Open Access
Add to BibSonomy
Abstract
Label-free optical imaging is valuable in biology and medicine because of its non-destructive nature. Quantitative phase imaging (QPI) and molecular vibrational imaging (MVI) are the two most successful label-free methods, providing morphological and biochemical information, respectively. These techniques have enabled numerous applications as they have matured over the past few decades; however, their label-free contrasts are inherently complementary and difficult to integrate due to their reliance on different light–matter interactions. Here we present a unified imaging scheme with simultaneous and in situ acquisition of quantitative phase and molecular vibrational contrasts of single cells in the QPI framework using the mid-infrared photothermal effect. The robust integration of subcellular morphological and biochemical label-free measurements may enable new analyses, especially for studying complex and fragile biological phenomena such as drug delivery, cellular disease, and stem cell development, where long-time observation of unperturbed cells is needed under low phototoxicity.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. INTRODUCTION
Optical imaging is indispensable in biological and medical science due to its non-destructive property. Label-free imaging, such as quantitative phase imaging (QPI) and molecular vibrational imaging (MVI), is particularly valuable for studying fragile systems where exogenous labeling that often spoils the sample is not preferred [1–18]. QPI yields sample-specific 2D optical-phase-delay [1] or 3D refractive-index (RI) distribution [2], which are the fundamental quantities used to visualize the morphology of transparent samples as in the cases of dark-field, phase-contrast, and differential-interference-contrast microscopy. Compared to these traditional methods, where the optical wavefront is perturbed to couple its phase into amplitude, direct quantification of optical phase delay or RI by QPI reveals high-contrast and smooth images with a higher spatial resolution, allowing for accurate cellular profiling [3]. Its essential capability is to translate the measured QP into cellular dry mass density, which enables quantification of cellular growth rate [4] and numerous other applications [5]. On the other hand, MVI yields comprehensive spectroscopic information of molecular bonds based on Raman scattering [6–10] or mid-infrared (MIR) absorption [11–18]. Coherent Raman [7–10] and, more recently, MIR photothermal [12–17] and MIR photoacoustic [18] imaging techniques have gained attention due to the high spatial resolution and detection sensitivity. The state-of-the-art systems can perform video-rate imaging of, e.g., intracellular proteins, lipids, or nucleic acids, allowing for high dimensional metabolic analysis [10].
Fig. 1. Concept of MV-QPI. (a) Principle of the MV-contrast acquisition in the QPI framework. The MIR light of a certain wavenumber is irradiated to the wide area of the sample, where the resonant biomolecules are selectively excited to their fundamental vibrational states. The vibrational energy is eventually transformed into heat that diffuses into the surrounding medium. The resulting photothermal RI decrease is detected by the QPI system with the spatial resolution of the VIS probe light. (b) Cross-correlative analysis enabled by MV-QPI. The phase or RI image obtained at the MIR OFF state reveals the quantitative and comprehensive morphology of the sample containing rich information about cellular shapes and distributions of intracellular organelles. Scanning the MIR wavenumber visualizes contrasts of various MV resonances at each spatial point in the FOV, which can be decomposed into individual biomolecular constituents through chemometric analysis. (c) Mechanism of the diffraction limit in the standard 2D QPI. The object is illuminated at the normal angle with a plane wave, and only a limited range of spatial-frequency information of the diffracted light is collected with the objective lens. (d) Mechanism of the depth- and super-resolution in the synthetic-aperture QPI. The object is illuminated with the angled plane wave such that higher-frequency contents that used to be outside the NA of the objective lens in (c) can be collected. Scanning the angle of the illumination allows us to computationally synthesize the 3D frequency aperture. The depth- and super-resolved imaging performance can be achieved with the expanded axial and lateral bandwidths of the 3D synthetic aperture, respectively. The black dotted curves in the frequency spectrum indicate the Ewald’s spherical cap, which determines the 3D coverage of the NA of the objective lens under a certain angle of illumination.
In spite of their independent technological maturity, the inherent limitations of QPI and MVI are still left unresolved. The RI does not offer chemical specificity [5], whereas the MV can be detected at specific sites where resonant molecules exist. It is also difficult to estimate the quantity of each molecular constituent based on MV spectroscopy without additional knowledge of interaction lengths, absorption or scattering cross sections [6], and molecular masses of unknown biological compositions. Combining the quantitative morphological and qualitative biochemical information can not only mitigate these limitations but also synergistically expand the capability of label-free imaging. Indeed, multimodal spontaneous Raman-QPI has shown the potential to decompose the local cellular dry mass density into the individual biochemical constituents [19]. However, merely combining the independent modalities is not a robust solution because there are always mismatches in temporal and spatial resolutions, sampling points, fields of view (FOVs), and so on. For instance, in the spontaneous Raman-QPI system, the lack of a depth resolution in the QPI prohibits accurate decomposition of cellular dry mass density into independent biomolecular components, whereas the slow acquisition speed of the spontaneous Raman imaging is accompanied by motion-blur artifacts. Accurate correlation of the spatiotemporal subcellular morphological and biochemical evolutions yields more comprehensive and robust pictures of complex biological systems, and a fully label-free implementation would have important implications when studying fragile phenomena.
Here we present a unified imaging scheme that bridges this technological gap between the two label-free modalities, realizing simultaneous and in situ acquisition of MV contrasts in the framework of QPI using the MIR photothermal effect. Preliminary results on this method, which we term MV-sensitive QPI (MV-QPI), have been recently reported [20,21]. The focus of this work is to further develop the MV-QPI method, proving its practical bioimaging capability in the broadband MIR fingerprint region while also pioneering the depth- and super-resolved imaging performance beyond the diffraction limit imposed in other MVI techniques [6–18]. To highlight MV-QPI’s versatility, we present two QPI implementations for single-cell imaging. The first is based on digital holography (DH), where we demonstrate live-cell 2D MV-QPI. The second is based on optical diffraction tomography (ODT), where we demonstrate, for the first time to our knowledge, the depth-resolved MV-QPI. We expect our MV-QPI method to allow for merging the independent knowledge of the QPI and MVI communities, leading to more comprehensive understanding of various biological phenomena.
Fig. 2. Experimental implementations. (a) Synchronization of the pulse trains and image sensor. The VIS and MIR lasers are electrically controlled to synchronize their pulse repetitions (${\sim}{1}\;{\rm kHz}$) and relative time delay. The MIR beam is intensity modulated to be in phase with the half-harmonic of the image sensor’s frame rate (${\sim}{100}\;{\rm Hz}$). (b) MV-DH system. The DH microscope is built based on a commercial microscope housing IX73 (Olympus). The collimated VIS laser beam is used as the plane-wave probe illumination, and the magnified image of the sample is formed at the output port of the microscope. The subsequent ${4}{f}$ system is used to perform the common-path off-axis interferometry. The MIR laser beam is loosely focused to the sample with a ${{\rm CaF}_2}$ lens. QCL, quantum cascade laser. (c) MV-ODT system. The collimated VIS laser beam is split into two paths to create the Mach–Zehnder off-axis interferometer. The deflection angle of the probe beam created by the wedge prism is magnified and relayed into the sample plane by the subsequent tube lens and the illumination objective lens. The resulting angled plane-wave illumination is then collected by another objective lens, and the subsequent tube lens forms the sample’s magnified image on the image sensor. The MIR and VIS beams are combined by the dichroic mirror (DM). To avoid absorption of the MIR light by, e.g., ${{\rm SiO}_2}$-based optics, a reflective objective lens is chosen for illumination.
2. MV-QPI
The concept of MV-QPI is illustrated in Fig. 1. MV-QPI is a super-resolved MIR imaging method based on wide-field visible (VIS) QPI detection of site-specific RI changes induced by the MIR photothermal effect. A wide area of a sample placed in the objective focus of a QPI system is illuminated by MIR light lasing at a certain wavenumber that excites the resonant molecular species to their fundamental vibrational states [see Fig. 1(a)]. Through non-radiative decay of the molecular vibrations, the local RIs in the vicinities of the resonant molecules decrease due to the rise of the temperature [12–17], which are detected by QPI with the spatial resolution of the VIS probe light [20,21]. The obtained MIR “OFF” QP image reveals the quantitative morphology of the sample, while the subtraction of the “OFF” from the “ON” QP image reveals the site-specific phase or RI decrease, which reflects the local MIR absorption property [see Fig. 1(b)]. Scanning the MIR wavenumber yields spectroscopic images of different MV resonances. The broadband MIR absorption spectrum can be obtained at each spatial point, which can be used to identify the local molecular compositions through chemometric analysis. Eventually, we can map the MV-based biomolecular distributions within the global morphology of the sample provided by the QP contrast.
Generally, the MIR photothermal imaging [12–17], including our MV-QPI method [20,21], offers (1) high MV detection sensitivity based on the MIR absorption having a cross-section more than 8 orders of magnitude larger than that of Raman scattering [17] and (2) low photodamage by the use of the low-photon-energy MIR excitation that most unlikely excites the electronic transitions [22] of biomolecules. In this work, we additionally harness the capability of QPI to computationally synthesize the 3D spatial-frequency aperture of the imaging system [2,23]. This enables us to further achieve (3) a higher lateral spatial resolution by the expanded lateral bandwidth of the synthetic aperture. It surpasses the numerical aperture (NA) of a single objective lens, which imposes the diffraction limit in other far-field MVI techniques [6–18,20,21]. The 3D spatial-frequency aperture also allows us to achieve (4) decoupling of the undesired MIR absorption effect caused by an aqueous medium surrounding the sample, which is a well-known problem of MIR microscopy, by depth-resolving power offered by the expanded axial bandwidth of the 3D synthetic aperture.
Figures 1(c) and 1(d) depict the mechanism of the diffraction limit when the sample is illuminated with a coherent wavefront [2,23,24]. Figure 1(c) illustrates the case of standard 2D QPI, where the synthetic aperture technique is not used [1,20,21]. The object is illuminated with a plane wave at the normal angle. A higher-frequency structure diffracts the light at a larger angle. The objective lens collects the transmitted and diffracted light with a limited angular range determined by its NA, imposing the diffraction limit due to the low-pass-filtering effect. Other MVI techniques that use a focused optical probe [6–13,16–18] share the same diffraction-limit mechanism determined by the NA of a single objective lens. In the synthetic-aperture QPI, this limitation can be overcome by using angled plane-wave illumination [2,23,24] [see Fig. 1(d)]. The propagation direction of the diffracted wavefront is accordingly angled, such that the higher-frequency contents can be collected with the same objective lens. Essentially, the illumination angle shifts the location of the NA of the objective lens in the spatial-frequency domain. The Fourier diffraction theorem [25] also allows us to map the 2D frequency aperture to the 3D frequency space when the illumination is monochromatic, which becomes a spherical cap called the Ewald’s sphere as shown in Figs. 1(c) and 1(d). The 3D position of the spherical cap can be stirred by scanning the angle of the illumination [25], allowing us to computationally fill a certain volume of the 3D frequency space [2,23]. The expanded axial and lateral bandwidths of the 3D synthetic aperture result in the depth- and super-resolved imaging performance, respectively. In an extreme case, half-pitch lateral and axial resolutions down to 90 and 150 nm have been realized, respectively [23].
Fig. 3. Basic performance of the MV-ODT system. The liquid oil sandwiched between two ${{\rm CaF}_2}$ substrates is used as the sample, which is excited by the MIR beam with the focus diameter of ${\sim}{30}\;\unicode{x00B5}{\rm m}$. (a) Linearity of the photothermal RI change with respect to the MIR excitation pulse energy. (b) Exponential temporal decay of the photothermal RI change with the decay constant of ${\sim}{130}\;\unicode{x00B5} {\rm s}$. (c) MIR spectrum of the liquid oil obtained by the MV-ODT system, showing good agreement with the FTIR reference spectrum. Each measurement point shown in (a)–(c) represents one voxel of the FOV.
3. EXPERIMENTAL SYSTEMS
Our experimental implementations are schematically shown in Fig. 2. The VIS (10 ns duration, 532 nm wavelength) and MIR (1 µs duration, tunable wavenumber over ${{1450} {-} {1645}}\;{{\rm cm}^{ - 1}}$) lasers produce electrically synchronized optical pulse trains at a rate of 1 kHz [see Fig. 2(a)]. We fix the time delay between the two pulse trains such that the VIS pulse arrives at the sample soon after the falling edge of the MIR pulse to maximize the photothermal signal due to the heat accumulation. The MIR light is intensity modulated by a square wave such that the image sensor alternately captures the MIR ON and OFF frames at ${\sim}{100}\;{\rm Hz}$. The MIR pulse fluence is ${\sim}{10}\;{{\rm pJ/\unicode{x00B5}{\rm m}}^2}$ (${\sim}{100}\;{\rm nJ}$ over ${\sim}{100\;\unicode{x00B5}{\rm m}} \times {100\; \unicode{x00B5}{\rm m}}$) but depends on the wavenumber. The VIS pulse fluence can be as low as ${\sim}{0.1}\;{{\rm pJ/\unicode{x00B5}{\rm m}}^2}$ (${\sim}{1}\;{\rm nJ}$ over ${\sim}{100\; \unicode{x00B5}{\rm m}} \times {100\;\unicode{x00B5}{\rm m}}$), which is 3–4 orders of magnitude lower than that used in, e.g., coherent Raman imaging [10].
Figure 2(b) shows our MV-DH system. DH is a wide-field interferometric technique to measure a 2D map of optical phase delay [1,20,21,26]. A collimated laser beam illuminates the sample, and its magnified complex-field image is replicated by a diffraction grating. The zeroth-order term is low-pass filtered to create a reference quasi-plane wave, while the first-order term is transmitted unperturbed, such that these two terms create an interferogram on an image sensor. The illumination optical power is ${\sim}{100}\;{\unicode{x00B5}\rm W}$, which is enough to use the full dynamic range of the image sensor that runs at the frame rate of 100 Hz. The mechanism of the diffraction limit of DH is illustrated in Fig. 1(c). The diffraction-limited half-pitch lateral resolution is ${\sim}{440}\;{\rm nm}$, which is determined by the NA of the objective lens = 0.6. The temporal phase sensitivity of our DH system is dominated by the optical shot noise with ${\sim}{10}\;{\rm mrad}$ in standard deviation without averaging.
Figure 2(c) shows our MV-ODT system. ODT provides a depth-resolved 3D RI map of the sample through multi-angle tomographic measurements and computational reconstruction incorporating the diffraction effect [2,23]. A collimated VIS laser beam illuminates the sample at various incident angles stirred by a rotating wedge prism, and its magnified complex-field image is detected with an image sensor in the configuration of Mach–Zehnder interferometry with a reference wave. A reflective objective lens is used to deliver the VIS and MIR pulses to the sample, while a glass-based one is used as a collection lens for the VIS imaging. We use a fixed illumination NA of 0.55 and scan nine azimuthal angles with an increment of 36 deg for the VIS light illumination. The VIS light illumination optical power is ${\sim}{1}\;{\unicode{x00B5}\rm W}$, which is enough to use the full dynamic range of the image sensor that runs at the frame rate of 60 Hz. The mechanism of the diffraction limit of ODT is illustrated in Fig. 1(d). The diffraction-limited half-pitch lateral and axial resolutions are ${\sim}{190}\;{\rm nm}$ and ${\sim}{2.3}\;{\unicode{x00B5}{\rm m}}$, respectively, as determined by the illumination and collection NAs of 0.55 and 0.85, respectively. The temporal RI sensitivity of our ODT system is dominated by the optical shot noise with ${\sim}{2} \times {{10}^{ - 5}}$ in standard deviation without averaging.
4. BASIC PERFORMANCE OF MV-QPI
We first characterize the basic performance of our MV-QPI systems. The performance of our MV-DH system is summarized in our prior work [20]. The performance of our MV-ODT system is summarized in Fig. 3. We measure liquid oil [Series A 1.54000 (Cargille) consisting of aliphatic/alicyclic hydrocarbons and hydrogenated terphenyl] sandwiched between two ${{\rm CaF}_2}$ substrates of 500 µm thickness by exciting it with a MIR beam with a focus diameter of ${\sim}{30}\;{\unicode{x00B5}{\rm m}}$. We average 500 pairs of MIR ON–OFF measurements for each photothermal tomogram. In Fig. 3(a), we verify that the signal (i.e., the RI decrease) varies linearly against the MIR excitation energy. In Fig. 3(b), we confirm exponential temporal decay of the signal with the decay constant of ${\sim}{130}\;{\unicode{x00B5} \rm s}$ by varying the time delay between the MIR and VIS pulses. Similar decay constants could be obtained with other experimental conditions because the thermal diffusivities of various liquids and polymers are in the order of ${{10}^{ - 7}}\;[{{\rm m}^2}/{\rm s}]$ [27,28]. In Fig. 3(c), we perform MIR spectroscopy of the oil by scanning the MIR wavenumber. The obtained spectrum shows good agreement with the spectrum of the same oil obtained by a commercial Fourier-transform infrared spectrometer (FTIR).
5. COMPARISON OF THE DEPTH-RESOLVING CAPABILITY BETWEEN MV-DH AND MV-ODT
In the context of MV-QPI, the depth-resolved quantitative RI imaging capability of ODT is critical for (1) decoupling the MIR photothermal effects in the out-of-focus aqueous layers and (2) quantifying the actual photothermal temperature change. We demonstrate these effects in Fig. 4 by comparing the photothermal DH and ODT images of fixed HEK293 cells immersed in ${{\rm D}_2}{\rm O}$-based phosphate-buffered saline (PBS). We measure the sample with the ODT system only and perform computational DH and ODT reconstructions with the same raw dataset using the synthetic-aperture DH [24] and ODT algorithms, respectively. In short, the synthetic-aperture DH enhances the lateral resolution of DH by the same principle as the multi-angle ODT but without mapping to the 3D frequency space. This ensures that any measurement conditions, such as the MIR fluence, FOV, lateral spatial resolution, etc., are the same for the two QPI cases, except for the depth-resolving capability. Figures 4(a) and 4(b) show the reconstructed QP and RI images of the HEK293 cells, respectively, whereas Figs. 4(c) and 4(d) show their photothermal counterparts, respectively, obtained with the MIR wavenumber tuned to ${1548}\;{{\rm cm}^{ - 1}}$. We average 2500 pairs of MIR ON–OFF measurements, resulting in the acquisition time of 12.5 min. In this experiment, the MIR fluence is intentionally made nonuniform within the FOV to make a clearer comparison. Specifically, the MIR beam diameter is reduced to be smaller than the lateral extension of the FOV (i.e., ${\sim}{50}\;{\unicode{x00B5}{\rm m}}$), while its center is positioned to the top right corner of the FOV. Such nonuniform MIR illumination creates a nonuniform background originating from the water’s photothermal phase change, which contaminates the cell’s photothermal signals with the DH algorithm as shown in Fig. 4(c). On the other hand, this background can be made more flattened by sectioning the in-focus layer with the ODT algorithm as shown by the blue regions of the cross-sectional profiles in Figs. 4(c) and 4(d). Also, some of the intracellular structures, such as those indicated by the red regions in the cross-sectional profiles, are resolved with higher contrasts in the ODT reconstruction. Furthermore, ODT quantifies the photothermal RI change to be ${\sim}{{10}^{ - 5}}$, from which the intracellular temperature rise can be estimated to be ${\sim}{0.1}\;{\rm K}$, assuming the water’s thermo-optic coefficient of ${\sim}{1.4} \times {{10}^{ - 4}}$ $[{1/\!K}]$ [29]. Note that the estimation of the temperature rise is generally not possible with MV-DH because the thickness and RI information are coupled in the obtained phase value.
Fig. 4. Comparison of the depth-resolving capability between MV-DH and MV-ODT. (a), (b) Raw phase and RI images of the fixed HEK293 cells in ${{\rm D}_2}{\rm O}$-based PBS at the MIR OFF state, respectively. The image shown in (b) is a cross-section of one particular height of the reconstructed 3D RI tomogram. (c), (d) Photothermal contrasts of the same FOVs as those shown in (a) and (b), respectively, obtained with the MIR wavenumber tuned to ${1548}\;{{\rm cm}^{ - 1}}$. In (c), the cellular structures are contaminated by the photothermal signals originating from the out-of-focus aqueous layers. In (d), the depth resolution provided by ODT results in the higher contrasts of the cellular structures (indicated by the red arrows) as well as the more uniform and flattened background distribution originating from the MIR absorption of the in-focus water layer (indicated by the blue arrows). The depth-resolved quantification of the RI values also allows for accurate estimation of the photothermal temperature rise inside the cells (${\sim}{0.1}\;{\rm K}$) using the thermo-optic coefficient of water (${\sim}{1.4} \times {{10}^{ - 4}}$ [${1/\!K}$]). In this experiment, the MIR fluence is intentionally made nonuniform within the FOV to make a clearer comparison.
6. LIVE-CELL, BROADBAND MIR-FINGERPRINT MV-QPI
We demonstrate live-cell, broadband MIR-fingerprint imaging of a COS7 cell immersed in ${{\rm H}_2}{\rm O}$-based culture medium with the MV-DH system (see Fig. 5). We average 2500 pairs of MIR ON-OFF frames for each spectral point, resulting in the acquisition time of 50 s. The water’s MIR absorption effect is independently measured and subtracted as explained in Supplement 1. The QP image provided in Fig. 5(a) reveals a comprehensive morphology of the cell where the global cellular shape and various intracellular structures, such as the nucleus, nucleoli, and small cytoplasmic particles, can be recognized. We acquire MV-spectral images at 27 spectral points between 1453 and ${1632}\;{{\rm cm}^{ - 1}}$. This spectral range is in the MIR fingerprint region where ${{\rm CH}_2}$ bending (1450–1500) and the peptide bond’s amide bands [1500–1580 (amide II) and 1580–1700 (amide I)] show characteristic spectral signatures, which are recognized to be abundant in lipids and proteins, respectively [11]. Figure 5(b) shows the obtained MIR spectrum at different locations in the FOV. We can observe different spectral signatures at different intracellular structures, e.g., compared to the cytoplasm, the nucleolus shows a stronger signal of the amide II band centered at ${\sim}{1550}\;{{\rm cm}^{ - 1}}$. Indeed, the MV image of ${1548}\;{{\rm cm}^{ - 1}}$ clearly visualizes the signal localizations at the nucleoli, which could represent rich proteins [see Fig. 5(d)]. Also, the small cytoplasmic localizations of ${1472}\;{{\rm cm}^{ - 1}}$ signal at the cellular boundary could represent the existence of lipid droplets [see Fig. 5(c)].
Fig. 5. Live-cell, broadband MIR-fingerprint MV-DH microscopy. (a) Raw phase image of the live COS7 cell in ${{\rm H}_2}{\rm O}$-based culture medium at the MIR OFF state. (b) MIR spectrum of the nucleolus (orange), cytoplasm (blue), and empty area (gray) indicated by the arrows of the respective colors in (a). The scanned MIR wavenumbers are in the MIR fingerprint region, where spectroscopic signatures of ${{\rm CH}_2}$ bending and peptide bond’s amide bands can be found, which are abundant in lipids and proteins, respectively. Compared to the cytoplasm, the nucleolus shows the stronger signal of the broad absorption centered at ${\sim}{1550}\;{{\rm cm}^{ - 1}}$, which coincides with the amide II band. Each spectral point represents the spatial average of ${3} \times {3}$ diffraction-limited pixels (${1.3}\;\unicode{x00B5}{\rm m} \times {1.3}\;\unicode{x00B5}{\rm m}$). (c), (d) MV images of the cell resonant to 1472 and ${1548}\;{{\rm cm}^{ - 1}}$, respectively, after the spatial and spectral normalization (see Supplement 1 for more detail). In (c), the small cytoplasmic localizations of the MV contrast at the cellular boundary could represent the existence of lipid droplets. In (d), the MV contrast shows strong selectivity on the nucleoli that could represent rich proteins.
7. DEPTH-RESOLVED, BROADBAND MIR-FINGERPRINT MV-QPI
We demonstrate depth-resolved, broadband MIR-fingerprint imaging of fixed HEK293 cells immersed in ${{\rm D}_2}{\rm O}$-based PBS with the MV-ODT system (see Fig. 6). We average 1500 pairs of MIR ON–OFF measurements for each spectral point, resulting in the acquisition time of 7.5 min. We acquire MV-spectral tomograms at 19 spectral points between 1502 and ${1632}\;{{\rm cm}^{ - 1}}$. The cross-sectional images of the sample’s reconstructed 3D RI distribution at two different heights are shown in Figs. 6(a) and 6(b), which section the podia (red arrow) and the nucleoli (red square) of the cells, respectively. The MIR spectrum obtained at the nucleolus resolves the spectroscopic signatures of the amide II band [see Fig. 6(e)], and maps of its resonance at ${1563}\;{{\rm cm}^{ - 1}}$ are shown in Figs. 6(c) and 6(d). The MV contrast is localized at the nucleoli in the top two cells, while it shows a characteristic cytoplasmic distribution surrounding the nucleus in the bottom cell, which reminds us of an endoplasmic reticulum. Such intercellular variation, mainly of protein distribution, could represent different phases of the cell cycle. Finally, we can observe the depth-resolved localization of the MV signal and the RI contrast originating from the nucleolus in Fig. 6(f). To the best of our knowledge, this is the first demonstration of depth-resolved MV-QPI that is enabled by the implementation of the ODT scheme. The temperature rise inside the cells can be estimated to be ${\sim}{0.1}\;{\rm K}$with the water’s thermo-optic coefficient of ${\sim}{1.4} \times {{10}^{ - 4}}$ [${1/\!K}$].
Fig. 6. Depth-resolved, broadband MIR-fingerprint MV-ODT microscopy. (a), (b) Cross-sectional images in two different axial planes of the reconstructed RI tomogram of the fixed HEK293 cells in ${{\rm D}_2}{\rm O}$-based PBS at the MIR OFF state. The images section the podia (red arrows) and nucleoli (red square) of the cells, respectively. (c), (d) MV contrasts of the same FOVs as those shown in (a) and (b), respectively, resonant to ${1563}\;{{\rm cm}^{ - 1}}$. The photothermal temperature rise inside the cells can be estimated to be ${\sim}{0.1}\;{\rm K}$ using the thermo-optic coefficient of water. (e) MIR spectrum at one voxel of the FOV in the nucleolus indicated by the white arrow in (d), resolving the characteristic signature of the amide II band. (f) Enlargement of the red-square regions in (a)–(d). At ${z}={0}\;{\unicode{x00B5}{\rm m}}$, the nucleolus indicated by the red arrow is not visible in the RI or the photothermal contrast. At ${z}={3.3}\;{\unicode{x00B5}{\rm m}}$, the nucleolus appears in the RI contrast, which also gives the signal in the photothermal contrast, demonstrating the depth-resolving capability.
8. DISCUSSION
The presented MV-QPI systems have room for improvement. First, the spatial resolutions of the photothermal images currently do not reach the diffraction limits of the QPI systems but are rather limited by the heat diffusion that occurs within the relatively long MIR pulse duration [20]. In our case, the photothermal heat diffuses by ${\sim}{700}\;{\rm nm}$ to the surrounding medium within the MIR pulse duration of 1 µs before the VIS pulse arrives. We estimate that the pulse duration of ${\sim}{10}\;{\rm ns}$ confines the diffusion within the diffraction limit of the QPI systems [30]. Second, a higher-energy MIR pulse source is desired to increase the photothermal signal by an order of magnitude to create temperature rise of ${\sim}{1}\;{\rm K}$. It is also desired to broaden the spectral tunability of the MIR light source so that various other biomolecules can be probed. In our experimental systems, these issues arise due to the use of the semiconductor MIR quantum cascade laser that offers low optical output power with a limited gain bandwidth, which can be solved by using, e.g., a broadly tunable high-energy nanosecond optical parametric oscillator. Third, a higher-full-well-capacity image sensor [31] can be used to enhance the shot-noise-limited sensitivities by at most 2 orders of magnitude. Fourth, any other QPI methods can be implemented to harness their versatility [5], robustness [32,33], and imaging speed [34–37], while higher-NA objective lenses can be used to enhance the lateral and axial spatial resolutions [23].
With the above-mentioned improvements, MV-QPI method has the potential to achieve high-sensitivity, low-photodamage, and super-resolved MVI. In total, nearly 3 orders of magnitude enhancement in the MV detection sensitivity can be expected. In this case, the VIS fluence at the sample plane increases by ${\sim}{3}$ orders of magnitude and becomes comparable to that used in coherent Raman imaging [10]. Our method can still reduce the photodamage associated with nonlinear electronic transitions of biomolecules [22] since the duration of the VIS pulse can be ${\sim}{10}\;{\rm ns}$, which is significantly longer than the picosecond or femtosecond pulses used in coherent Raman imaging. Also, with the reduced MIR pulse duration, the spatial resolution of the photothermal images can, in principle, reach the diffraction limit of the synthetic-aperture QPI system, which surpasses the diffraction limit imposed in other MVI techniques [6–18,20,21].
We expect that our MV-QPI method could pioneer several interesting research directions. The combination of the MV and QP contrasts could enable us to quantify the dry mass distribution of each biomolecular component inside cells [19] such as proteins, lipids, and nucleic acids, providing novel insights in cellular disease, growth, development, and so on. MV-QPI may also be useful for thermal biology, where the thermal regulation mechanisms of cellular functions are studied [38]. The diffusion process of the photothermal heat can be visualized by scanning the time delay between the MIR and VIS pulses, which should give us information regarding the intracellular thermal diffusivities. MV-QPI could also be used to probe intracellular water molecules. Water molecules are considered to significantly affect bioactivities, as they interact with most of the intracellular and extracellular biomolecules. Raman scattering spectroscopy [39,40] has been recently used as a water-sensing tool to map the distribution and chemical states of intracellular water molecules. Similar applications could be explored with MV-QPI but with the inherently higher sensitivity to the O-H vibrational modes.
Funding
Precursory Research for Embryonic Science and Technology (JPMJPR17G2); Japan Society for the Promotion of Science (17H04852, 17K19071).
Acknowledgment
We thank Makoto Kuwata-Gonokami and Junji Yumoto for letting us use their equipment.
Disclosures
The authors declare no conflicts of interest.
See Supplement 1 for supporting content.
REFERENCES
1. P. Marquet, B. Rappaz, P. J. Magistretti, E. Cuche, Y. Emery, T. Colomb, and C. Depeursinge, “Digital holographic microscopy: a noninvasive contrast imaging technique allowing quantitative visualization of living cells with subwavelength axial accuracy,” Opt. Lett. 30, 468–470 (2005). [CrossRef]
2. Y. Sung, W. Choi, C. Fang-Yen, K. Badizadegan, R. R. Dasari, and M. S. Feld, “Optical diffraction tomography for high resolution live cell imaging,” Opt. Express 17, 266–277 (2009). [CrossRef]
3. R. Kasprowicz, R. Suman, and P. O’Toole, “Characterising live cell behaviour: traditional label-free and quantitative phase imaging approaches,” Int. J. Biochem. Cell Biol. 84, 89–95 (2017). [CrossRef]
4. G. Popescu, Y. K. Park, N. Lue, C. Best-Popescu, L. Deflores, R. R. Dasari, M. S. Feld, and K. Badizadegan, “Optical imaging of cell mass and growth dynamics,” Am. J. Physiol. 295, C538–C544 (2008). [CrossRef]
5. Y. K. Park, C. Depeursinge, and G. Popescu, “Quantitative phase imaging in biomedicine,” Nat. Photonics 12, 578–589 (2018). [CrossRef]
6. D. W. Shipp, F. Sinjab, and I. Notingher, “Raman spectroscopy: techniques and applications in the life sciences,” Adv. Opt. Photon. 9, 315–428 (2017). [CrossRef]
7. C. H. Camp and M. T. Cicerone, “Chemically sensitive bioimaging with coherent Raman scattering,” Nat. Photonics 9, 295–305 (2015). [CrossRef]
8. J. X. Cheng and X. S. Xie, “Vibrational spectroscopic imaging of living systems: an emerging platform for biology and medicine,” Science 350, aaa8870 (2015). [CrossRef]
9. F. Hu, L. Shi, and W. Min, “Biological imaging of chemical bonds by stimulated Raman scattering microscopy,” Nat. Methods 16, 830–842 (2019). [CrossRef]
10. Y. Wakisaka, Y. Suzuki, O. Iwata, A. Nakashima, T. Ito, M. Hirose, R. Domon, M. Sugawara, N. Tsumura, H. Watarai, T. Shimobaba, K. Suzuki, K. Goda, and Y. Ozeki, “Probing the metabolic heterogeneity of live Euglena gracilis with stimulated Raman scattering microscopy,” Nat. Microbiol. 1, 16124 (2016). [CrossRef]
11. M. J. Baker, J. Trevisan, P. Bassan, R. Bhargava, H. J. Butler, K. M. Dorling, P. R. Fielden, S. W. Fogarty, N. J. Fullwood, K. A. Heys, C. Hughes, P. Lasch, P. L. Martin-Hirsch, B. Obinaju, G. D. Sockalingum, J. Sulé-Suso, R. J. Strong, M. J. Walsh, B. R. Wood, P. Gardner, and F. L. Martin, “Using Fourier transform IR spectroscopy to analyze biological materials,” Nat. Protoc. 9, 1771–1791 (2014). [CrossRef]
12. J. M. Lim, C. Park, J. S. Park, C. Kim, B. Chon, and M. Cho, “Cytoplasmic protein imaging with mid-infrared photothermal microscopy: cellular dynamics of live neurons and oligodendrocytes,” J. Phys. Chem. Lett. 10, 2857–2861 (2019). [CrossRef]
13. P. D. Samolis and M. Y. Sander, “Phase-sensitive lock-in detection for high-contrast mid-infrared photothermal imaging with sub-diffraction limited resolution,” Opt. Express 27, 2643–2655 (2019). [CrossRef]
14. Y. Bai, D. Zhang, L. Lan, Y. Huang, K. Maize, A. Shakouri, and J. X. Cheng, “Ultrafast chemical imaging by widefield photothermal sensing of infrared absorption,” Sci. Adv. 5, eaav7127 (2019). [CrossRef]
15. K. Toda, M. Tamamitsu, Y. Nagashima, R. Horisaki, and T. Ideguchi, “Molecular contrast on phase-contrast microscope,” Sci. Rep. 9, 9957 (2019). [CrossRef]
16. Z. Li, K. Aleshire, M. Kuno, and G. V. Hartland, “Super-resolution far-field infrared imaging by photothermal heterodyne imaging,” J. Phys. Chem. B 121, 8838–8846 (2017). [CrossRef]
17. D. Zhang, C. Li, C. Zhang, M. N. Slipchenko, G. Eakins, and J. X. Cheng, “Depth-resolved mid-infrared photothermal imaging of living cells and organisms with submicrometer spatial resolution,” Sci. Adv. 2, e1600521 (2016). [CrossRef]
18. J. Shi, T. T. W. Wong, Y. He, L. Li, R. Zhang, C. S. Yung, J. Hwang, K. Maslov, and L. V. Wang, “High-resolution, high-contrast mid-infrared imaging of fresh biological samples with ultraviolet-localized photoacoustic microscopy,” Nat. Photonics 13, 609–615 (2019). [CrossRef]
19. N. Pavillon, A. J. Hobro, and N. I. Smith, “Cell optical density and molecular composition revealed by simultaneous multimodal label-free imaging,” Biophys. J. 105, 1123–1132 (2013). [CrossRef]
20. M. Tamamitsu, K. Toda, R. Horisaki, and T. Ideguchi, “Quantitative phase imaging with molecular vibrational sensitivity,” Opt. Lett. 44, 3729–3732 (2019). [CrossRef]
21. D. Zhang, L. Lan, Y. Bai, H. Majeed, M. E. Kandel, G. Popescu, and J. X. Cheng, “Bond-selective transient phase imaging via sensing of the infrared photothermal effect,” Light Sci. Appl. 8, 116 (2019). [CrossRef]
22. Y. Fu, H. Wang, R. Shi, and J. X. Cheng, “Characterization of photodamage in coherent anti-Stokes Raman scattering microscopy,” Opt. Express 14, 3942–3951 (2006). [CrossRef]
23. Y. Cotte, F. Toy, P. Jourdain, N. Pavillon, D. Boss, P. Magistretti, P. Marquet, and C. Depeursinge, “Marker-free phase nanoscopy,” Nat. Photonics 7, 113–117 (2013). [CrossRef]
24. V. Micó, J. Zheng, J. Garcia, Z. Zalevsky, and P. Gao, “Resolution enhancement in quantitative phase microscopy,” Adv. Opt. Photon. 11, 135–214 (2019). [CrossRef]
25. E. Wolf, “Three-dimensional structure determination of semi-transparent objects from holographic data,” Opt. Commun. 1, 153–156 (1969). [CrossRef]
26. B. Bahduri, H. Pham, M. Mir, and G. Popescu, “Diffraction phase microscopy with white light,” Opt. Lett. 37, 1094–1096 (2012). [CrossRef]
27. J. Wang and M. Fiebig, “Measurement of the thermal diffusivity of aqueous solutions of alcohols by a laser-induced thermal grating technique,” Int. J. Thermophys. 16, 1353–1361 (1995). [CrossRef]
28. A. Salazar, “On thermal diffusivity,” Eur. J. Phys. 24, 351–358 (2003). [CrossRef]
29. P. Schiebener, J. Straub, J. M. H. Levelt Sengers, and J. S. Gallagher, “Refractive index of water and steam as function of wavelength, temperature and density,” J. Phys. Chem. Ref. Data 19, 677–717 (1990). [CrossRef]
30. V. P. Zharov and D. O. Lapotko, “Photothermal imaging of nanoparticles and cells,” IEEE J. Sel. Top. Quantum Electron. 11, 733–751 (2005). [CrossRef]
31. P. Hosseini, R. Zhou, Y.-H. Kim, C. Peres, A. Diaspro, C. Kuang, Z. Yaqoob, and P. T. C. So, “Pushing phase and amplitude sensitivity limits in interferometric microscopy,” Opt. Lett. 41, 1656–1659 (2016). [CrossRef]
32. S. Chowdhury, M. Chen, R. Eckert, D. Ren, F. Wu, N. Repina, and L. Waller, “High-resolution 3D refractive index microscopy of multiple-scattering samples from intensity images,” Optica 6, 1211–1219 (2019). [CrossRef]
33. F. Merola, P. Memmolo, L. Miccio, R. Savoia, M. Mugnano, A. Fontana, G. D’Ippolito, A. Sardo, A. Iolascon, A. Gambale, and P. Ferraro, “Tomographic flow cytometry by digital holography,” Light Sci. Appl. 6, e16241 (2017). [CrossRef]
34. R. Horisaki, K. Fujii, and J. Tanida, “Diffusion-based single-shot diffraction tomography,” Opt. Lett. 44, 1964–1967 (2019). [CrossRef]
35. K. Lee, K. Kim, G. Kim, S. Shin, and Y. Park, “Time-multiplexed structured illumination using a DMD for optical diffraction tomography,” Opt. Lett. 42, 999–1002 (2017). [CrossRef]
36. J. A. Rodrigo, J. M. Soto, and T. Alieva, “Fast label-free microscopy technique for 3D dynamic quantitative imaging of living cells,” Biomed. Opt. Express 8, 5507–5517 (2017). [CrossRef]
37. R. Horisaki, R. Egami, and J. Tanida, “Single-shot phase imaging with randomized light (SPIRaL),” Opt. Express 24, 3765–3773 (2016). [CrossRef]
38. K. Okabe, N. Inada, C. Gota, Y. Harada, T. Funatsu, and S. Uchiyama, “Intracellular temperature mapping with a fluorescent polymeric thermometer and fluorescence lifetime imaging microscopy,” Nat. Commun. 3, 705 (2012). [CrossRef]
39. L. Shi, F. Hu, and W. Min, “Optical mapping of biological water in single live cells by stimulated Raman excited fluorescence microscopy,” Nat. Commun. 10, 4764 (2019). [CrossRef]
40. M. Nuriya, H. Yoneyama, K. Takahashi, P. Leproux, V. Couderc, M. Yasui, and H. Kano, “Characterization of intra/extracellular water states probed by ultrabroadband multiplex coherent anti-stokes Raman scattering (CARS) spectroscopic imaging,” J. Phys. Chem. A 123, 3928–3934 (2019). [CrossRef]
