System and method for bond-selective imaging using vibrational relaxation encoded fluorescence

Abstract

A vibrationally enhanced fluorescence microscopy system includes a narrowband pulsed mid-infrared laser configured to excite selected chemical bond vibrations in a sample. A continuous-wave visible laser is coaligned with the mid-infrared laser and configured to induce fluorescence from at least one fluorescent reporter bound to the sample. An external pulse generator temporally synchronizes emission of the mid-infrared pulses with emission of the visible laser. A reflective objective is arranged to deliver the mid-infrared pulses to a focal plane within the sample. A water-immersion objective is coaxially aligned with the reflective objective and arranged to collect the fluorescence from the focal plane. A photon-counting detector is coupled to receive the collected fluorescence. Control electronics are configured to output a vibrational-contrast signal derived from counts produced by the photon-counting detector.

Description

SYSTEM AND METHOD FOR BOND-SELECTIVE IMAGING USING VIBRATIONAL RELAXATION ENCODED FLUORESCENCETECHNICAL FIELD

[0001] This disclosure relates to vibrational spectroscopy and imaging technologies and, in particular, to methods and systems for vibrational spectroscopy and imaging technologies using vibrational relaxation encoded fluorescence.BACKGROUND

[0002] Fluorescence microscopy has revolutionized life science and biomedical research by enabling the visualization of cells, organelles, and biomolecules with high specificity and spatial resolution. Targeted fluorescent probes can track biological processes in real time, establishing fluorescence microscopy as the most widely used optical imaging tool in biology and medicine. Single molecule sensitivity and versatility have made fluorescence indispensable in studies ranging from basic cell biology to clinical diagnostics. Despite these strengths, fluorescence microscopy lacks chemical information, limiting functional analysis such as metabolic activity, antibiotic response, disease progression, and organellar heterogeneity.SUMMARY

[0003] According to a first aspect, a vibrationally enhanced fluorescence microscopy system is provided. A narrowband pulsed mid-infrared laser is configured to excite selected chemical bond vibrations in a sample. A continuous-wave visible laser is coaligned with the mid-infrared laser and configured to induce fluorescence from at least one fluorescent reporter bound to the sample. An external pulse generator temporally synchronizes emission of the mid-infrared pulses with emission of the visible laser. A reflective objective is arranged to deliver the mid-infrared pulses to a focal plane within the sample. A waterimmersion objective is coaxially aligned with the reflective objective and arranged to collect the fluorescence from the focal plane. A photon-counting detector is coupled to receive the collected fluorescence. Control electronics are configured to output a vibrational-contrast signal derived from counts produced by the photon-counting detector.

[0004] In some exemplary embodiments, the system further comprises a pair of galvanometer mirrors arranged to synchronously scan the coaligned mid-infrared and visible beams across the sample.

[0005] In some exemplary embodiments, the system further comprises a quad-band dichroic mirror and at least one optical filter positioned to separate the fluorescence from excitation light before the fluorescence reaches the photon-counting detector.

[0006] In some exemplary embodiments, the system further comprises an acousto-optic modulator positioned in the path of the continuous -wave visible laser and driven to intensity-modulate the visible laser at a reference frequency.

[0007] In some exemplary embodiments, the photon-counting detector is a single-photon avalanche diode operatively connected to a digitizer that records photon arrival times with a temporal resolution of 150 picoseconds or finer.

[0008] In some exemplary embodiments, the narrowband pulsed mid-infrared laser is tunable, and the control electronics are further configured to acquire photon-count data at a plurality of mid-infrared wavenumbers.

[0009] In some exemplary embodiments, the plurality of mid-infrared wavenumbers lie within a fingerprint region from 1500 cm1to 1780 cm ■,

[0010] According to a second aspect, a method for single-molecule vibrational spectroscopy is provided. The method comprises illuminating a fluorescent reporter attached to a target molecule with a narrowband pulsed mid-infrared beam to resonantly excite a chosen molecular vibration; simultaneously illuminating the fluorescent reporter with a continuous-wave visible beam whose timing is synchronized to the mid-infrared pulses; detecting fluorescence emitted by the fluorescent reporter in response to the combined illumination; and determining presence of the chosen molecular vibration from a magnitude of the detected fluorescence.

[0011] In some exemplary embodiments, the method further comprises differentiating photon counts acquired during successive cycles in which the mid-infrared beam is alternatelypresent and absent; and using the count difference as the magnitude in step of determining presence of the chosen molecular vibration.

[0012] In some exemplary embodiments, the method further comprises synchronously scanning the mid-infrared and visible beams across a sample by means of galvanometer mirrors; collecting fluorescence at each scan position; and constructing from the collected fluorescence a vibrational-contrast image having sub-micron spatial resolution.

[0013] According to a third aspect, a photon-counting acquisition method for vibrationally enhanced fluorescence is provided. The method comprises directing fluorescence, emitted following concurrent exposure of a sample to synchronized mid-infrared pulses and visible excitation, onto a single-photon detector; recording arrival times of photons detected during first time windows in which a mid-infrared pulse is incident on the sample and during second time windows in which no mid-infrared pulse is incident on the sample; and extracting vibrational-contrast data by differential counting of photons recorded in the first and second time windows.BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figs. 1(a), 1(b). and 1(c) include schematic illustrations schematically illustrating theory, principle and spectral reliability of vibration relaxation encoded fluorescence (VREF), according to some exemplary embodiments.

[0015] Figs. 2(a), 2(b), and 2(c) include schematic illustrations schematically illustrating the VREF microscope and spectral reliability, according to some exemplary embodiments.

[0016] Figs. 3(a), 3(b), 3(c), 3(d), 3(e), and 3(f) include schematic illustrations schematically illustrating hyperspectral VREF imaging of live cells labelled with Nile Red, according to some exemplary embodiments.

[0017] Figs. 4(a), 4(b), 4(c), 4(d), 4(e), 4(f), 4(g), 4(h), 4(i), and 4(j) include schematic illustrations schematically illustrating hyperspectral widefield VREF imaging of E. coli and 5. aureus bacteria labelled with Nile Red and dried on CaF2 coverslip, according to some exemplary embodiments.

[0018] Figs. 5(a), 5(b), 5(c), and 5(d) include schematic diagrams schematically illustrating bacterial response to different concentrations of antibiotic.

[0019] Figs. 6(a), 6(b), and 6(c) include schematic illustrations schematically illustrating hyperspectral wide-field VREF imaging of dried E. coli and S. aureus, according to some exemplary embodiments.

[0020] Figs. 7(a) and 7(b) include graphs illustrating fluorescence counts comparison of both 5. aureus and E. coli bacteria using VREF and F-MIP, according to some exemplary embodiments.

[0021] Fig. 8 includes schematic illustrations schematically illustrating comparison of VREF and FRET mechanisms, according to some exemplary embodiments.DETAILED DESCRIPTION

[0022] To address the limitations of fluorescence microscopy, vibrational spectroscopic imaging techniques based on infrared (IR) absorption and Raman scattering have been developed to directly probe chemical bonds in molecules. While these approaches enable molecular fingerprinting and have been applied in studies of drug response, fungal infections, metabolism, and cancer identification, their weak signals, limited sensitivity, and tradeoffs between speed and spatial resolution restrict broader application. Nonlinear vibrational microscopy overcomes this challenge by providing nonlinear optical or thermal readouts of molecular vibrations. In particular, recently developed Mid-IR photothermal (MIP) microscopy combines the large absorption cross-sections of mid-IR with pump-probe detection to achieve submicron resolution and live-cell compatibility. Furthermore, fluorescence-detected MIP (F-MIP) takes advantage of the thermal sensitivity of fluorescence quantum yield to enhance modulation depth by over two orders of magnitude, enabling high-contrast organelle-level chemical imaging with thermal-sensitive dyes.

[0023] Conventional fluorescence microscopy enables versatile visualization of cells and organelles but lacks the ability to show molecular or chemical content for functional analysis. According to the present technology disclosed and described in detail herein, to address thislimitation, Vibrational Relaxation Encoded Fluorescence (VREF) microscopy is provided. VREF is a photothermal chemical imaging approach coding vibrational selectivity into fluorescence. In this approach, a laser excites molecular vibrations within cells or organelles, and subsequent vibrational relaxation into heat raises the local temperature. This thermal effect alters the Boltzmann equilibrium of nearby reporters, resulting in thermally activated fluorescence. In contrast to recent methods that suppress fluorescence quantum yield through thermally accelerated dynamic quenching, VREF utilizes an anti-Stokes excitation scheme, leading to a positive modulation of fluorescence emission. This method offers broad compatibility with common, thermal insensitive dyes. Applications to mammalian cells and bacteria demonstrate its potential for high-quality functional imaging with reduced background noise. Importantly, VREF allows detection of subtle biochemical and metabolic changes in bacteria exposed to antibiotics, including concentrations below the minimum inhibitory concentration. These findings establish VREF as a unique tool for live-cell imaging, functional single-cell analysis, and characterization of antibiotic resistance.

[0024] According to the cunent disclosure, the current technology is referred to herein as VREF microscopy, which is a new photothermal chemical imaging approach that encodes vibrational selectivity into fluorescence emission while ensuring broad dye compatibility. The underlying principle of VREF is rooted in the Boltzmann distribution of vibrational states. At thermal equilibrium, a fraction of molecules in the ground electronic state occupy higher vibrational levels, with a population that decreases exponentially as energy increases. The ratio of populations between the excited and ground vibrational states is determined by the Boltzmann equilibrium, establishing a statistical connection between vibrational populations and temperature.

[0025] The concept of exciting a system with lower-energy photons and detecting higher-energy emission has been recognized in spectroscopic methods such as anti-Stokes fluorescence and coherent anti-Stokes Raman spectroscopy. However, the VERF microscopy disclosed herein is different from stimulated Raman excited fluorescence or fluorescence-encoded infrared spectroscopy, in which up-converting fluorescence is used to retrieve vibrational information from a fluorescent molecule. Instead, VREF microscopy employs an anti-Stokes excitation scheme to generate thermally activated fluorescence and retrieve information from molecules surrounding the fluorescent reporter. Unlike F-MIP that is restricted to thermally sensitive dyes and constrained by photobleaching as the modulationrelies on excited fluorophores, the VREF approach enhances the photon yield via off-resonance excitation, creating a positive contrast while reducing photobleaching. Importantly, it extends compatibility to a wider range of common dyes that are otherwise thermal insensitive. This approach enables chemically resolved imaging of live cells, organelles, and bacteria, expanding fluorescence microscopy into the domain of functional chemical analysis with direct relevance to cellular physiology.

[0026] The principle of VREF of the current disclosure can be understood in relation to F-MIP, which couples IR absoiption to fluorescence via the photothermal effect. F-MIP imaging couples IR absorption to fluorescence emission by photothermal modulation of fluorescence quantum yield, thereby revealing the chemical signatures of biomolecules. Figs.2(a), 1(b), and 1(c) schematically illustrate theory, principle and spectral reliability of vibration relaxation encoded fluorescence (VREF), according to some exemplary embodiments. Specifically, Fig. 1(a) schematically illustrates the working principle of VREF and F-MIP. Referring to Fig. 1(a), the target molecule is excited by IR and relaxes in a non-radiative way, transferring heat to the nearby sensor molecule. In F-MIP, the sensor molecule undergoes ground-state excitation, while in VREF, the sensor molecule undergoes excited-state excitation. S: Stokes fluorescence; AS: anti-Stokes fluorescence. Referring to Fig. 1(b), fluorescence excitation records the entire IR-induced photothermal signal in F-MIP and only signal at the “hot” state in VREF, thus reducing the photobleaching. Fig. 1(c) is a graph illustrating calculations based on Boltzmann distribution highlighting the exponential population ratio and modulation depth sensitivity per Kelvin.

[0027] As can be seen in Fig. 1(a), the working principle behind F-MIP relies on mid-IR photons which excite vibrational modes of target molecules, and the subsequent relaxation generates localized heating. This local heating enhances the nonradiative decay pathways of nearby fluorescent probes, reducing their quantum yield through thermally induced nonradiative relaxation. The increase in temperature intensifies the dynamic quenching process of fluorophores, resulting in reduced fluorescence emission. This effect was implemented by exciting the fluorescent molecules with a continuous- wave visible laser to record fluorescence changes correlated with molecular vibrational absorption features.

[0028] VREF according to the current technology follows a similar excitation-detection scheme but with an anti-Stokes mechanism, where thermally populated vibrational states areexcited by the probe beam, leading to enhanced fluorescence emission. As illustrated in Fig.1(a), the process begins with thermal activation to promotes a fraction of molecules in the ground electronic state into higher vibrational levels of the sensor molecule, as determined by the Boltzmann distribution. Then, a visible probe excites these thermally populated states into the electronic excited state. As a result, the emission includes conventional Stokes fluorescence at longer wavelengths as well as anti-Stokes fluorescence at shorter wavelengths, thus encoding vibrational population information into the fluorescence signal. This mechanism enables chemical imaging with broad compatibility across dyes, including thermal insensitive ones.

[0029] While both F-MIP and VREF rely on modulation by the pulsed IR beam, their fluorescence intensity differs in contrast as shown in Fig. 1(b). Whereas a negative modulation is produced in F-MIP, VREF produces fluorescence only when the IR is on (“hot” state), while negligible emission occurs when the IR is off (“cold” state), resulting in a positive modulation. Technically, chopping the probe into pulses could reduce the overall excitation dose, thereby mitigating photobleaching while preserving signal fidelity.

[0030] The temperature-dependent population ratio between excited and ground vibrational states, N2 / N , follows the Boltzmann distribution:where N2 / N denotes the ratio of populations in the excited and ground vibrational states, respectively; AE is the energy difference between these states (in electronvolts, eV); ks is the Boltzmann constant (8.617xlO-5eV / K); and T is the absolute temperature in kelvin (K). This ratio decreases exponentially with the vibrational energy gap and determines how thermal activation modulates the available population for excitation.

[0031] Fig. 1(c) illustrates this relationship, showing how population ratios vary with transition energy AE and temperature T, as well as the corresponding sensitivity per K. F-MIP exhibits approximately 1 % per K modulation of fluorescence intensity. For VREF, assuming a transition energy of 0.2 eV corresponding to 1613.1 cm'1and room temperature of 300 K,the calculated change in population corresponds to a modulation depth of ~3% per K (Fig.1(c)). This relatively large modulation ensures the high sensitivity of VREF for vibrational imaging.

[0032] VREF microscope and spectral fidelity

[0033] Based on the above principle, a scanning VREF microscope is provided according to the current technology. Figs. 2(a), 2(b), and 2(c) include schematic diagrams schematically illustrating the VREF microscope and spectral reliability, according to some exemplary embodiments. Fig. 2(a) includes a schematic functional block diagram illustrating a point scanning VREF microscope, according to some exemplary embodiments. In Fig. 2(a), GM: galvo mirrors; CM: concave mirrors; DM: dichroic mirrors; SiPM: silicon photomultiplier. Fig. 2(b) is a graph illustrating averaged VREF spectra of dimethyl sulfoxide (DMSO) supplemented with various fluorescent dyes. FTIR spectrum was used for comparison. Fig.2(c) is a graph illustrating averaged VREF spectra of different supplements with Nile Red dye representing the different biological spectral features. The errors in the curves were calculated as the standard deviation of three independent measurements at each wavenumber and presented as shadow areas.

[0034] In one exemplary demonstrative embodiment, the microscope integrated a tunable mid-IR quantum cascade laser (QCL) (IR Laser) for vibrational excitation and a visible laser (Probe Laser) for fluorescence excitation, enabling synchronized counter-propagating dualbeam excitation. The fluorescence collected in epi-detection mode was filtered and detected by a SiPM. Ilyperspectral imaging was performed by tuning the QCL across the fingerprint region. Scanning is performed by two synchronized galvo pairs. The scanning VREF microscope setup incorporated mid-IR excitation together with fluorescence detection. A pulsed mid-IR pump beam was generated by the tunable QCL (for example, MIRcat-2400), Daylight Solutions). Fluorescence excitation was provided by an acousto-optic modulator (AOM) modulated 532 nm continuous-wave laser (Probe Laser) (for example, Samba, Cobolt) or a 632.8 nm He-Ne laser (for example, 05-LHP, Melies Griot), depending on the fluorophore. The IR pulse width was set to 500 ns at a repetition rate of 200 kHz, while the visible excitation operated with a 30% duty cycle at a 400 kHz modulation frequency.Fluorescence excitation scanning was achieved using a pair of dual-axis galvanometer mirrors (GM1, GM1) (for example, GVS002, Thorlabs). The scanned beam passedsequentially through a scan lens (for example, f = 100 mm, AC508-100-A, Thorlabs) and a tube lens (for example, f = 200 mm, TTL200-A, Thorlabs) before being reflected by a dichroic mirror (DM) into a 60x water-immersion objective (for example, UPlanSApo, Olympus, NA = 1.2), which focused the beam onto the sample. The IR beam was scanned independently by a second pair of galvanometer mirrors (GM3, GM4) (for example, GVS002, Thorlabs). Its optical path included a concave scan lens (for example, f = 200 mm, CM508-200-P01, Thorlabs) and a tube lens (for example, f = 500 mm, CM508-500-P01, Thorlabs) to project the beam into the back pupil of a reflective objective (for example, PIKE, 40x, NA = 0.78), arranged in counter-propagation with the visible beam.

[0035] Prior to imaging, the IR focus was carefully aligned to overlap with the visible focus. During acquisition, both beams were synchronously scanned to ensure uniform excitation and detection across the field of view. Synchronization between the two galvanometer pairs was calibrated according to the focal lengths of the objectives and adjusted for the beam expansion ratio of the relay optics. Fluorescence emitted backward from the sample was collected by the water-immersion objective, transmitted through the DM, and spectrally filtered (Filterl) before detection with a silicon photomultiplier (for example, SiPM, Cl 3366-3050GA, Hamamatsu). The electrical signal was recorded using a MokmPro system (for example, Liquid Instruments, Multi-Instrument Mode). For hyperspectral imaging, the QCL was operated in multi-spectral mode with a preset scanning list, covering the fingerprint region (1000-1800 cm’1) in 126 frames.

[0036] To validate the spectral fidelity of VREF according to the current technology, the 1000 to 1500 cm'1region was scanned. Two thermal sensitive dyes (for example, FITC, Nile Red) and a thermal insensitive dye (for example, FAM) dissolved in DMSO were used as testbeds. As shown in Fig. 2(b), the obtained VREF spectra for all dyes accurately reproduced the characteristic absorption bands of DMSO, with excellent agreement in both peak position and width. Notably, since the dye concentration (100 jiM) was far lower than that of the solvent, the presence of the fluorophores did not distort the spectra. For validation, the corresponding FTIR spectmm of pure DMSO is included, demonstrating that VREF reliably captures the same vibrational features as FTIR spectroscopy.

[0037] To demonstrate that VREF can capture vibrational spectra of surrounding biomolecules, VREF spectra were acquired in the IR fingerprint region from differentbiomolecular compounds stained with Nile Red (Fig. 2(c)). These samples were selected as chemical markers of major biomolecular classes, including Bovine Serum Albumin (BSA) to represent proteins, DNA for nucleic acids, glycogen for carbohydrates, and glyceryl trioleate (TAG) for lipids. Each compound exhibits distinct spectral features at characteristic wavenumbers that correspond closely to known IR vibrational bands, such as amide modes in proteins, phosphate vibrations in DNA, and C-0 or C-H vibrational modes in carbohydrates and lipids. Together, the agreement of these spectral signatures with conventional IR assignments demonstrates both molecular specificity and spectral fidelity of VREF spectroscopy of the current technology in capturing chemically meaningful vibrational fingerprints.

[0038] Scanning VREF imaging of live HeLa cells

[0039] Based on the spectral validation, the VREF system of the current disclosure was further applied to image live cell samples beyond model compounds to evaluate its performance in a biologically relevant context. This is demonstrated by probing intracellular vibrational signatures through fluorescence signal, the method enables visualization of molecular distributions in living systems. Live HeLa cells were stained with Nile Red and subsequently imaged with VERF microscopy of the disclosure to acquire hyperspectral images across the mid-IR fingerprint region (1000-1800 cm1). A probe laser at 632.8 nm was used to excite fluorophores from thermally populated higher vibrational states, thereby generating both normal and anti-Stokes fluorescence. A pixel integration time of 30 ps was used, enabling rapid data acquisition while maintaining a sufficiently signal-to-noise ratio (SNR) to resolve vibrationally specific features at the single-cell level.

[0040] Figs. 3(a), 3(b), 3(c), 3(d), 3(e), and 3(f) include schematic diagrams schematically illustrating hyperspectral VREF imaging of live cells labelled with Nile Red, according to some exemplary embodiments. Fig. 3(a) illustrates fluorescence images of live HeLa cells acquired with the scanning system at 1649 cm1and at 1797 cm1(Fig. 3(b)) Fig. 3(c) illustrates corresponding VREF images at 1649 cm1and at 1797 cm1(Fig. 3(d), with three ROIs selected from the nucleus and three from the cytoplasm of the same cell. Fig. 3(e) includes a graph illustrating normalized VREF spectra from the nucleus (lower curve), and cytoplasm (upper curve). Errors are calculated as standard deviations and shown as shaded areas. Fig. 3(f) includes a graph illustrating ratios of Phosphate / Amide I for cytoplasm and nucleus, respectively (t-test: ****p < 0.0001). Scale bar: 10 pm.

[0041] Figs. 3(a) and 3(b) display the conventional fluorescence images of a single living HeLa cell obtained with the IR pump tuned to 1648 cm1(on-resonance) and to 1797 cm1(off-resonance). In both cases, the fluorescence intensity excited at 632.8 nm appears nearly identical, highlighting that the probe dye itself does not provide direct vibrational contrast. In contrast, the corresponding VREF images shown in Figs. 3(c) and 3(d) illustrate a striking difference: at 1649 cm the cell can be clearly visualized, demonstrating vibrationally encoded fluorescence, whereas at 1797 cm1the signal disappears, consistent with the absence of absorption band at this off-resonance wavenumber. This on / off resonance behavior confirms that VREF selectively reports vibrational absorption and directly links the fluorescence modulation to the underlying molecular vibrational fingerprint.

[0042] To demonstrate intracellular VREF spectroscopy of the current disclosure, regions of interest (ROIs) in Fig. 3(c) were selected to compare nuclear and cytoplasmic signals within the same cells. Three ROIs corresponding to the nucleus (blue circles — lower right) and three corresponding to the cytoplasm (red circles-up and to the left from blue circles) were chosen for each cell. Fig. 3(e) presents the averaged spectra from seven individual HeLa cells, where for each cell, three nuclear and three cytoplasmic ROIs were selected. The extracted spectra were subsequently processed by baseline correction, power normalization, and spectral normalization to the highest peak, followed by offset adjustment to facilitate comparison. The resulting vibrational spectra clearly reveal distinct chemical contributions to nuclear and cytoplasmic regions, enabling direct ratio-metric analysis of key chemical groups, most notably the Amide I band (1649 cm1) and the nucleic acid-associated phosphate band (1080 cm1).

[0043] To quantify these differences, the ratio of 21 selected ROIs — three from the nucleus and three from the cytoplasm of each of the seven cells — is presented in Fig. 3(f). The phosphate (1080 cnr1) to Anri de I (1649 cm1) intensity ratio was calculated for each ROI, providing a direct metric to compare protein with nucleic acid content between subcellular compartments. The results show that nuclear regions consistently yield higher phosphate (nucleic acid) / Amide I (protein) ratios compared to cytoplasmic regions, reflecting the phosphate-enriched material in the nucleus. This quantitative ratio metric analysis demonstrates the ability of VREF to resolve chemical heterogeneity at the subcellular scale, enabling functional insights beyond qualitative imaging.

[0044] Widefield VREF microscopy and imaging of bacterial species

[0045] Figs. 4(a), 4(b), 4(c), 4(d), 4(e), 4(f), 4(g), 4(h), 4(i), and 4(j) include schematic illustrations schematically illustrating hyperspectral widefield VREF imaging of E. coli and .S', aureus bacteria labelled with Nile Red and dried on CaF2 coverslip, according to some exemplary embodiments. Fig. 4(a) includes a schematic functional diagram of the wide-field VREF microscope. Fig. 4(b) includes wide-field fluorescence images of bacterial cells. Fig.4(c) includes VREF images acquired at the Amide I - on-resonance (1648 cm ' Fig. 4(d) includes VREF images acquired off-resonance (1756 cm ). Fig. 4(e) includes a graph illustrating average fluorescence intensity comparison between hot (IR excitation on) and cold (IR excitation off) frames from 12 bacterial cells from each of 5. aureus and E. coli. Fig.4(f) illustrates normalized VREF spectra of S. aureus (upper curve at wavenumber 1100 cm ' ) and E. coli (lower curve at wavenumber 1100 cm (i obtained from the same 12 cells. Fig. 4(g) illustrates phosphate / Amide I ratio of both bacteria (t-test: ****P < 0.0001). Fig.4(h) illustrates modulation depth of 10 .S', aureus samples measured by VREF (upper curve) and F-M1P (lower curve). Fig. 4(i) illustrates modulation depth of 10 E. coli samples measured by VREF (upper curve) and F-MIP (lower curve). Fig. 4(j) illustrates photobleaching percentage of the same 10 samples of each bacterium measured by VREF (first and third curves from the left) and F-MIP (second and fourth curves from the left) (t-test: ****P < 0.0001). Errors are shown as standard deviations (shaded areas in spectra, error bars in plots). Scale bars: 10 pm.

[0046] By pulsed excitation and camera-based detection, widefield VREF enhances spectral fidelity in hyperspectral imaging by averaging a larger number of IR-pump / fluorescence-probes cycles within single frames, providing field-of- view-wide consistency for quantitative comparisons. The optical arrangement of the widefield system is shown in Fig. 4(a) and described in detail below.

[0047] The widefield VREF microscope employed synchronized mid-IR excitation with fluorescence detection. Mid-IR pulses were generated by a QCL (for example, MIRcat, Daylight Solution) and focused onto the sample with a parabolic mirror (for example, f = 15 mm, MPD00M9-M01, Thorlabs) in transmission geometry. A visible probe laser at either 520 nm (for example, NPL52C, Thorlabs, 129 ns) or 638 nm (for example, 06-MLD, Cobolt, electrically modulated to 200 ns pulses) was coupled through a fiber de-Speckler andcollimation optics (for example, L5-L2) and then relayed to the back focal plane of a 50x objective (for example, Nikon, NA = 0.8). A DM directed the probe beam to the sample while transmitting the epi-fluorescence to pass. The emitted fluorescence was collected by the same objective, passed through a bandpass filter to reject scattered excitation light, and imaged onto an sCMOS camera (for example, Andor Zyla 5.5).

[0048] System synchronization was provided by a digital pulse generator (for example, Emerald 9254-TZ50-US, Quantum Composers), which supplied a 50 kHz master clock to externally trigger the QCL, probe laser, and camera. The camera operated with exposures of tens of milliseconds, integrating thousands of pulse cycles. This virtual lock-in camera scheme enabled pixel-wise demodulation of IR modulated fluorescence intensity, yielding F-MIP / VREF images by tuning QCL wavelength across the mid-IR fingerprint region, hyperspectral VREF imaging provided molecularly specific contrast in both bacterial and cellular samples.

[0049] As a focused application, the widefield VREF system was harnessed to examine bacteria dried on a CaF2 substrate. Nile Red labeled bacterial cells were imaged across the mid-IR fingerprint region (900-1800 cm1). Similar to the scanning system, a probe laser at 638 nm was employed to excite fluorophores from thermally populated higher vibrational states, enabling the generation of both normal and anti-Stokes fluorescence.

[0050] Fig. 4(b) presents the widefield fluorescence images of Staphylococcus aureus (S. aureus) and Escherichia coli (E. coli), which serve as reference signals for the subsequent vibrationally encoded fluorescence analysis. The widefield VREF images of S. aureus and E. coli are presented at two IR excitation wavenumbers: 1648 cml, corresponding to the Amide I band and representing the on-resonance state (Fig. 4(c)), and at 1756 cman off-resonance state outside major absorption feature (Fig. 4(d)). The VREF images reveal strong bacterial contrast at 1648 crn1and the contrast disappears at 1756 cm-1. This clear on / off resonance behavior demonstrates the specificity of VREF for vibrational absorption signatures and highlights its capability to resolve functional group associated bands in bacterial samples. Importantly, the approach is not limited to Nile Red; comparable VREF signals were also observed with other common fluorophores, including Cy2, FITC, and Rhodamine 6G (R6G), as shown in Figs. 6(a), 6(b), and 6(c), which include schematic illustrations schematically illustrating hyperspectral wide-field VREF imaging of dried E. coli and .S', aureus, accordingto some exemplary embodiments. VREF images were acquired at the Amide I on- resonance (1648 crn1) and off-resonance (1756 crn ') conditions for both bacterial species labeled with Cy2 (Fig. 6(a)), FITC (Fig. 6(b)), and R6G (Fig. 6(c)). Scale bars: 10 pm.

[0051] Fig. 4(e) presents the fluorescence counts of both .S', aureus and E. coli under hot and cold frames across the entire fingerprint region. As shown, the fluorescence intensity decrease gradually over time, and in both species the hot signal remains consistently higher than the cold signal. This behavior confirms the positive contrast of VREF, in which measurable modulation occurs only when the IR pump is resonant and synchronized with the probe laser. The plotted curves represent an average of 12 individual bacterial cells for each species. The VREF spectra was obtained by subtracting the cold curve from the hot curve. For comparison, similar plots for F-MIP are presented in Figs. 7(a) and 7(b), which include graphs illustrating fluorescence counts comparison of both .S', aureus (Fig. 7(a)) and E. coli (Fig. 7(b)) bacteria using VREF and F-MIP, according to some exemplary embodiments. Figs. 7(a) and 7(b) illustrate fluorescence counts recorded under hot and cold frames across the fingerprint region (900-1800 cmVREF were acquired using ~3 mW laser power, while F-MIP were acquired using ~5 mW laser power. Both hot and cold conditions are shown for comparison.

[0052] Fig. 4(f) shows the VREF spectra in the fingerprint region, revealing distinct vibrational features for the two bacterial types, after IR normalization and weighted leastsquares normalization. Notably, clear differences are observed in the nucleic acid region, consistent with the structural differences between Gram-positive .S', aureus and Gramnegative E. coli. The thicker peptidoglycan layer of S. aureus contains more carbohydrate components, leading to higher relative intensity in this spectral region and providing a biochemical basis for species differentiation. For quantitative analysis to differentiate between the two bacterial species, the intensity ratio of the phosphate (1080 cm1) to Amide I (1648 cm1) was calculated for the same 12 bacterial cells of each type. The results, presented in Fig. 4(g), show a clear separation between 5. aureus and E. coli. This difference is consistent with the spectral features observed in the VREF spectra (Fig. 4(f)) and confirms that ratio metric analysis of VREF signals can provide a reliable metric for distinguishing bacterial species based on their biochemical composition.

[0053] Comparison of VREF and F-MIP modalities

[0054] Using the widefield system, the performance of VREF was evaluated against F-MIP by analyzing modulation depth and photostability, using a set of 10 bacterial cells for each species. The VREF and F-MIP signals were obtained by fluorescence excitation of the same Nile Red dye at 638 nm and 520 nm, respectively. The modulation depth, defined as hwt icoid^ |s ngative in F-MIP due the thermal reduction of fluorescence quantumJ coldyield. In contrast, VREF yields a positive contrast since fluorescence arises only when the IR pump is resonant and synchronized with the probe laser. As shown in Figs. 4(h) and 4(i), the VREF modulation depth for Nile Red was ~7-fold higher in S. aureus and ~3-fold higher in E. coli than F-MIP, demonstrating its superior signal contrast.

[0055] Fig. 4(j) compares photobleaching percentages for the same set of 10 bacterial cells. In VREF, the bleaching average percentages were 34.6% for 5. aureus and 28.1% for E. coli, corresponding to photostability values of 65.4% and 71.9%, respectively. By contrast, F-MIP showed greater photobleaching — 83.9% in S. aureus and 75.1% in E. coli — leaving photostability at 16.1% and 24.9%, respectively. Thus, VREF improves photostability by approximately 4-fold in S. aureus and 3-fold in E. coli, highlighting a major practical advantage of the technique for prolonged hyperspectral imaging.

[0056] VREF assessment of bacterial response and antibiotic susceptibility

[0057] The growing misuse of antibiotics has accelerated the emergence of multidrugresistant bacteria, posing a major global health challenge. Conventional susceptibility tests, typically require up to 48 hours, delaying appropriate treatment and encouraging empirical use of broad-spectrum antibiotics. Rapid and accurate determination of bacterial response to antibiotics is therefore critical for guiding timely therapy, reducing resistance development, and improving patient outcomes. Furthermore, since bacteria exhibit significant cell-to-cell heterogeneity, assessing single bacterial response to antibiotics is crucial.

[0058] Figs. 5(a), 5(b), 5(c), and 5(d) include schematic diagrams schematically illustrating bacterial response to different concentrations of antibiotic. Fig. 5(a) illustrates MIC measurement of erythromycin against .S'. aureus. Optical density at 600 nm (ODeoo) was measured after 2 h of incubation, and bacterial growth percentage was calculated from colony counts after 24 h, across five erythromycin concentrations (0.06, 0.12, 0.25, 0.50, and 1.0pg / mL) and a control (no antibiotic). Fig. 5(b) illustrates average VREF spectra (n = 10 single cells) of 5. aureus under the same five concentrations and control. Fig. 5(c) illustrates quantification of VREF peak intensities corresponding to Amide I (1648 cnr1), Amide II (1548 crn1). and Phosphate (1080 cm (i from the same 10 samples, across all concentrations and control. Fig. 5(d) illustrates ratios of Phosphate / Amide I and Phosphate / Amide II, calculated from the peaks shown in Fig. 5(c) for all concentrations and control. Errors bars are shown as standard deviations (box plots).

[0059] Hyperspectral VREF imaging was applied to rapidly assess bacterial response to antibiotics at single cell level. To investigate the antibiotic response, .S'. aureus cultures were exposed to erythromycin at a range of concentrations, then labeled with Nile Red for spectroscopic analysis. In parallel, the minimum inhibitory concentration (MIC) of erythromycin against 5. aureus was determined following the standard broth dilution method. The MIC was defined as the lowest concentration at which no visible bacterial growth occurred after 24 hours of antibiotic exposure, which in the experiments / demonstrations was 0.25 pg / mL. Growth inhibition was quantified by measuring optical density at 600 nm (ODeoo) and by counting colony-forming units (CFUs), with growth percentages normalized to untreated controls (Fig. 5(a)). As shown, ODeoo measurements closely matched the CFU-based growth percentages across concentrations, validating the consistency of the assay. A detailed description of the experiment is provided herein.

[0060] Using VREF, vibrational spectra of .S', aureus treated with a series of erythromycin concentrations (Fig. 5(b)) was recorded. The quantified intensities of Amide I (1648 cm1), Amide II (1548 cm1), and phosphate (1080 cm1) bands are shown in Fig. 5(c). With increasing antibiotic concentration, the Amide II band showed a clear gradual decrease, while the Amide I band decreased more modestly, both indicating reduced protein synthesis. In contrast, the phosphate increased progressively reflecting the buildup of untranslated RNA, and reached a plateau near the MIC (0.25 pg / mL). This is in close agreement with the ODeoo measurements and growth percentage shown in Fig. 5(a). To further quantify these spectral changes, we calculated two ratio metric markers of bacterial activity: phosphate I Amide I and phosphate I Amide II. Both ratios decreased steadily from the control to the MIC and then remained at similar levels up to 4 x MIC (highest concentration), reflecting suppressed bacterial replicative activity (Fig. 5(d)).

[0061] These results correlate well with ODeoo measurements and growth percentages obtained by traditional assays. Importantly, while standard AST requires at least 24 hours of incubation, VREF enabled detection of antibiotic-induced metabolic changes within only 2 hours. This demonstrates the potential of VREF as a rapid and sensitive approach for determining bacterial susceptibility to antibiotics.

[0062] The technology of the current disclosure demonstrates VREF as a versatile photothermal imaging approach that offers chemical specificity with the convenience of fluorescence detection. VREF achieves higher modulation depth, improved photostability, and broader dye compatibility compared to F-MIP, enabling vibrational imaging with both thermally sensitive and insensitive dyes.

[0063] The intcrmolccular heat transfer allows VREF to bridge the gap between fluorescence microscopy and vibrational spectroscopy. In this sense, VREF can be compared with Forster resonance energy transfer (FRE I ), a widely used fluorescence-based method for probing local molecular environments. In FRET, energy is transferred non-radiatively from a donor to an acceptor fluorophore, a process that requires spectral overlap and operates over nanoscale distances (<10 nm). In contrast, VREF encodes vibrational information into fluorescence through heat transfer from host molecules to surrounding dye reporters, eliminating the need for donor-acceptor pairs, with reference to Fig. 8, which includes schematic illustrations schematically illustrating comparison of VREF and FRET mechanisms, according to some exemplary embodiments.. In FRET, the energy transfer occurs between the electronically excited states, while in VREF, the heat transfer occurs between the electronic ground states. Like single molecule FRET, single molecule VREF is possible with picosecond pump / probe pulses. FRET is a non-radiative dipole-dipole coupling process in which an excited donor fluorophore transfers energy to a nearby acceptor fluorophore, dependent on spectral overlap and intermolecular distance (<10 nm). In contrast, VREF relies on IR excitation of molecular vibrations, where vibrational relaxation into localized heat enhances fluorescence emission of nearby dyes.

[0064] On the application side, the ability to resolve protein-to-nucleic acid ratios in nuclei and cytoplasm highlights its capacity to probe functional heterogeneity in living cells (Fig.3(d)). Species-level discrimination between Gram-positive and Gram-negative bacteriademonstrates its robustness in capturing biologically meaningful differences in molecular composition (Fig. 4(g)). Moreover, VREF is able to detect antibiotic-induced metabolic alterations within two hours, a timescale far shorter than conventional AST, underscoring its potential as a rapid diagnostic tool (Fig. 5(d)). While further optimization of imaging throughput and in vivo compatibility can be accomplished, the present technology of the disclosure demonstrate that VREF extends the scope of fluorescence microscopy into functional chemical analysis, opening new opportunities for studying dynamic biological processes and accelerating antibiotic susceptibility testing.

[0065] As described herein in detail, VREF microscopy is a new platform that encodes vibrational spectroscopy into fluorescence through infrared pump, heat transfer, and antiStokes fluorescence probe. By converting vibrational population dynamics into fluorescence intensity modulation, VREF achieves broad dye compatibility, enhanced modulation depth, and improved photostability compared to fluorescence-detected MIP microscopy. High spectral fidelity in both solution-phase and biomolecular model systems waw demonstrated, subcellular chemical heterogeneity in live HeLa cells was validated, and Gram-positive was distinguished from Gram-negative bacteria based on their biochemical composition.Furthermore, VREF enables rapid assessment of antibiotic response in S. aureus, detecting metabolic changes within only 2 hours. Together, these results highlight VREF as a chemical imaging approach that extends the reach of fluorescence microscopy into functional vibrational analysis, offering new opportunities for live-cell biology, diagnostics, and rapid antibiotic susceptibility testing.

[0066] Data Processing

[0067] For hyperspectral scanning VREF imaging, the Self-Permutation Noise2Noise Denoiser (SPEND) algorithm was applied to enhance the quality of the datasets and suppressing frame-to-frame noise and enhancing spectral fidelity. After SPEND processing, the spectra were baseline-corrected and subsequently normalized by the IR power to ensure accurate recovery of molecular absorption features. In the case of widefield VREF imaging, the output spectra were likewise normalized by the IR power and subsequently processed using a weighted least-squares fitting model with total variation (TV) and Ll-bias regularization, efficiently solved with the Fast Iterative Shrinkage-Thresholding Algorithm (FISTA), to minimize noise and refine spectral resolution. Finally, all VREF spectra wereuniformly normalized to their highest peak intensity and offset-corrected for consistent comparison across datasets.

[0068] Cell Culture and Staining

[0069] HeLa cells were cultured on C'al;2 slide at a density of 1 x 105cells / mL in 2 mL of high-glucose DMEM supplemented with 10% fetal bovine serum (FBS) and penicillinstreptomycin, then incubated for 24 hours at 37 °C in a humidified incubator with 5% CO2. Following this initial incubation, the growth medium was replaced with fresh serum-free DMEM containing Nile Red at a final concentration of 400 ng / mL. The cells were then incubated under the same conditions for 30 minutes to allow dye uptake. Following staining, the samples were gently washed three times with pre-warmed phosphate-buffered saline (PBS) to remove excess unbound dye. For imaging, the Cal ’2 slide was mounted with a coverslip in PBS, and fluorescence was recorded using the scanning VREF microscope.

[0070] Bacterial Culture and Staining

[0071] Both bacterial strains used in this technology, Escherichia coli (E. coli) BW25113 and Staphylococcus aureus (S. aureus) ATCC6538, were treated using the same procedure.Frozen stocks stored at -80 °C were streaked onto Mueller-Hinton agar (MHA) plates and incubated overnight at 37 °C. A single colony was then picked and inoculated into 3 mL of Mueller-Hinton broth (MHB), followed by incubation at 37 °C for 2 hours to reach the logarithmic growth phase.

[0072] Bacterial cells were purified by centrifugation (5000 x g, 5 minutes), washed three times with PBS, and fixed in 4% formalin solution for 30 minutes at room temperature. After fixation, Nile Red or Cy2 (final concentration: 1 x 104M) was added directly to the bacterial pellet. The pellet was resuspended in the dye solution and incubated for 30 minutes in the dark. Following staining, bacterial cells were washed three times with sterile water to remove excess dye. A 5 pL aliquot of the bacterial suspension was then deposited onto a CaF2 slide and allowed to air-dry for 10-15 minutes at room temperature to promote bacterial adhesion ready for imaging.

[0073] The MIC of erythromycin against .S', aureus was determined using the broth microdilution method. Log-phase cultures were adjusted to ~5 x 105CFU per mL and inoculated into 96-well plates containing serial dilutions of erythromycin in MHB. Plateswere incubated at 37 °C for 24 hours, and bacterial growth was assessed by measuring ODeoo. The MIC was defined as the lowest antibiotic concentration yielding no visible growth. To verify results, serial dilutions from each well were plated on MHA in triplicate, and CFU were counted to calculate growth percentage relative to untreated controls.

[0074] DEFINITIONS

[0075] Vibrational Relaxation Encoded Fluorescence (VREF):A technique in which vibrational energy absorbed by a molecule is rapidly redistributed via intramolecular vibrational redistribution (IVR), resulting in a vibrationally “hot” ground state. Subsequent visible excitation of a fluorescent reporter leads to anti-Stokes fluorescence, the intensity of which encodes the vibrational absorption signature.

[0076] Mid-Infrared (MIR) Excitation:Irradiation of a sample with mid-infrared laser pulses, typically in the 1000-1800 cm1range, to selectively excite molecular vibrations.

[0077] IR-tag:A chemical group within a molecule that exhibits strong absorption in the MIR region and serves as the site for selective vibrational excitation.

[0078] Fluorescent Reporter:A dye or fluorophore attached to the molecule of interest, capable of absorbing visible light and emitting fluorescence. In VREF, the reporter’s emission is modulated by vibrational excitation.

[0079] Anti-Stokes Fluorescence:Emission of photons at a shorter wavelength (higher energy) than the excitation light, resulting from excitation of a vibrationally “hot” ground state.

[0080] Intramolecular Vibrational Redistribution (IVR):The rapid dispersal of vibrational energy throughout a molecule following MIR absorption, leading to a thermalized ground state.

[0081] Photon-Counting Detector:A device capable of detecting and counting individual photons with high temporal resolution, such as a single -photon avalanche diode (SPAD) or silicon photomultiplier (SiPM).

[0082] Galvanometer Mirrors:Motorized mirrors used to scan laser beams across a sample.

[0083] Dichroic Mirror:An optical component that selectively transmits or reflects light based on wavelength, used to separate fluorescence from excitation light.

[0084] Acousto-Optic Modulator (AOM):A device that modulates the intensity of a laser beam using sound waves, enabling synchronized excitation.

[0085] Vibrational-Contrast Signal:The differential fluorescence signal obtained by comparing photon counts during MIR “hot” and “cold” states, reflecting the presence of specific molecular vibrations.

[0086] Fingerprint Region:The MIR spectral range (approximately 1000-1800 crn ' i containing characteristic vibrational modes of functional groups.

[0087] Single-Molecule Sensitivity:The ability to detect signals originating from individual molecules.

[0088] Sub-Micron Spatial Resolution:The capability to resolve features smaller than one micron in size.

[0089] Hot and Cold States:Time windows during which the sample is exposed to MIR excitation (“hot”) or not (“cold”), used for differential signal extraction.

[0090] Counter / Digitizer:An electronic device that records photon arrival times with high temporal resolution.

[0091] Reflective Objective Lens:An optical element for focusing MIR excitation onto the sample.

[0092] Water Immersion Objective Lens:An optical element for collecting fluorescence emission with high numerical aperture.

[0093] Quad-Band Optical Filter:A filter that transmits specific wavelength bands while blocking others, used to isolate fluorescence.

Claims

CLAIMSWhat is claimed herein is:

1. A vibrationally enhanced fluorescence microscopy system, comprising:a narrowband pulsed mid-infrared laser configured to excite selected chemical bond vibrations in a sample:a continuous-wave visible laser coaligned with the mid-infrared laser and configured to induce fluorescence from at least one fluorescent reporter bound to the sample;an external pulse generator that temporally synchronizes emission of the mid-infrared pulses with emission of the visible laser;a reflective objective arranged to deliver the mid-infrared pulses to a focal plane within the sample;a water-immersion objective coaxially aligned with the reflective objective and arranged to collect the fluorescence from the focal plane;a photon-counting detector coupled to receive the collected fluorescence; andcontrol electronics configured to output a vibrational-contrast signal derived from counts produced by the photon-counting detector.

2. The system of claim 1 , further comprising:a pair of galvanometer mirrors arranged to synchronously scan the coaligned mid-infrared and visible beams across the sample.

3. The system of claim 1, further comprising:a quad-band dichroic mirror and at least one optical filter positioned to separate the fluorescence from excitation light before the fluorescence reaches the photon-counting detector.

4. The system of claim 1, further comprising:an acousto-optic modulator positioned in the path of the continuous-wave visible laser and driven to intensity-modulate the visible laser at a reference frequency.

5. The system of claim 1, wherein:the photon-counting detector is a single-photon avalanche diode operatively connected to adigitizer that records photon arrival times with a temporal resolution of 150 picoseconds or finer.

6. The system of claim 1, wherein:the narrowband pulsed mid-infrared laser is tunable; andthe control electronics are further configured to acquire photon-count data at a plurality of mid-infrared wavenumbers.

7. The system of claim 6, wherein:the plurality of mid-infrared wavenumbers lie within a fingerprint region from 1500 cm1to 1780 cm1.

8. A method for single-molecule vibrational spectroscopy comprising: illuminating a fluorescent reporter attached to a target molecule with a narrowband pulsed mid-infrared beam to resonantly excite a chosen molecular vibration;simultaneously illuminating the fluorescent reporter with a continuous-wave visible beam whose timing is synchronized to the mid-infrared pulses;detecting fluorescence emitted by the fluorescent reporter in response to the combined illumination; anddetermining presence of the chosen molecular vibration from a magnitude of the detected fluorescence.

9. The method of claim 8, further comprising:differentiating photon counts acquired during successive cycles in which the mid-infrared beam is alternately present and absent; andusing the count difference as the magnitude in step of determining presence of the chosen molecular vibration.

10. The method of claim 8, further comprising:synchronously scanning the mid-infrared and visible beams across a sample by means of galvanometer mirrors;collecting fluorescence at each scan position; andconstructing from the collected fluorescence a vibrational-contrast image having sub-micron spatial resolution.

11. A photon-counting acquisition method for vibrationally enhanced fluorescence comprising:directing fluorescence, emitted following concurrent exposure of a sample to synchronized mid-infrared pulses and visible excitation, onto a single-photon detector;recording arrival times of photons detected during first time windows in which a mid-infrared pulse is incident on the sample and during second time windows in which no mid-infrared pulse is incident on the sample; andextracting vibrational-contrast data by differential counting of photons recorded in the first and second time windows.

Images