Mid-infrared photothermal imaging flow cytometry

Abstract: Imaging flow cytometry (IFC) enables high-throughput single-cell analysis but largely relies on fluorescence labeling to obtain molecular specificity. Label-free vibrational imaging can provide intrinsic chemical contrast, yet coherent Raman-based methods interrogate only a limited axial volume, which restricts quantitative whole-cell analysis under flow. Mid-infrared photothermal (MIP) microscopy offers a promising route to overcome this limitation by combining linear mid-infrared (MIR) absorption-based chemical contrast with visible-light detection, allowing chemical imaging of a broader axial volume of each cell in a wide-field configuration. However, applying MIP microscopy to rapidly flowing cells has been difficult because conventional frame-sequential acquisition of MIR-ON and MIR-OFF images is highly susceptible to motion-induced subtraction artifacts. Here we demonstrate MIP-IFC, a label-free imaging flow cytometry platform based on single-shot nanosecond-dual-pulse MIP (SNAP-MIP) microscopy. SNAP-MIP encodes the MIR-ON and MIR-OFF states into separate holographic channels within a single camera exposure, reducing their temporal separation to 20 ns. This single-shot acquisition suppresses motion artifacts and increases the allowable sample velocity for artifact-free MIP imaging by five orders of magnitude compared with conventional frame-sequential MIP imaging. Leveraging this capability, MIP-IFC acquired chemical images at 500 frames per second and achieved a cellular event rate up to ~70 events s^-1. We demonstrate quantitative chemical discrimination of flowing microbeads and apply MIP-IFC to single-cell profiling of oleic-acid-induced lipid accumulation, adipocyte differentiation, and confluence-dependent cellular heterogeneity. These results establish MIP-IFC as a high-throughput, quantitative, label-free chemical imaging platform for single-cell phenotyping under flow.

• The paper presents SNAP-MIP microscopy, a label-free technique that overcomes motion artifacts in mid-IR imaging flow cytometry using nanosecond dual-pulse holography.
• It employs spatial-frequency multiplexed off-axis digital holography to separate MIR-ON/OFF channels, achieving high frame rates (500 fps) and reliable chemical discrimination.
• Experimental results include effective lipid profiling in cells and precise volumetric imaging, demonstrating enhanced throughput and artifact-free performance.

Mid-Infrared Photothermal Imaging Flow Cytometry: A Technical Review

Background and Motivation

Imaging flow cytometry (IFC) has become integral to high-throughput cellular phenotyping, enabling robust extraction of morphological and molecular features at single-cell resolution. Traditional IFC methods predominantly depend on fluorescence labeling for molecular specificity, which poses cytotoxicity risks and often introduces physiological perturbations. Label-free approaches, such as quantitative phase imaging (QPI), provide biophysical contrast but lack molecular discrimination. Vibrational microscopy modalities—particularly coherent Raman and mid-infrared (MIR) absorption—offer chemical specificity, yet suffer from confining signal generation to thin axial volumes, yielding incomplete cellular characterization under flow.

The principal challenge addressed in this paper is the development of a label-free IFC platform capable of both quantitative whole-cell chemical imaging and artifact suppression at high flow velocities.

SNAP-MIP Microscopy: Methodological Advancements

The authors introduce the single-shot nanosecond-dual-pulse MIP (SNAP-MIP) microscopy, which fundamentally reconfigures conventional MIR photothermal imaging for IFC applications. Traditional MIP-DH acquisition is challenged by motion-induced subtraction artifacts due to millisecond-scale separation between MIR-ON and MIR-OFF images. SNAP-MIP mitigates these artifacts by encoding the MIR-ON and MIR-OFF states into independent holographic channels with a 20 ns temporal separation, enabling near-simultaneous acquisition within a single sensor exposure.

Key technical features:

• Spatial-Frequency Multiplexed Off-Axis Digital Holography (DH): Encodes MIR-ON/OFF channels into distinct frequency bands within the hologram, allowing computational separation and independent reconstruction.
• Nanosecond Dual-Pulse Design: Synchronization of visible probe pulses before and after MIR excitation permits rapid interrogation of photothermal-induced refractive index (RI) changes.
• Robust Computational Refocusing: Utilizes DH to accommodate axial variations, extending volumetric imaging without mechanical adjustment and diminishing requirements for hydrodynamic focusing.
• High Frame Rates and Throughput: Achieves 500 fps and cellular event rates up to ~70 events/s, becoming competitive with fluorescence-based platforms.

Experimental Validation and Quantitative Analysis

The manuscript provides rigorous validation of artifact suppression:

• Motion Artifact Elimination: SNAP-MIP eliminates spatial mismatch artifacts observed in conventional MIP-DH at velocities up to 8 cm/s, thus increasing allowable flow speed by five orders of magnitude.
• Chemical Discrimination of Microbeads: Flowing PMMA and silica beads are reliably distinguished based on MIR-fluence-normalized MIP signal intensities, achieving signal separation consistent with theoretical refractive index predictions.
• Computational Refocusing Accuracy: Accurate phase and MIP contrast for beads is retained across axial positions, confirming robustness against sample displacement.

In biological contexts:

• Lipid Accumulation Profiling in COS-7 Cells: MIP-IFC discriminates lipid-rich regions induced by oleic acid treatment, yielding a ~6.8-fold increase in MIP signal density compared to untreated controls, whereas dry mass density varies modestly. Entropy analyses reveal increasingly concentrated intracellular lipid heterogeneity post-treatment.
• Single-Cell Adipocyte Differentiation Analysis: Differentiated 3T3-L1 cells exhibit broad heterogeneity in both cell size and lipid-rich area; MIP-IFC captures the emergence of large lipid storage structures and delineates size-dependent limits on lipid area fraction.
• Culture Confluence Profiling: As confluence approaches 100%, average MIP signal density and pixel-wise entropy change, reflecting increased lipid-associated content and population heterogeneity under high-density culture conditions.

Implications, Limitations, and Future Directions

SNAP-MIP-based MIP-IFC represents a functional advance in label-free chemical cytometry, providing high-throughput, volumetric, and quantitative molecular imaging under flow at velocities relevant to practical IFC. The system's noise resilience and artifact suppression fundamentally enable robust analysis of morphological and chemical phenotypes.

Theoretical implications concern single-cell biochemical profiling in contexts where fluorescent labels are undesirable or unreliable, such as primary immune cells or live-cell metabolic studies. Practically, this supports applications in functional cell-state profiling, differentiation monitoring, environmental adaptation, and diagnostic screening.

Future directions include:

• Throughput Enhancement: By adopting faster image sensors and increasing MIR pulse repetition rates (currently 1 kHz to potentially 10 kHz), throughput may approach 10,000 fps, substantially exceeding coherent Raman IFC benchmarks.
• Spectral Multiplexing: The spatial-frequency domain remains under-utilized; expansion to encode multiple MIR-ON states at different excitation wavelengths within a single frame enables simultaneous multispectral chemical imaging, facilitating multi-target molecular diagnostics.
• Broader Biological Applications: The demonstrated workflow for lipid-related phenotyping is readily extensible to other molecular targets (proteins, nucleic acids) as MIR excitation spectral engineering advances.

Conclusion

This study demonstrates a high-throughput, label-free IFC platform leveraging SNAP-MIP microscopy to achieve quantitative chemical and morphological single-cell profiling in flowing samples. By suppressing motion artifacts, extending volumetric imaging to the whole cell, and capturing chemical heterogeneity, the approach marks a significant advancement in vibrational cytometry. The methodology promises scalable, multispectral, and nondestructive phenotyping for functional and diagnostic applications in cellular biology and bioengineering (2606.19842).