# Interferometric scattering microscopy
Interferometric scattering microscopy (iSCAT) is a label-free optical technique that detects unlabeled nanoparticles and single molecules by letting their weakly scattered light interfere with a reference beam of the same light. Because the interference signal scales linearly with particle polarizability (how easily a particle's electrons shift in light) rather than with the sixth power of particle size, iSCAT reaches single-molecule sensitivity and supports three main uses: mass determination of single biomolecules, high-speed single-particle tracking, and imaging of nanoscale structures without fluorescent labels.[1](https://www.nature.com/articles/s43586-025-00391-1)
| Key fact | Value |
|---|---|
| Contrast mechanism | Interference of scattered light with a reference field; signal linear in polarizability and particle volume[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) |
| Smallest label-free protein detected | 9 kDa at SNR ≈ 1.4 with machine-learning analysis; ~40 kDa at SNR ≈ 3 with conventional processing[3](https://www.nature.com/articles/s41592-023-01778-2) |
| Mass photometry performance | 2% sequence-mass accuracy, 19 kDa resolution, 1 kDa precision[4](https://www.science.org/doi/10.1126/science.aar5839) |
| Live-cell tracking | 3 nm average localization error at 2–40 kHz frame rates with 40 nm gold labels[5](https://discovery.ucl.ac.uk/id/eprint/10052649/7/091736.full.pdf) |
| Fastest demonstrated imaging | Up to about 1 MHz frame rate for small gold nanoparticles[6](https://arxiv.org/html/1812.10765) |
| Noise floor | Shot-noise limited; SNR scales as \( S \times N^{1/2} \) with detected photon number \( N \)[7](https://www.annualreviews.org/content/journals/10.1146/annurev-physchem-050317-021247) |
## How it works
In the common reflection geometry, the light reflected from the sample interface serves as the reference field \( E_{\mathrm{r}} \), and the field scattered by the particle is \( E_{\mathrm{s}} \). The detector records the superposition,[8](https://pubs.acs.org/doi/pdf/10.1021/nl500234t)
\[ I_{\mathrm{det}} = |E_{\mathrm{r}} + E_{\mathrm{s}}|^2 = |E_{\mathrm{i}}|^2 \left( r^2 + |s|^2 + 2 r |s| \sin\varphi \right), \]
where \( r \) is the interface reflectivity, \( s \) the scattering amplitude, and \( \varphi \) the relative phase. For small scatterers the pure scattering term \( |s|^2 \) is negligible and the cross-term \( 2 r |s| \sin\varphi \) dominates, so the measured contrast scales linearly with polarizability and particle volume rather than with the sixth power of size.[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) This linear scaling is the whole reason single molecules are visible: pure scattering drops with the sixth power of diameter, so a 5 nm particle scatters one million times less light than a 50 nm particle, which is why small particles vanish below background in dark-field methods.[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) • [9](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.93.037401)
Because the scattering cross section of a protein such as albumin is as small as \( 10^{-19} \) cm² at visible wavelengths, sensitivity is set by the signal-to-noise ratio, which is shot-noise limited and proportional to \( \sqrt{N} \); it can be improved with higher illumination power or longer exposure within photodamage limits.[1](https://www.nature.com/articles/s43586-025-00391-1) • [7](https://www.annualreviews.org/content/journals/10.1146/annurev-physchem-050317-021247)
## How it is done
A typical setup uses linearly polarized continuous-wave illumination (a 532 nm laser in one implementation, a 445 nm laser in another; LEDs also suffice when the particle-reference distance is within the coherence length), wide-field illumination of a few micrometers in diameter, and a camera in the image plane.[10](https://iopscience.iop.org/article/10.1088/0022-3727/49/27/274002) • [3](https://www.nature.com/articles/s41592-023-01778-2) • [6](https://arxiv.org/html/1812.10765) A back-focal-plane mask attenuating the reflected light by a factor of 100 or more matches the strong reference to the weak scattering signal on the camera's dynamic range without changing the shot-noise-limited SNR.[7](https://www.annualreviews.org/content/journals/10.1146/annurev-physchem-050317-021247) Cameras run from about 1.7 kHz for a 104 × 104 pixel field to 5–15 kHz with 20 μs exposures for single proteins, and up to roughly 1 MHz for small gold labels.[8](https://pubs.acs.org/doi/pdf/10.1021/nl500234t) • [3](https://www.nature.com/articles/s41592-023-01778-2) • [6](https://arxiv.org/html/1812.10765)
Background removal is the central processing step. The static speckle-like pattern of the clean substrate is removed by differential subtraction of consecutive frame batches, temporal median subtraction, iterative estimation, or rolling-window averaging; a pixelwise ratiometric frame computed with a rolling window highlights landing molecules as spots whose fitted 2D Gaussian amplitude gives the iSCAT contrast.[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) • [11](https://beta.iopscience.iop.org/article/10.1088/1361-6463/ac2f68) • [12](https://www.cell.physiol.med.uni-muenchen.de/studium_lehre/laboratory-courses/v2b_manual.pdf) The Python package PiSCAT implements this analysis pipeline, and calibration ladders of proteins of known mass convert contrast to mass, analogous to gel electrophoresis.[11](https://beta.iopscience.iop.org/article/10.1088/1361-6463/ac2f68) • [3](https://www.nature.com/articles/s41592-023-01778-2)
Sample requirements are demanding: coverslips cleaned by bath sonication and blow-drying or UV/ozone treatment, passivation (for example PEG-PLL with BSA blocking), and low sample concentrations, typically 5–50 nM for mass photometry, so that single molecules land sparsely on the surface.[8](https://pubs.acs.org/doi/pdf/10.1021/nl500234t) • [12](https://www.cell.physiol.med.uni-muenchen.de/studium_lehre/laboratory-courses/v2b_manual.pdf)
## Origin
Interference had been used as a contrast mechanism long before iSCAT: phase-contrast microscopy was an early explicit example, followed by differential interference contrast in the 1950s.[7](https://www.annualreviews.org/content/journals/10.1146/annurev-physchem-050317-021247) Earlier single-particle work included detection of single dye molecules through interference between their scattering and a residual reflection of the excitation beam, room-temperature photothermal detection of small gold particles, and lock-in assisted transmitted-light detection of gold particles down to 10 nm.[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) The method was introduced by K. Lindfors and colleagues in 2004 in [Physical Review Letters](https://www.edgepedia.org/physical-review-letters), combining confocal microscopy, supercontinuum illumination, and an interferometric detection scheme to identify single gold nanoparticles of diameter below 10 nm and record a single particle's plasmon resonance.[9](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.93.037401) In 2006, Filipp V. Ignatovich and [Lukas Novotny](https://www.edgepedia.org/lukas-novotny) reported background-free real-time interferometric detection of single viruses and 5 nm gold particles in a microfluidic channel with an external reference beam and 1 ms time resolution.[13](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.96.013901) The name iSCAT entered the literature with a study of the diffusional dynamics of individual viruses on supported lipid bilayers.[7](https://www.annualreviews.org/content/journals/10.1146/annurev-physchem-050317-021247) Single-protein detection was reported for myosin 5a HMM (502 kDa, expected contrast about 0.15%) and proteins down to BSA at 66 kDa.[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) • [8](https://pubs.acs.org/doi/pdf/10.1021/nl500234t)
## Variants
Because contrast is proportional to polarizability, and polarizability is proportional to particle volume at constant density, the iSCAT signal of a protein is linearly proportional to its molecular mass; refractive index and specific volume vary by only about 0.3% and 1.2% across more than \( 10^{5} \) protein sequences.[11](https://beta.iopscience.iop.org/article/10.1088/1361-6463/ac2f68) • [7](https://www.annualreviews.org/content/journals/10.1146/annurev-physchem-050317-021247) This readout, mass photometry, was reported by Gavin Young and colleagues in Science in 2018 as interferometric scattering mass spectrometry (iSCAMS), quantifying single biomolecule mass in solution with 2% sequence-mass accuracy, up to 19 kDa resolution, and 1 kDa precision, and resolving oligomeric distributions and small-molecule binding; proteins as small as 20 kDa were measured accurately.[4](https://www.science.org/doi/10.1126/science.aar5839) [Mass photometry](https://www.edgepedia.org/mass-photometry) has been commercialized (Refeyn Ltd., Oxford, UK) and is used to study self-assembly, interaction strengths, and oligomerization.[3](https://www.nature.com/articles/s41592-023-01778-2) • [1](https://www.nature.com/articles/s43586-025-00391-1)
A family of related interferometric techniques differs mainly in geometry and readout: interference reflectance imaging sensing (IRIS), rotating coherent scattering (ROCS), interference plasmonic imaging (iPM), coherent bright-field imaging (COBRI), stroboscopic interference scattering imaging (stroboSCAT), and iSCAMS.[6](https://arxiv.org/html/1812.10765) Amplified iSCAT (a-iSCAT), reported by Matz Liebel, James T. Hugall, and Niek F. van Hulst in 2017 in Nano Letters, boosts the signal for ultrasensitive label-free sensing and high-speed tracking of single proteins.[14](https://doi.org/10.1021/acs.nanolett.6b05040) Combining mass readout with tracking on supported lipid bilayers produced mass-sensitive particle tracking (MSPT), reported by Tamara Heermann and colleagues in 2021 in [Nature Methods](https://www.edgepedia.org/nature-methods) to follow the membrane-associated MinDE reaction cycle, and a mass photometry implementation for label-free tracking and mass measurement of single proteins on lipid bilayers reported by Eric D. B. Foley and colleagues the same year.[15](https://doi.org/10.1038/s41592-021-01260-x) • [16](https://doi.org/10.1038/s41592-021-01261-w)
## Applications
On live cell membranes, 20 nm gold nanoparticles have been tracked in 3D at 50 μs temporal and about 5 nm spatial resolution; with 40 nm gold-nanoparticle-labeled membrane proteins, average localization error of 3 nm was achieved at 0.5 ms exposure and 2–40 kHz frame rates, revealing GPI-GFP confinement events lasting up to 50 ms in regions a few tens of nanometers across with a 180–200 nm periodicity consistent with actin-spectrin rings.[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) • [5](https://discovery.ucl.ac.uk/id/eprint/10052649/7/091736.full.pdf) Label-free imaging has covered lipid nanodomains as small as 50 nm, microtubule assembly and disassembly, viral capsid self-assembly around an RNA scaffold, bacterial pili dynamics, and protein oligomerization.[10](https://iopscience.iop.org/article/10.1088/0022-3727/49/27/274002) • [6](https://arxiv.org/html/1812.10765) • [1](https://www.nature.com/articles/s43586-025-00391-1)
## Limitations and alternatives
Sensitivity has improved in steps: about 66 kDa in the first single-protein experiments, then roughly 40 kDa, the lowest value reported in the literature before machine learning.[11](https://beta.iopscience.iop.org/article/10.1088/1361-6463/ac2f68) • [3](https://www.nature.com/articles/s41592-023-01778-2) • [2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) Self-supervised machine learning (an [isolation forest](https://www.edgepedia.org/isolation-forest) for anomaly detection with FastDVDNet) pushed detection by a factor of 4 to 9 kDa at SNR ≈ 1.4, where conventional processing reaches unity SNR for about 15 kDa proteins after about 2 s of integration.[3](https://www.nature.com/articles/s41592-023-01778-2) • [11](https://beta.iopscience.iop.org/article/10.1088/1361-6463/ac2f68)
The main failure modes are technical. [Shot noise](https://www.edgepedia.org/shot-noise) is the intrinsic limit, but in practice speckle-like background from surface roughness and adsorbates, readout noise, laser intensity fluctuations, and phase noise from stage drifts dominate; contrasts below about \( 10^{-3} \) require referencing or normalization, and balanced photodiode pairs reach stability on the order of \( 10^{-7} \).[11](https://beta.iopscience.iop.org/article/10.1088/1361-6463/ac2f68) • [2](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/) In cells, background fluctuations from cell material reach 1–2% RMS versus about 0.3% RMS shot noise, limiting the use of smaller labels.[5](https://discovery.ucl.ac.uk/id/eprint/10052649/7/091736.full.pdf) On rough substrates, background heterogeneities coupled with tiny stage movements produce features that traditional computer vision algorithms misidentify as particles; a mask R-CNN trained with experimental backgrounds markedly reduced such false positives.[17](https://pubs.acs.org/doi/full/10.1021/acs.jpcb.3c00097)
Against dark-field microscopy, the decisive difference is scaling: iSCAT's detected signal depends linearly on the scattered-field amplitude, or equivalently on polarizability and particle volume, whereas dark-field relies on pure scattering intensity, which is proportional to the scattering cross section and therefore falls with the sixth power of diameter for small Rayleigh particles; this linear dependence is the main reason for iSCAT's higher sensitivity to very small particles.[18](https://www.biorxiv.org/content/10.1101/2024.06.18.599482v1) • [9](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.93.037401) Compared with labeled fluorescence and TIRF microscopy, iSCAT is label-free, and combined iSCAT-fluorescence protocols exist, but it demands clean, well-defined surfaces and low concentrations, and in cells it is constrained by background fluctuations from cell material.[19](https://doi.org/10.1038/nprot.2016.022) • [5](https://discovery.ucl.ac.uk/id/eprint/10052649/7/091736.full.pdf)
Developments since 2023 include machine-learning sensitivity gains below 10 kDa,[3](https://www.nature.com/articles/s41592-023-01778-2) defocus-integration iSCAT, which suppresses speckle from sub-nanometer substrate undulations and improves SNR by 5.4 dB for dielectric and 6.9 dB for gold nanoparticles without hardware change,[20](https://arxiv.org/html/2402.16527) and interferometric image scanning microscopy (iISM), which reaches about 120 nm lateral resolution inside live cells at roughly tenfold lower illumination power per spot (~0.5 μW), visualizing endoplasmic reticulum, actin, mitochondria, and vesicles at essentially unlimited observation times.[21](https://link.springer.com/article/10.1038/s41377-026-02210-y)
## References
1. [Interferometric scattering microscopy, Nature Reviews Methods Primers (2025)](https://www.nature.com/articles/s43586-025-00391-1)
2. [Taylor & Sandoghdar, Interferometric Scattering Microscopy: Seeing Single Nanoparticles and Molecules via Rayleigh Scattering (Nano Lett. 2019 review)](https://pmc.ncbi.nlm.nih.gov/articles/PMC6750867/)
3. [Self-supervised machine learning pushes the sensitivity limit in label-free detection of single proteins below 10 kDa (Nature Methods 2023)](https://www.nature.com/articles/s41592-023-01778-2)
4. [Young et al., Quantitative mass imaging of single biological macromolecules (Science 2018)](https://www.science.org/doi/10.1126/science.aar5839)
5. [Revealing compartmentalised membrane diffusion in living cells with interferometric scattering microscopy (UCL repository copy)](https://discovery.ucl.ac.uk/id/eprint/10052649/7/091736.full.pdf)
6. [Taylor & Sandoghdar, iSCAT Microscopy & Related Techniques (arXiv review)](https://arxiv.org/html/1812.10765)
7. [Interferometric Scattering Microscopy (Annual Review of Physical Chemistry)](https://www.annualreviews.org/content/journals/10.1146/annurev-physchem-050317-021247)
8. [Ortega Arroyo et al., Label-Free, All-Optical Detection, Imaging, and Tracking of a Single Protein (Nano Lett. 2014)](https://pubs.acs.org/doi/pdf/10.1021/nl500234t)
9. [Detection and Spectroscopy of Gold Nanoparticles Using Supercontinuum White Light Confocal Microscopy (Lindfors et al., PRL 93, 037401, 2004)](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.93.037401)
10. [Visualization of lipids and proteins at high spatial and temporal resolution via iSCAT microscopy (J. Phys. D 2016, Spindler et al.)](https://iopscience.iop.org/article/10.1088/0022-3727/49/27/274002)
11. [Optimized analysis for sensitive detection and analysis of single proteins via interferometric scattering microscopy (J. Phys. D 2022)](https://beta.iopscience.iop.org/article/10.1088/1361-6463/ac2f68)
12. [Sizing Proteins with Mass Photometry (LMU laboratory course manual)](https://www.cell.physiol.med.uni-muenchen.de/studium_lehre/laboratory-courses/v2b_manual.pdf)
13. [Ignatovich & Novotny, Real-Time and Background-Free Detection of Nanoscale Particles (PRL 96, 013901, 2006)](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.96.013901)
14. [Matz Liebel, James T. Hugall, Niek F. van Hulst (2017). Ultrasensitive Label-Free Nanosensing and High-Speed Tracking of Single Proteins. Nano Letters.](https://doi.org/10.1021/acs.nanolett.6b05040)
15. [Tamara Heermann and colleagues (2021). Mass-sensitive particle tracking to elucidate the membrane-associated MinDE reaction cycle. Nature Methods.](https://doi.org/10.1038/s41592-021-01260-x)
16. [Eric D. B. Foley and colleagues (2021). Mass photometry enables label-free tracking and mass measurement of single proteins on lipid bilayers. Nature Methods.](https://doi.org/10.1038/s41592-021-01261-w)
17. [Enhancing Nanoparticle Detection in iSCAT Microscopy Using a Mask R-CNN (J. Phys. Chem. B)](https://pubs.acs.org/doi/full/10.1021/acs.jpcb.3c00097)
18. [Homogeneous large field-of-view and compact iSCAT-TIRF setup for dynamic single molecule measurements (bioRxiv 2024)](https://www.biorxiv.org/content/10.1101/2024.06.18.599482v1)
19. [Jaime Ortega Arroyo, Daniel Cole, Philipp Kukura (2016). Interferometric scattering microscopy and its combination with single-molecule fluorescence imaging. Nature Protocols.](https://doi.org/10.1038/nprot.2016.022)
20. [Defocus-integration interferometric scattering microscopy for speckle suppression and enhancing nanoparticle detection on substrate (arXiv / Opt. Lett. 2024)](https://arxiv.org/html/2402.16527)
21. [Interferometric Image Scanning Microscopy for label-free imaging at 120 nm lateral resolution inside live cells (Light: Science & Applications 2026)](https://link.springer.com/article/10.1038/s41377-026-02210-y)
---
*Topic: Encyclopedia › Life and health › Biological foundations*
*Initially written Sep 29, 2026 · Reviewed: Sep 30, 2026 · Edited: Sep 30, 2026 · Last review: Sep 30, 2026*
*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*
License: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license