at a single point (focal volume) in space. Lateral misalignment (y-direction) between the dual-axis beams [Figs. 2(a) and 2(b)] causes a decay in signal [Fig. 2(c)] as demonstrated with diffraction theory. This micron-scale modulation in the alignment of the illumination and collection beams causes the in-focus signal to modulate significantly, whereas the background – due to out-of-focus and multiply scattered photons originating over a large tissue volume – is negligibly modulated in intensity by these relatively miniscule spatial variations. Thus, by modulating the alignment of the two beams at a frequency f, the microscope signal is modulated at a frequency of 2f, which can be separated from non-modulated background signal using lock-in detection.
Fig. 2 Principles of modulated-alignment dual-axis (MAD) confocal microscopy. By periodically sweeping the illumination beam into (a) and out of alignment (b) with the collection beam at a given frequency f, the signal generated at the focal volume experiences an amplitude modulation at 2f (c) and can therefore be distinguished from background signal that remains largely unmodulated.
2. Methods
2.1 System architecture
The microscope system presented here is a modified version of a tabletop DAC microscope described previously [15]. The device’s optical circuit [Fig. 3] consists of two parallel collimated beams (illumination and collection beams) that focus to and intersect at a single point (the focal volume). The focusing half angle (α) of each beam and the intersection half angle (θ) between the beams are 0.11 radians (1∕e2 intensity) and 30.0 degrees, respectively. Illumination is provided by a fiber-pigtailed diode laser at 658 nm (57ICS054/SP/HS, Melles Griot). The illumination light is collimated (L1, EFL = 4.51 mm; 1∕e2 diameter of 1.6mm; NT64-806, Edmund Optics, Barrington, NJ) and passes through the AOD’s aperture before being re-focused (L1, EFL = 4.51 mm; 1∕e2 diameter of 1.6 mm) prior to injection into the illumination arm of the system. The illumination beam is steered laterally (y-direction) by as much as +/− 0.5 mrad using the acousto-optic deflector (1205C-2, Isomet Corporation, Springfield, VA). The first-order diffracted light exiting the AOD is isolated using a 50-μm slit (S50R, ThorLabs, Newton, NJ), ensuring that all other diffraction orders are blocked. Using a function generator (DS345, Stanford Research Systems, Sunnyvale, CA) to control the AOD’s analog driver (630C-80, Isomet Corporation, Springfield, VA), the maximum rate of alignment modulation was set to 500 kHz since nonlinearities in the AOD output were observed at rates above 1 MHz. A 1D galvanometric scanning mirror (6210H Series, Cambridge Technology, Bedford, MA) deflects these beams toward the specimen, which rests on a solid immersion lens (SIL) that is mounted on a linear piezoelectric actuator (P-601.4SL, Physik Instrumente LP, Auburn, MA). The galvanometric mirror is used to scan the dual-axis beams together in the y-direction at a rate of up to 160 Hz (fast axis) while the linear piezoelectric actuator is used to scan the sample in the vertical (z) direction at a rate of up to 0.4 Hz (the imaging frame rate for vertical sections). In addition, a motorized micrometer (LTA-HL, Newport Corporation, Irvine, CA) may be used to scan the sample in the x-direction for the acquisition of three-dimensional data sets.
Fig. 3 MAD architecture and optical circuit. (a) The MAD system is built by modifying a typical dual-axis confocal (DAC) tabletop microscope. An acousto-optical deflector (AOD), which is optimized to produce a 1st-order diffracted beam, is inserted into the optical path upstream of the illumination arm. The 1st diffracted order serves as the illumination beam in the MAD system. A slit (S1) oriented perpendicularly to the dispersion direction is used to isolate the 1st-order beam and to block all other diffracted orders. Pixel intensities correspond to the magnitude of the 2f signal. (b) A photograph of the tabletop MAD setup. (c) Motion of the illumination beam is minimized in relation to the motion of the focus (right side of ray-trace diagram) by ensuring that the pivot point about which the illumination beam rotates (for alignment modulation) is located near or at the focusing objective. This ensures that the motion of the illumination beam at the focus is larger than the motion of the beam prior to the focus.
Fiber-coupled signals are detected by a photomultiplier tube (H7422-40, Hamamatsu, Edison, NJ) with a transimpedance amplifier (DHPCA-100, FEMTO, Berlin, Germany), which generates a voltage output that is fed into a spectrum analyzer operating as a lock-in amplifier (FSEA20, Rohde&Schwarz, Munich, Germany) with a resolution bandwidth and video bandwidth both set at 100 kHz and a center frequency of 1 MHz (zero span). The spectrum analyzer provides an analog voltage output that scales logarithmically with the magnitude of the 2f signal, which is collected by an analog signal connector block (BNC-2110, National Instruments, Austin, TX) and recorded to a PC using a digitizer (PCI-6115, National Instruments Corporation, Austin, TX). A custom frame grabber written in LabVIEW (National Instruments, Austin, TX) is synchronized with the fast- and slow-axis scanners of the microscope system and provides 2D vertical images.
A subtlety in the MAD design is ensuring that the motion of the illumination beam away from the focus is minimized in relation to the motion of the beam very close to the focus. This is necessary to minimize the modulation of the out-of-focus background signal compared to the modulation of the signal at the focal volume. To achieve this, the illumination optics are arranged such that the illumination beam’s center of rotation (pivot point for alignment modulation) is located near the focusing objective [Fig. 3(c), L2*; Edmund Optics NT49-660, Barrington, NJ]. This guarantees that the lateral motion of the illumination beam at the focus is larger than the motion of the beam prior to the focus.
2.2 System operating modes
Regarding comparative experiments in scattering media to investigate the difference in signal-to-background ratio between MAD and DAC in the axial and transverse directions, it was important to ensure that the spectrum analyzer could be used for measurements in both MAD and DAC modes even though DAC is not classically considered to have any predictable signal modulation. Using the same detection system (PMT and spectrum analyzer) is useful in the direct comparison of MAD to DAC performance since it avoids the need to convert between different detection systems that each possess different gain settings, noise characteristics and output formats.
To record DAC measurements using the lock-in amplifier, a sinusoidal voltage signal (DS345, Stanford Research Systems, Sunnyvale, CA) at a frequency of 1 MHz (equivalent to the MAD 2f modulation frequency) was applied to the intensity modulation input of the illumination laser without any input to the AOD (i.e., no spatial modulation between the illumination and detection paths). The sine wave was DC-biased (E3610A, Agilent, Santa Clara, CA) to the empirically measured voltage corresponding to the 50% power setting for the laser using a bias-tee connector (ZX85-12G-S + , MiniCircuits, Brooklyn, NY).
Results from axial and transverse scans obtained in AM-DAC (amplitude-modulated DAC) mode were compared to results obtained using standard DAC mode (i.e., no laser intensity modulation and no spatial modulation between the illumination and detection paths), where a basic optical power meter (S130C, ThorLabs, Newton, NJ) was used to record the optical throughput of the DAC microscope. Since there was no appreciable difference observed in SBR values between AM-DAC and standard DAC, we claim that the amplitude-modulated DAC (AM-DAC) system provides equivalent optical-sectioning performance (resolution and contrast) when compared to a conventional (unmodulated) DAC microscope (data not shown). Therefore, MAD and AM-DAC data were collected using the same spectrum analyzer settings (resolution bandwidth and video bandwidth were both set at 30 Hz for axial/transverse scans or 100 kHz for imaging; the center frequency was 1 MHz (zero span) in all experiments).
2.3 Monte-Carlo simulations
FRED software (Photon Engineering, Tucson, AZ) was used for Monte-Carlo scattering simulations. To model the DAC and MAD systems with single-mode fibers as illumination and collection pinholes, two Gaussian beams with a 1∕e2 focusing half-angle (α) of 0.11 radians were aligned in such a way that their intersection half-angle (θ) was 30 degrees. The illumination beam was generated by a Gaussian point source at 658 nm, which was imaged without magnification into homogeneous scattering media (μs = 30 mm−1) through a pair of matched aspheric lenses with a 25-mm focal length (NT49-660, Edmund Optics, Barrington, Canada). The collection path was identical to the illumination path and symmetric with respect to the z-axis. The pinhole size at the detector was set to 3 μm, which corresponds to the mode-field diameter of a singlemode collection fiber at ~660 nm. In order to measure the signal-to-background ratio for the DAC system configuration, a mirror was placed at the focal plane behind a variable thickness of homogeneous scattering medium with all refractive indices set to unity to avoid aberrations and index-matching issues. Signal-to-background ratios were computed by recording the peak signal obtained when the mirror was located precisely at the focus of the microscope as well as the background signal obtained when removing the mirror from the simulation. The imaging depth in these simulations was defined as the “perpendicular optical length,” L = 2μsd (where μs is the scattering coefficient and d is the penetration depth or distance between the tissue surface and focal volume), which is a dimensionless value describing the total number of mean free paths that ballistic (nonscattered) photons traverse in a perpendicular roundtrip path between the tissue surface and the mirror. Signal-to-background ratios were measured in the MAD system configuration by computing the ratio between the difference in peak signal seen (i.e., with a mirror at the focus) in perfect alignment vs. maximal alignment offset and the difference in background signal (i.e., no mirror) seen in perfect alignment vs. maximal alignment offset. Expressed in equation form, this corresponds to SBRMAD = (mirrorsignalaligned – mirrorsignalmisaligned) / (nomirroraligned – nomirrormisaligned). Monte-Carlo simulations showed an improvement of ~9 dB in MAD’s ability to enhance signal-to-background ratio over the standard DAC setup [Fig. 4(b)].
Fig. 4 Optimizing modulation depth. (a) In response to a mirror placed at the focal plane, the optical throughput of the microscope system decays as the alignment offset between the illumination and detection paths increases. (b) When sinusoidally modulating the alignment offset at a modulation depth of amplitude Δy/ω0, the maximum 2f signal is generated at Δy/ω0 ~1.5 - 1.8. Example time traces (c) and their frequency content (d) at three distinct modulation depths demonstrate the dependency of the 2f signal on modulation depth.
2.4 Phantom preparation
Custom gel-based phantoms were prepared using a modified version of a commonly used recipe [24]. Specifically, a stock solution was prepared using 47.5% deionized water, 47.5% PBS, 3.0% Intralipid (20%) and 2% w/v agarose powder. The powder, PBS and water were heated to 90 degrees C and allowed to cool to 45 degrees C before adding Intralipid and stirring. Maintaining the temperature at 45 degrees C, the solution was poured into a mould to cast phantoms corresponding in size to the specimen stage of the microscope (~4.5 mm diameter). Fluorescent microspheres (λex/em = 660/690 nm) measuring 1.01 μm in diameter (FS06F/9874, Bangs Laboratories, Fishers, IN) and fluorescent dye (NF647-5-01, Cytodiagnostics, Burlington, Canada) were added to serve as bright targets of interest and a competing source of background signal, respectively. After quickly cooling to 4 degrees C, phantoms were extracted from the mould and imaged immediately using 658-nm excitation (0.5 – 1 mW incident power) and a 665-690-nm bandpass filter (ER690/50m, Chroma, Bellows Falls, VT).
2.5 Tissue samples
Fresh tissues from euthanized mice were stained with fluorescent dye (NF647-5-01, Cytodiagnostics, Burlington, Canada) for up to 15 min and rinsed in PBS to remove excess fluorophore. Tissues were kept hydrated with PBS, stored on ice and were imaged within 3 hours of euthanization. As in phantom imaging, three-dimensional DAC and MAD microscopy was performed using 658-nm excitation (0.5 - 1 mW incident power) and a 665-690-nm bandpass filter.
3. Results
3.1 Simulations
In the MAD approach, we take advantage of the fact that the DAC microscope signal is highly sensitive to the precise spatial alignment between the illumination and collection paths while the background signal is much less sensitive to this alignment. Using an acousto-optic deflector (AOD), these paths are driven cyclically from perfect alignment [Fig. 2(a)] towards a defined maximum offset [Fig. 2(b)], which causes strong amplitude modulation at the focal volume [Fig. 2(c), red trace].
System behavior was simulated using both diffraction theory as well as Monte-Carlo tissue-scattering simulations. We assumed that the beam waists were identical for the illumination (λ = 658 nm) and collection paths (i.e., a symmetric system with identical illumination and collections NAs) and performed a diffraction-theory analysis in MATLAB (Mathworks) assuming an experimentally measured Gaussian beam waist (ω0) of 1.9 μm and a Rayleigh length (zR) of 17.2 μm. When placing a mirror at the focus of the microscope either in simulations or experiments, the optical throughput is maximized as the alignment offset (Δy/ω0) approaches zero (i.e., perfect alignment) and decreases monotonically as the alignment offset increases. Diffraction theory simulations were found to agree to within 10% of experimental values [Fig. 4(a)].
We used the relationship between optical throughput and alignment offset [Fig. 4(a)] to simulate the 2f signal generated by a modulated alignment offset. Specifically, the 2f power in the microscope signal, in response to a 1f sinusoidal modulation of the alignment offset, was simulated and measured for various modulation depths (Δy/ω0) from 0 to 3. Both theory and experiments show that the 2f signal from a mirror is maximized at a modulation depth between 1.5 and 1.8 [Fig. 4(b)]. We reason that as modulation depth is increased, there comes a point beyond which the amplitude of the modulated signal remains nearly constant while the duty cycle (the fraction of the time in which there is appreciable signal) progressively shortens. These larger modulation depths reduce the amount of power at 2f by redistributing power into higher harmonics [Figs. 4(c) and 4(d)].
In order to assess MAD system performance in the presence of scattering, Monte-Carlo simulations were performed in FRED (Photon Engineering, Tucson, AZ), which employs the Henyey–Greenstein approximation of Mie scattering theory [25] without accounting for polarization, diffraction, absorption, or refractive beam steering and lensing introduced by heterogeneous structures. Despite these model simplifications, Monte-Carlo simulations provide a realistic approximation of optical-sectioning performance in tissues that we have previously validated experimentally with Intralipid phantoms [15, 26]. Our homogeneous tissue model [Fig. 5(a)] has a scattering coefficient (μs) of 30 mm−1 and an anisotropy factor (g) of 0.9 and is designed to estimate tissue-imaging performance. Monte-Carlo simulations show an improvement of ~9 dB in MAD’s ability to enhance signal-to-background ratio (SBR) over the standard DAC setup [Fig. 5(b)]. Furthermore, the optimal modulation depth for maximizing SBR in tissues is relatively insensitive to depth (d) and is consistently in the range of Δy/ω0 ~1.5 - 1.8 [Fig. 5(c)]. Details regarding the computation of signal-to-background ratios in the MAD and DAC configurations are described in the Methods section.
Fig. 5 Monte-Carlo simulations. (a) Typical DAC microscope system geometry where θ is the crossing half-angle between the beams and α is the focusing half-angle of each beam. (b) A ~9-dB improvement in optical-sectioning contrast is achieved at all depths (perpendicular optical lengths. Lp, ranging from 2 to 10). (c) Monte-Carlo simulations show that the optimal modulation depth for maximizing contrast (signal to background ratio, SBR) is consistently Δy/ω0 ~1.5 - 1.8 over a range of imaging depths (or perpendicular optical lengths, Lp).
3.2 System performance
The MAD system’s maximal signal-to-background ratio in the axial and lateral directions was assessed by placing a reflective target within a certain depth of 20% Intralipid (Fresenius Kabi, Uppsala, Sweden) corresponding to a perpendicular optical length of ~7 - 9 and translating this mirror or knife-edge target along the axial or transverse directions, respectively, at a rate of 0.1 μm/sec as driven by a linear actuator (TRA12CC, Newport Corporation, Newton, NJ). A low-pass filter (SR560, Stanford Research Systems, Sunnyvale, CA) was used to remove measurement noise without broadening the axial and transverse response measurements. Compared to DAC, it was found that the MAD technique preserves axial resolution (DACFWHM = MADFWHM = 2.9 - 3.0 μm) and lateral resolution (DAC-3dB = MAD-3dB = 1.9 - 2.0 μm) while improving the rejection of background signal due to out-of-focus and multiply scattered light [Fig. 6]. Specifically, under conditions corresponding to Lp = 2μsd = 2(30 to 40 mm−1)(0.116 mm) ~7 to 9, we consistently observed an improvement in signal-to-background (SBR) of 5 - 6 dB in the MAD technique over DAC when testing the system’s performance in the axial direction and a ~4 dB improvement in the in-plane response to a knife-edge transition (chrome-to-glass edge).
Fig. 6 Axial and transverse responses of MAD and DAC in scattering media. Intralipid experiments demonstrate consistent improvement in optical sectioning contrast in the MAD technique over DAC. (a) Contrast improves by 5 - 6 dB axially without compromising optical section thickness. (b) In-plane resolution is maintained while improving contrast by up to ~4 dB. The signal labeled as 'Water' refers to characterization of the system in the absence of scattering (whereas scattering is achieved using Intralipid).
3.3 Imaging
A phantom consisting of bright fluorescent beads within a fluorescent background was constructed to compare the contrast (signal to background ratio) of MAD and DAC microscopy at various depths. Vertical and en face images of this fluorescent phantom demonstrate the qualitative advantages of MAD over DAC while en face line profiles in the y-direction through selected fluorescent beads show evidence of a quantitative improvement in contrast at a fixed imaging depth [Fig. 7].
Fig. 7 MAD imaging performance in fluorescent phantoms. Vertical and en face images acquired at a depth of ~200 μm demonstrate the MAD technique’s ability to reduce the amount of background fluorescence due to out-of-focus and multiply scattered light, thereby improving the contrast when visualizing bright fluorescent beads in a dim fluorescent matrix. Line profiles (bottom left) in the y-direction through selected fluorescent beads from the en face images show improved contrast. Scale bars measure 20 μm.
Initial tissue-imaging studies were performed using mouse tissues stained for 15 minutes with Cyto647 fluorescent dye, and then rinsed for 30 seconds with PBS, and also demonstrate the ability of the MAD technique to reject more out-of-focus and multiply-scattered light than DAC. Clear enhancements in contrast were seen [Fig. 8, white arrows].
Fig. 8 MAD imaging performance in fresh tissues. En face images from fresh mouse tissues demonstrate distinct contrast enhancement in MAD compared with DAC (white arrows indicate example features). Imaging depths for the intestine, renal capsule and renal medulla examples are 25 μm, 80 μm, and 60 μm, respectively. Scale bars measure 40 μm.
4. Discussion
We have presented a new strategy to enable optical-sectioning microscopy at large depths that benefits from the inherent strengths of dual-axis confocal (DAC) microscopy. We leverage the fact that the DAC architecture’s intersecting illumination and collection beams significantly improves the spatial-filtering and optical-sectioning performance compared with conventional single-axis confocal microscopy. This performance is further improved by modulating the spatial alignment of the dual-axis beams at a frequency f such that the signal generated at the focal volume of the microscope is modulated at 2f, with minimal modulation of the background signal, thus providing a substantial improvement in optical-sectioning contrast (SBR) without sacrificing resolution.
Several factors play a role regarding the discrepancy in contrast (SBR) when comparing MAD to DAC in simulations vs. experiments (i.e., Monte-Carlo simulations predict 9 dB of contrast improvement axially whereas Intralipid experiments yield 5 - 6 dB improvement). For example, the first-order light used as a spatially modulated illumination source in this system is known to vary in intensity over time since AOD diffraction efficiency is a function of the RF driving frequency (deflection angle of the output beam). This leads to background signals at 2f, which raises the background floor leading to an overall reduction in SBR. Future experiments will utilize AODs in which this effect is minimized through various compensation methods. Secondly, although an AOD driven at a constant frequency produces a well-defined effective grating spacing, the driving acoustic wave has a finite propagation speed (3.63 mm/μsec). When driving the AOD at a sufficiently high modulation frequency, the acoustic grating spacing is not constant over the transverse extent of the beam, leading to undesired spatial dispersion and a non-ideal beam profile. However, we found that limiting our modulation rate to 0.5 MHz or less helped to mitigate these effects. Future investigations will explore the use of optimized AODs and alternative spatial-modulation approaches to increase the modulation speed, and therefore the frame rate, of MAD. Finally, the MAD technique may be limited in heterogeneous tissues that cause aberrations and misalignment of the beams due to refractive beam steering [27]. The use of adaptive optics, or self-reconstructing beams (e.g., Bessel beams) that are less sensitive to beam steering, may be potential strategies for minimizing these effects [28, 29].
Despite various limitations requiring further attention, the MAD technique synergizes the benefits of two promising approaches (DAC and focal modulation) to push the imaging depth of a linear optical-sectioning microscope to greater limits. This technique utilizes simple low-power diode lasers rather than pulsed lasers (as in multi-photon microscopy). Since DAC microscopy reliably attains imaging depths of up to 300 - 400 μm in epithelial tissues [16], extending the range further with the MAD technique could potentially allow for an additional 100 - 150 μm of imaging (future work). Such an extension of the bound on imaging depth in confocal microscopy is particularly useful in examining the full thickness and basement membranes of clinical mucosal specimens through which epithelial-based tumors may begin to metastasize. MAD microscopy fills a unique niche in that it retains the resolution of confocal microscopy while increasing its penetration depth with minimal change to the optical configuration of DAC microscopy. Our approach is purely optical and is therefore capable of diffraction-limited sub-cellular resolution and can also leverage the large arsenal of fluorescent contrast agents available for multiplexed molecular imaging.
Acknowledgments
The authors acknowledge funding support from the National Institute for Biomedical Imaging and Bioengineering R00EB008557 (Liu), the National Institute of Dental and Craniofacial Research R01DE023497 (Liu), the National Cancer Institute R01CA175391 (Liu) and the Office of the Vice President for Research at Stony Brook University.
References and links
1. S. Y. Leigh and J. T. C. Liu, “Multi-color miniature dual-axis confocal microscope for point-of-care pathology,” Opt. Lett. 37(12), 2430–2432 (2012). [CrossRef] [PubMed]
2. J. T. C. Liu, N. O. Loewke, M. J. Mandella, R. M. Levenson, J. M. Crawford, and C. H. Contag, “Point-of-care pathology with miniature microscopes,” Anal. Cell Pathol. (Amst.) 34(3), 81–98 (2011). [PubMed]
3. J. T. C. Liu, M. J. Mandella, N. O. Loewke, H. Haeberle, H. Ra, W. Piyawattanametha, O. Solgaard, G. S. Kino, and C. H. Contag, “Micromirror-scanned dual-axis confocal microscope utilizing a gradient-index relay lens for image guidance during brain surgery,” J. Biomed. Opt. 15(2), 026029 (2010). [CrossRef] [PubMed]
4. O. De Wever and M. Mareel, “Role of tissue stroma in cancer cell invasion,” J. Pathol. 200(4), 429–447 (2003). [CrossRef] [PubMed]
5. M. Bates, B. Huang, G. T. Dempsey, and X. Zhuang, “Multicolor super-resolution imaging with photo-switchable fluorescent probes,” Science 317(5845), 1749–1753 (2007). [CrossRef] [PubMed]
6. E. Betzig, G. H. Patterson, R. Sougrat, O. W. Lindwasser, S. Olenych, J. S. Bonifacino, M. W. Davidson, J. Lippincott-Schwartz, and H. F. Hess, “Imaging intracellular fluorescent proteins at nanometer resolution,” Science 313(5793), 1642–1645 (2006). [CrossRef] [PubMed]
7. S. Berning, K. I. Willig, H. Steffens, P. Dibaj, and S. W. Hell, “Nanoscopy in a living mouse brain,” Science 335(6068), 551 (2012). [CrossRef] [PubMed]
8. M. G. L. Gustafsson, “Nonlinear structured-illumination microscopy: wide-field fluorescence imaging with theoretically unlimited resolution,” Proc. Natl. Acad. Sci. U.S.A. 102(37), 13081–13086 (2005). [CrossRef] [PubMed]
9. M. Rajadhyaksha, G. Menaker, T. Flotte, P. J. Dwyer, and S. González, “Confocal examination of nonmelanoma cancers in thick skin excisions to potentially guide Mohs micrographic surgery without frozen histopathology,” J. Invest. Dermatol. 117(5), 1137–1143 (2001). [CrossRef] [PubMed]
10. J. Huisken, J. Swoger, F. Del Bene, J. Wittbrodt, and E. H. Stelzer, “Optical sectioning deep inside live embryos by selective plane illumination microscopy,” Science 305(5686), 1007–1009 (2004). [CrossRef] [PubMed]
11. W. Denk, J. H. Strickler, and W. W. Webb, “Two-photon laser scanning fluorescence microscopy,” Science 248(4951), 73–76 (1990). [CrossRef] [PubMed]
12. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Hee, T. Flotte, K. Gregory, C. A. Puliafito, and et, “Optical coherence tomography,” Science 254(5035), 1178–1181 (1991). [CrossRef] [PubMed]
13. E. H. Stelzer and S. Lindek, “Fundamental reduction of the observation volume in far-field light microscopy by detection orthogonal to the illumination axis: confocal theta microscopy,” Opt. Commun. 111(5-6), 536–547 (1994). [CrossRef]
14. L. K. Wong, M. J. Mandella, G. S. Kino, and T. D. Wang, “Improved rejection of multiply scattered photons in confocal microscopy using dual-axes architecture,” Opt. Lett. 32(12), 1674–1676 (2007). [CrossRef] [PubMed]
15. J. T. C. Liu, M. J. Mandella, J. M. Crawford, C. H. Contag, T. D. Wang, and G. S. Kino, “Efficient rejection of scattered light enables deep optical sectioning in turbid media with low-numerical-aperture optics in a dual-axis confocal architecture,” J. Biomed. Opt. 13(3), 034020 (2008). [CrossRef] [PubMed]
16. J. T. C. Liu, M. J. Mandella, S. Friedland, R. Soetikno, J. M. Crawford, C. H. Contag, G. S. Kino, and T. D. Wang, “Dual-axes confocal reflectance microscope for distinguishing colonic neoplasia,” J. Biomed. Opt. 11(5), 054019 (2006). [CrossRef] [PubMed]
17. 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,” Nat. Biotechnol. 21(7), 803–806 (2003). [CrossRef] [PubMed]
18. D. A. Boas, D. H. Brooks, E. L. Miller, C. A. DiMarzio, M. Kilmer, R. J. Gaudette, and Q. Zhang, “Imaging the body with diffuse optical tomography,” IEEE Sig. Proc. Mag. 18(6), 57–75 (2001). [CrossRef]
19. A. Bilenca, A. Ozcan, B. Bouma, and G. Tearney, “Fluorescence coherence tomography,” Opt. Express 14(16), 7134–7143 (2006). [CrossRef] [PubMed]
20. Y. Chen, D. Wang, and J. T. C. Liu, “Assessing the tissue-imaging performance of confocal microscope architectures via Monte Carlo simulations,” Opt. Lett. 37(21), 4495–4497 (2012). [CrossRef] [PubMed]
21. N. Chen, C. H. Wong, and C. J. R. Sheppard, “Focal modulation microscopy,” Opt. Express 16(23), 18764–18769 (2008). [CrossRef] [PubMed]
22. K. Isobe, H. Kawano, T. Takeda, A. Suda, A. Kumagai, H. Mizuno, A. Miyawaki, and K. Midorikawa, “Background-free deep imaging by spatial overlap modulation nonlinear optical microscopy,” Biomed. Opt. Express 3(7), 1594–1608 (2012). [CrossRef] [PubMed]
23. K. Isobe, H. Kawano, A. Kumagai, A. Miyawaki, and K. Midorikawa, “Implementation of spatial overlap modulation nonlinear optical microscopy using an electro-optic deflector,” Biomed. Opt. Express 4(10), 1937–1945 (2013). [CrossRef] [PubMed]
24. G. Wagnières, S. Cheng, M. Zellweger, N. Utke, D. Braichotte, J. P. Ballini, and H. van den Bergh, “An optical phantom with tissue-like properties in the visible for use in PDT and fluorescence spectroscopy,” Phys. Med. Biol. 42(7), 1415–1426 (1997). [CrossRef] [PubMed]
25. L. G. Henyey and J. L. Greenstein, “Diffuse radiation in the galaxy,” Astrophys. J. 93, 70–83 (1941). [CrossRef]
26. D. Wang, Y. Chen, Y. Wang, and J. T. C. Liu, “Comparison of line-scanned and point-scanned dual-axis confocal microscope performance,” Opt. Lett. 38(24), 5280–5283 (2013). [CrossRef] [PubMed]
27. D. Wang, Y. Chen, and J. T. C. Liu, “A liquid optical phantom with tissue-like heterogeneities for confocal microscopy,” Biomed. Opt. Express 3(12), 3153–3160 (2012). [CrossRef] [PubMed]
28. F. O. Fahrbach and A. Rohrbach, “Propagation stability of self-reconstructing Bessel beams enables contrast-enhanced imaging in thick media,” Nat. Commun. 3, 632 (2012). [CrossRef] [PubMed]
29. N. Ji, D. E. Milkie, and E. Betzig, “Adaptive optics via pupil segmentation for high-resolution imaging in biological tissues,” Nat. Methods 7(2), 141–147 (2010). [CrossRef] [PubMed]
