Research

We develop new tools for biomedical imaging to expand the capabilities of optical imaging and apply these tools to address new and pressing biological questions. New capabilities include expanding our ability to image deeper into tissues, to image with higher spatial resolution, to image at higher speeds, and to obtain information that enables us to follow dynamics and interaction in cells and tissues. To develop these capabilities we work across computational, chemical, multiscale, and hyperspectral imaging with our collaborators. Above all, we seek to take measurements and observe dynamics that have not been able to be studied or in which previous techniques could only provide limited information. This work provides the foundation for studying new biological questions and ultimately for the study and diagnosis of diseases. Moreover, as active investigations into biological phenomena run into new barriers, they will produce technical challenges that drive our development of novel optical imaging solutions. Ongoing projects center on the following areas.

  • Dynamic super-resolution microscopy using quantum correlations

    To study the dynamics of interactions between organelles in the cell, we are developing a new approach to enable dynamic (high-speed) super-resolution microscopy using quantum correlations. Super-resolution microscopy refers to a range of microscopy techniques that improve the spatial resolution of optical imaging that is normally imposed in conventional nonlinear microscopy by the properties of light propagation, i.e. the diffraction limit.

    In the first super-resolution microscopy methods based on stimulated emission depletion (STED), a confocal microscope is modified to include both a diffraction-limited focal spot to initially excite a set of fluorescent molecules, followed by a time-delayed depletion doughnut beam that de-excite molecules on the periphery. Another class of super-resolution microscopy methods, called single molecule localization microscopy (SMLM), pinpoints the location of single fluorescent molecules by imaging fluorescent light emitted by a molecule onto a very sensitive camera. But both of these super-resolution methods are limited in the field of view over which they can image and dynamics of imaging. In the STED case, only one small point is imaged at a time, thus scanning large volumes is slow, restricting dynamics. In the case of SMLM, only a small number of fluorophores can be active (i.e., excited and emitting light) for any image from which the location of emitters are estimated. To reveal structures, 1000s of cycles of excitation and imaging are required. In addition, axial localization information is restricted to very shallow depths. Consequently, SMLM is also restricted in imaging volume and speed.

    We are developing a new concept in localization microscopy, where localization of fluorophores is determined by a spatially varying illumination pattern that labels each fluorophore position with a unique intensity pattern in time. The illumination light excites fluorophores and when excited, those fluorophores emit light in proportion to the strength of the excitation by the illumination light. As a result, the emitted fluorescent light from each molecule serves as a beacon, emitting a temporal pattern of fluorescent light that follows the temporal pattern of the illumination light at the location of the emitter. A portion of the fluorescent light power emitted by a set of emitters is recorded by an optical detector (e.g., a single pixel detector), producing an electrical signal that is the sum of the temporal patterns emitted by the fluorescent molecules in the illumination volume.

    We have analyzed the full electronic signal for a wide array of computational imaging modalities, including holography and diffraction tomography using fluorescent light. However, the spatial resolution of those strategies was limited by light propagation, even when implementing super-resolution imaging. Here, we are developing a new strategy: single pixel localization microscopy (SPLM), enabled by matching the known illumination pattern that varies spatially to the recorded fluorescent signal patterns recorded on the single pixel detector. In this approach, we use spatio-temporal modulation of the illumination light to illuminate a large volume of a sample simultaneously. We have shown through an analysis with statistical information theory that it is possible to estimate the location of emitters with <5 nm precision over imaging volumes orders of magnitude larger than is possible with conventional super-resolution imaging methods. SPLM is ultimately limited by emitter density. To circumvent that density limitation, we are expanding our SPLM by detecting emitted photons in a generalized Hanbury Brown-Twiss intensity correlation experiment. Quantum correlations are extracted from these data to provide additional information from the quantum correlations exhibited by single quantum emitters. This strategy is a generalization of an approach for quantum-enhanced computational super-resolution imaging method that we recently explored theoretically.

    Bartels et al. 2022 | Conceptual rendering of the experiment showing a spatiotemporally modulated optical illumination by a sparse set of mutually coherent beams. Beam interference produces the spatially structured illumination as illustrated in the main figure and insets (A) and (B). A large focal volume is achieved because each beam encompasses a small region of spatial frequency support in the pupil plane. The center spatial frequency of the beams scans across the pupil, cycling through a set of complex illumination patterns with spatial frequency structure that samples the full numerical aperture of the illumination objective lens throughout the full temporal modulation cycle. The figure shows an unfolded microscope; however, epi detection is possible. (A) A zoomed-in example of the structured illumination light intensity at one time sample. The specimen is placed in the region of the slide. (B) The spatial structure of the illumination intensity in the plane of the slide for two time points. (C) Examples of generalized HBT detection showing cases of two and three simultaneous photon detection events.

    Representative Publications

    • Bartels RA, Murray G, Field J, Squier J. Super-Resolution Imaging by Computationally Fusing Quantum and Classical Optical Information. Intelligent Computing. 2022 Jan;2022:0003.
    • Field JJ, Winters DG, Bartels RA. Single-pixel fluorescent imaging with temporally labeled illumination patterns. Optica, OPTICA. 2016 Sep 20;3(9):971–4.
    • Stockton PA, Field JJ, Squier J, Pezeshki A, Bartels RA. Single-pixel fluorescent diffraction tomography. Optica, OPTICA. 2020 Nov 20;7(11):1617–20.
    • Xiu M, Field J, Bartels R, Pezeshki A. Fisher information and the Cramér–Rao lower bound in single-pixel localization microscopy with spatiotemporally modulated illumination. J Opt Soc Am A, JOSAA. 2023 Jan 1;40(1):185–203.
  • Computational super-resolution imaging with coherent nonlinear interactions

    Nonlinear microscopy relies on longer-wavelength light, resulting in imaging resolution that is lower than an equivalent conventional imaging system using shorter wavelength light. The Bartels lab pioneers super-resolution imaging methods for resolving finer spatial features in nonlinear microscopy by developing new methods in sum frequency generation imaging, second and third harmonic generation imaging, and coherent anti-Stokes Raman (CARS) spectroscopy.

    For example, we developed the first super-resolution imaging approach for coherent nonlinear scattering, cutting the imaging resolution limit of THG microscopy in half.

    Another super-resolution imaging method used spatial-frequency (Fourier) structured light and single pixel (bucket) detection to enable super-resolution of fluorescent nonlinear and coherent scattering nonlinear imaging simultaneously. We were also the first to demonstrate super resolution imaging with second harmonic generation (SHG), using spatial frequency modulation for imaging, or SPIFI. SPIFI originates as a method to improve the imaging speed of standard point scanning nonlinear microscopy, while still using single pixel detection, which is beneficial for imaging deep within tissues where optical scattering is strong. This is also the first method that works for both coherent nonlinear scattering and for fluorescent emission for super-resolution imaging.

    Fantuzzi et al., 2023 | (a) Multimodal imaging of connective tissue of a 20-μm-thick human skin sample. The CARS-RIM and SFG-RIM images (right column) are compared with average CARS and average SFG images. The composite image corresponds to the superposition of the SFG and CARS signals. (b) Spectroscopic CARS-RIM images of PS beads in a suspension of PP powder diluted in D2O. The CARS signal is detected at Raman resonances of 2,400, 2,850 and 3,050 cm−1, matching the stretching vibrations of D2O, methylene (–CH2–) group within PP and the aromatic =C–H breathing vibration of PS, respectively. The average power in each beam is ~250 mW.

    Most recently, we have begun extending super-resolution imaging to wide-field, aberration-free SHG holographic microscopy and collaborated with a group at the Fresnel Institute in Marseille, France on wide-field super-resolution CARS random illumination microscopy.

    Representative Publications

    • Field JJ, Wernsing KA, Domingue SR, Allende Motz AM, DeLuca KF, Levi DH, et al. Superresolved multiphoton microscopy with spatial frequency-modulated imaging. Proceedings of the National Academy of Sciences. 2016 Jun 14;113(24):6605–10. https://doi.org/10.1073/pnas.1602811113
    • Fantuzzi EM, Heuke S, Labouesse S, Gudavičius D, Bartels R, Sentenac A, et al. Wide-field coherent anti-Stokes Raman scattering microscopy using random illuminations. Nat Photon. 2023 Dec;17(12):1097–104. https://doi.org/10.1038/s41566-023-01294-x
    • Masihzadeh O, Schlup P, Bartels RA. Enhanced spatial resolution in third-harmonic microscopy through polarization switching. Opt Lett, OL. 2009 Apr 15;34(8):1240–2. https://doi.org/10.1364/OL.34.001240

    SPIFI tutorial

    • Stockton P, Murray G, Field JJ, Squier J, Pezeshki A, Bartels RA. Tomographic single pixel spatial frequency projection imaging. Optics Communications. 2022 Oct 1;520:128401. https://doi.org/10.1016/j.optcom.2022.128401
  • Ultra-sensitive coherent and low-frequency Raman spectroscopy

    The vibrational frequency of materials and molecules represents a unique fingerprint by which to identify them and to explore their dynamics and interactions. Thanks to narrow Raman vibrational spectroscopic lines, Raman spectroscopy is a long-sought method of intrinsic detection of molecules across biological and material sciences. For example, it has become an increasingly valuable tool for label-free imaging of metabolism, particularly for metabolomics and cancer tissue classification, and for identifying species and strains of bacteria.

    However, conventional methods suffer from long integration times and lack sufficient sensitivity to detect low concentrations of small molecules in cells, thanks in large part to the weak Raman scattering cross-section and incoherent nature of spontaneous Raman scattering.

    We develop improved methods and instrumentation for coherent Raman scattering, which stimulates the scattering to drive a larger response and then uses measurements of the spatial variations of the strength of Raman scattering at vibrational frequencies to enable the study of changes in chemical composition within a specimen. The goal of this work is to improve detection sensitivity for low concentration molecules and push the new frontier of low frequency Raman forward. Low-frequency vibrational modes have been historically neglected but are biologically important. They are associated with large reduced mass and correspond to vibrational motion that occurs over an extended region, including protein deformation and virus capsid vibrations. Implementation of Raman scattering at low vibrational frequencies remains a persistent challenge (even in coherent Raman approaches) because inelastic scattering at small offsets is so challenging to measure.

    Our approaches make use of impulsive stimulated Raman scattering (ISRS) to detect extremely weak Raman spectroscopy signals and probe low-frequency vibrational modes with time-domain spectroscopy. ISRS, a third-order nonlinear pump–probe technique, produces intense signals by coherently exciting all Raman vibrations with a period longer than the pulse duration. We pioneer a phase detection method, called Doppler Raman Spectroscopy, that improves the sensitivity of ISRS. With Doppler Raman spectroscopy, we use a short laser pulse followed by a time-delayed probe pulse that experiences a small frequency shift because of scattering from the coherent charge density oscillations produced by the pump pulse. The frequency shift imparted on the probe pulse is converted into a phase shift of a radio frequency electronic signal. Doppler Raman is the only Raman spectroscopy method that amplifies the Raman signal outside of the interaction volume between the laser(s) and the molecule. This capability is important because all Raman methods scale with illumination intensity, and thus detection sensitivity is constrained by limits to optical power that are needed to prevent damage to the sample. By contrast, Doppler Raman allows chemical detection that is more sensitive to both spatial and temporal chemical dynamics than other coherent Raman techniques. Indeed, this technique has demonstrated unprecedented sensitivity for coherent Raman spectroscopy, reaching well below mM concentration with sub-second integration times.

    Doppler Raman scattering uses a pair of femtosecond laser pulses. a) The first pulse excites Raman vibrations, producing a time-varying refractive index. A second pulse passes through the sample at a later time. b) This second pulse is spectrally scattered, which leads to a shift in the center of mass of the probe spectrum. Ultrasensitive Raman detection is achieved by converting this spectral shift into a time delay, which is measured using an approach adapted from precision metrology of laser pulse train jitter. c) Doppler Raman can detect the Raman spectrum of many metabolites. * indicates published peaks with spontaneous Raman measurements. d) Raman spectrum of coenzymes at different redox states. e) Because fluorescent signal from cytochrome c buries the spontaneous Raman signal, vibrational information about cytochrome c is nearly impossible to access under normal conditions.​

    Representative Publications

    • Bartels RA, Oron D, Rigneault H. Low frequency coherent Raman spectroscopy. J Phys Photonics. 2021 Sep;3(4):042004. https://doi.org/10.1088/2515-7647/ac1cd7
    • Smith DR, Shivkumar S, Field J, Wilson JW, Rigneault H, Bartels RA. Nearly degenerate two-color impulsive coherent Raman hyperspectral imaging. Opt Lett, OL. 2022 Nov 15;47(22):5841–4. https://doi.org/10.1364/ol.467970
    • Smith DR, Wilson JW, Shivkumar S, Rigneault H, Bartels RA. Low frequency coherent Raman imaging robust to optical scattering. arXiv; 2024 Available from: http://arxiv.org/abs/2402.07006 https://doi.org/10.48550/arXiv.2402.07006
    • Smith DR, Field JJ, Winters DG, Domingue SR, Rininsland F, Kane DJ, et al. Phase noise limited frequency shift impulsive Raman spectroscopy. APL Photonics. 2021 Feb 19;6(2):026107. https://doi.org/10.1063/5.0038624
    • Heuke S, Sivankutty S, Scotte C, Stockton P, Bartels RA, Sentenac A, et al. Spatial frequency modulated imaging in coherent anti-Stokes Raman microscopy. Optica, OPTICA. 2020 May 20;7(5):417–24. https://doi.org/10.1364/optica.386526
    • Domingue SR, Winters DG, Bartels RA. Time-resolved coherent Raman spectroscopy by high-speed pump-probe delay scanning. Opt Lett, OL. 2014 Jul 15;39(14):4124–7. https://doi.org/10.1364/ol.39.004124
  • Widefield nonlinear holographic microscopy for deep tissue imaging

    We are developing holography nonlinear microscopy and computational adaptive optics for imaging deep inside (uncleared) tissue and working with pathologists to look at human tissues for deeper diagnostic imaging. Nonlinear, second harmonic generation (SHG) microscopy targets deeper tissue imaging but normally relies on point scanning, which is limited in speed and in signal-to-noise ratio. By using widefield SHG and THG holography, we are able to boost image signals as well as estimate and correct for optical aberrations accumulated from optics and specimen distortions, including correcting for scattering. With this approach, we are pushing to imaging depths beyond the ballistic imaging barrier. We send in powerful laser light at one frequency, and structures such as collagen will double (or triple, in THG) the frequency of that light. These structures scatter light, much like clouds or milk. This scattering distorts and weakens the imaging process—and has limited conventional methods to the shallow depths in tissue that ballistic (unscattered) light can travel. But by taking a sequence of holograms, we can undo the light scattering and recover high resolution images of the organization of collagen in tissues.

    Murray et al., 2023 | Experimental results for SHG synthetic aperture with aberration correction in the epi (a–d) and transmission (e–h) configurations for a 10 micrometer thick section of sheep tendon. From left to right: a) and e) show the reconstruction before any corrections are applied; b) and f) show the intensity after correction with the input and output pupil phase corrections inset; c) and g) show the phase of the synthetic aperture reconstructions after the corrections are applied, and the reconstructed spectrum in images d) and h) have dashed outlines showing the original spatial frequency support of the system shifted to a few example positions stitching together an expanded spatial frequency support containing more information. Scale bar is 50 micrometers.

    Representative Publications

    • Farah Y, Murray G, Field J, Xiu M, Wang L, Pinaud O, et al. Synthetic aperture holographic third harmonic generation microscopy [Internet]. arXiv; 2024 [cited 2024 Apr 3]. Available from: http://arxiv.org/abs/2402.04077
    • Masihzadeh O, Schlup P, Bartels RA. Label-free second harmonic generation holographic microscopy of biological specimens. Opt Express, OE. 2010 May 10;18(10):9840–51. https://doi.org/10.1364/oe.18.009840
    • Smith DR, Winters DG, Bartels RA. Submillisecond second harmonic holographic imaging of biological specimens in three dimensions. Proceedings of the National Academy of Sciences. 2013 Nov 12;110(46):18391–6. https://doi.org/10.1073/pnas.1306856110
    • Fantuzzi EM, Heuke S, Labouesse S, Gudavičius D, Bartels R, Sentenac A, et al. Wide-field coherent anti-Stokes Raman scattering microscopy using random illuminations. Nat Photon. 2023 Dec;17(12):1097–104. https://doi.org/10.1038/s41566-023-01294-x
    • Murray G, Field J, Xiu M, Farah Y, Wang L, Pinaud O, et al. Aberration free synthetic aperture second harmonic generation holography. Opt Express, OE. 2023 Sep 25;31(20):32434–57. https://doi.org/10.1364/oe.496083
    • Hu C, Field JJ, Kelkar V, Chiang B, Wernsing K, Toussaint KC, et al. Harmonic optical tomography of nonlinear structures. Nat Photonics. 2020 Sep;14(9):564–9. https://doi.org/10.1038/s41566-020-0638-5