Custom Picosecond IR Laser Enables Routine 3 µm Matrix-Free MSI of Tissue
A custom 2.94 µm picosecond laser now enables routine 3 µm-pixel mass spectrometry imaging of fresh-frozen and formalin-fixed tissue without matrix deposition, researchers report in Nature Communications.
The Tech TL;DR:
- Sub-cellular molecular mapping achieved at 3 µm pixels using ambient LA-REIMS, eliminating matrix crystallization artifacts.
- Low-fluence, high-repetition-rate IR ablation preserves metabolite and phospholipid ion yield at diffraction-limited spot sizes.
- Platform validated on fresh-frozen sections and FFPE biopsies, correlating with fluorescence histology in single acquisitions.
Breaking the Matrix Bottleneck
Mass spectrometry imaging has long forced a trade-off: shrink the pixel size and you lose ion yield; keep the signal and you accept matrix-assisted laser desorption ionization (MALDI) preparation steps that blur sub-cellular detail.
The study published 20 November in Nature Communications demonstrates that a near-diffraction-limited 2.94 µm optical parametric amplifier (OPA) — pulse width ~30 ps, high repetition rate — coupled to a rapid evaporative ionization (REIMS) interface sidesteps both compromises. The laser produces ablation craters of 5 µm to 10 µm diameter at low fluence, feeding a commercial time-of-flight analyzer that captures metabolites, phospholipids, and small molecules in a single run.
Laser Architecture and Ion Yield
The team benchmarked three infrared sources: a CO2 laser (10.6 µm, ~100 µs pulse), an optical parametric oscillator (2.94 µm, 8 ns), and the OPA (3.0 µm, ~30 ps). Only the picosecond OPA delivered the beam quality and thermal confinement necessary for few-micrometer pixels without carbonization.
According to the Nature Communications paper, the short pulse duration limits thermal diffusion, suppressing the condensation reactions and neutral-particle dominance that have plagued laser desorption since the LAMMA era. The REIMS interface then ionizes the ablated plume at atmospheric pressure, yielding multiplexed ion maps that correlate with fluorescence and H&E histology.
Clinical Samples Without Extra Prep
Fresh-frozen sections and formalin-fixed, paraffin-embedded (FFPE) biopsies were imaged directly — no matrix deposition, no derivatization. Oversampled datasets preserved sub-cellular morphology, enabling molecular distributions to be overlaid on histological features. The authors note this “clinically compatible, fully ambient format” opens a path toward intra-operative MS-guided surgery where a single platform generates both histological and molecular data in real time.
Comparative Laser Performance
| Laser Source | Wavelength | Pulse Duration | Imaging Outcome |
|---|---|---|---|
| CO2 | 10.6 µm | ~100 µs | Ambient MSI feasible; limited spatial resolution |
| OPO | 2.94 µm | 8 ns | Improved resolution; thermal effects persist |
| OPA | 3.0 µm | ~30 ps | Routine 3 µm pixels; sub-cellular morphology preserved |
Implementation Reality Check
The platform relies on a custom OPA light source — not an off-the-shelf component — and a commercial TOF analyzer.
Nature Communications, the open-access journal from Nature Portfolio that has published continuously since 2010, released the study on 20 November. The journal operates from editorial offices in London, Berlin, New York City and Shanghai and, since January 2016, has made all content freely accessible after initially mixing subscription and open-access articles.
# Pseudo-code for oversampled LA-REIMS registration pipeline
import numpy as np
from skimage import registration, transform
def register_oversampled_msi(msi_cube, histology_ref, pixel_size_um=3, oversample_factor=2):
"""
Align oversampled MSI ion maps to histology reference.
msi_cube: (n_pixels_y, n_pixels_x, n_mz)
histology_ref: (H, W, 3) RGB histology tile
"""
# Sum ion images for structural channel
structural = np.sum(msi_cube, axis=2)
# Upsample to match histology resolution
target_shape = (histology_ref.shape[0], histology_ref.shape[1])
structural_up = transform.resize(structural, target_shape, order=3, preserve_range=True)
# Rigid + affine registration
tform = registration.estimate_transform('affine', structural_up, histology_ref[:,:,0])
registered_cube = np.stack([
transform.warp(msi_cube[:,:,i], tform.inverse, output_shape=target_shape)
for i in range(msi_cube.shape[2])
], axis=2)
return registered_cube, tform
Next Steps Focus on Beam Stability and Commercial Source
The work establishes a “robust operational regime” for matrix-free MSI at few-micrometer pixels, but the authors identify beam stability and repetition-rate scaling as the next engineering frontiers. Transitioning from a custom OPA to a commercially ruggedized picosecond IR source will determine whether the technique moves from demonstration to routine pathology workflow. The paper was published online 20 November 2026 in Nature Communications (doi:10.1038/s41467-026-79489-3).