FOCUS
Smart illumination for 3D-imaging of biological tissues
the projection; when full volumetric reconstruction is required, 3D-RIM remains the method of choice.
We could validate the instrument on biologically relevant thick samples, including mouse intestine tissue, Drosophila embryos, and human cultured cells, illustrating both robustness and live-imaging compatibility.
SMART SCANNING Camera-based 3D imaging offers speed and high resolution, providing gentler, more distributed illumination than point-scanning confocal systems. However, even with parallel acquisition, very sensitive tissues can be damaged. Background fluorescence also poses a fundamental limitation: the photon noise associated with background can overwhelm the speckle-based variations used for super-resolved reconstruction, undermining the capacity for super-resolution.
To overcome these issues, we have developed a smart microscope that focuses illumination only where necessary, placing the imaging process under algorithmic control and enabling structures of interest to be targeted progressively [ 7 ]. Real-time analysis learns from the measurements already acquired and chooses where to probe next, so the illumination converges toward relevant features such as curved embryonic surfaces( Fig. 4). When we have a strong prior knowledge of the structure of interest-as is the case with surfaces, for example- it may take as few as three steps: performing a very sparse pre-scan; estimating a spatial model of the object( in this case, a surface); and scanning the surface. In more complex situations, a longer iterative approach may be required to converge. In our initial study, we focused on cell sheets, which are curved 3D surfaces commonly found in embryos, but the concept is general.
In the future, the detection could also be trained through examples, using machine-learning and artificial intelligence approaches. Temporal continuity is also crucial: information from one time point could inform the next, allowing the acquisition to concentrate on changes and maximize new information per irradiation photon.
The benefits of smart scanning are first in the reduction of the light dose. Because the structures of interest often occupy only a small fraction of the total volume, the illuminated voxels – and thus the total light dose impinging on the sample – can be reduced by up to two orders of magnitude. Lower dose means less phototoxicity and less bleaching, enabling much longer observations of living tissues. At the same time, restricting illumination to the informative regions reduces background and its associated photon noise, which improves reconstruction quality. A last practical side effect is strong data compression: by measuring less but measuring smarter, we store and process only what matters.
PERSPECTIVES Using light more intelligently can improve contrast, resolution, speed, and sample health in 3D imaging. By placing excitation photons only where they are informative, we extend what
REFERENCES live fluorescence microscopy can do without harming the specimen.
As a perspective, crossmodality is the natural next step. We can pair 3D-RIM or EDF-RIM with adaptive illumination patterns that target regions of interest in real time. Cross modality with point scanning microscopy( non-linear point scanning is still the way to go at really great depth), and light sheet microscopy will also be explored.
In all cases, the aim is to maximize information per photon, minimize dose, and go faster when imaging thick tissues.
ACKNOWLEDGEMENTS These results are the outcome of a nation-wide interdisciplinary collaboration. We are especially grateful to Nicolas Bertaux, Simon Labouesse, Thomas Mangeat, Jérôme Idier, Benoît Rogez, Faris Abouakil, Marie- Anne Burcklen, Hervé Rigneault, Guillaume Giroussens. This project is funded by the " France 2030 " investment plan managed by the French National Research Agency( ANR-21-ESRE-0002), and from Excellence Initiative of Aix- Marseille University— A * MIDEX, as well as Agence Nationale de la Recherche ANR-22-CE13-0039, ANR- 22-CE42-0010, ANR-24-CE13-2163.
[ 1 ] A. M. Gest, A. Z. Sahan, Y. Zhong, W. Lin, S. Mehta and J. Zhang, Chemical Reviews 124( 22), 12573( 2024)
[ 2 ] M. G. Gustafsson, Journal of microscopy 198( 2), 82( 2000)
[ 3 ] T. Mangeat, L. Mazzella, B. Rogez, G. Giroussens, C. Bernard, P. Vargas, M. Allain, S. Labouesse, J. Idier, L. LeGoff and A. Sentenac,( 3D-RIM)( 2025). BioRxiv. DOI: 10.1101 / 2025.08.06.668879
[ 4 ] S. Labouesse, J. Idier, M. Allain, G. Giroussens, T. Mangeat and A. Sentenac, Physical Review A 109( 3), 033525( 2024)
[ 5 ] T. Mangeat, S. Labouesse, M. Allain, A. Negash, E. Martin, A. Guénolé and A. Sentenac, Cell Reports Methods 1( 1),( 2021)
[ 6 ] L. Mazzella, T. Mangeat, G. Giroussens, B. Rogez, H. Li, J. Creff, M. Saadaoui, C. Martins, R. Bouzignac, S. Labouesse, J. Idier, F. Galland, M. Allain, A. Sentenac and L. Legoff light: Science & applications 13( 1), 285( 2024)
[ 7 ] F. Abouakil, H. Meng, M. A. Burcklen, H. Rigneault, F. Galland and L. LeGoff, light: Science & applications 10( 1), 210( 2021)
40 www. photoniques. com I Photoniques 134