Recurrent neural network-based volumetric fluorescence microscopy

Luzhe Huang, Hanlong Chen, Yilin Luo, Yair Rivenson, Aydogan Ozcan

Light Science & Applications 2021 article cited by 41

Source Open-access PDF More in Physics & Astronomy

importance 5/10 solved Advanced Fluorescence Microscopy Techniques microscopy imaging stage: research

Problem

Volumetric fluorescence microscopy normally requires many axial scans of a sample, making 3D imaging slow. The paper reconstructs a full 3D volume from only a few sparsely captured 2D images.

Why it matters

Volumetric fluorescence imaging plays an important role across physical, medical and life sciences, so fewer scans means faster imaging.

What the AI did

A recurrent convolutional neural network (Recurrent-MZ) takes 2D images sparsely captured at arbitrary axial positions and digitally reconstructs the sample volume over an extended depth-of-field.

Results

30-fold reduction in axial scans for the same sample volume; significantly increased depth-of-field of a 63×/1.4NA objective (C. elegans and nanobead experiments)

Impact

Demonstrated on C. elegans and nanobeads that the same volume can be imaged with 30× fewer axial scans.

Record

Venue	Light Science & Applications
Published	2021
Work type	article
Authors	Luzhe Huang, Hanlong Chen, Yilin Luo, Yair Rivenson, Aydogan Ozcan
Citations	41 (OpenAlex, at import time)
References	67
Model or method	Recurrent-MZ (recurrent convolutional neural network)
Organization	not stated in source
Primary source	https://doi.org/10.1038/s41377-021-00506-9
Open access	gold
Keywords	deep learning, recurrent neural network, fluorescence microscopy, volumetric imaging, depth-of-field

Importance rating

5 / 10

real but narrower advance

Volumetric fluorescence imaging spans physical, medical and life sciences; 30-fold scan reduction speeds 3D imaging broadly, but abstract states no scale of impact.

Basis: inferred. Judged from the source text, which gives no explicit scale. Stated scale: not stated in source. Method and limits.

More in Physics & Astronomy

#	Paper	Year	Citations	Importance
1	Magnetic control of tokamak plasmas through deep reinforcement learning Jonas Degrave, Federico Felici, Jonas Buchli, Michael Neunert, Brendan...	2022	739	7
2	Solving the quantum many-body problem with artificial neural networks Giuseppe Carleo, Matthias Troyer	2017	2,205	7
3	Learning high-accuracy error decoding for quantum processors Johannes Bausch, Andrew Senior, Francisco J. H. Heras, Thomas Edlich,...	2024	90	7
4	Optimizing Quantum Error Correction Codes with Reinforcement Learning Hendrik Poulsen Nautrup, Nicolas Delfosse, Vedran Dunjko, Hans J. Brie...	2019	158	7
5	Machine-learning-assisted correction of correlated qubit errors in a topological code Paul Baireuther, Thomas E. O’Brien, Brian Tarasinski, C. W. J. Beenakk...	2018	141	7