A spatio-temporally adaptive microscope for the imaging of biological tissues – TemporalSmartScan

The confocal fluorescence microscope is an essential tool in biology. It generates volumetric images of living tissues with high spatial resolutions, by scanning the volume point by point with a laser beam. However, the technique is slow and the implied light dose is harmful for the imaged cells.
In many situations, in particular in the case of embryos and developing tissues, cells are organized in curved sheets (epithelia). Conventionally, such objects are imaged by scanning the entire volume plane by plane, which is highly inefficient in terms of photon budget.
In the TemporalSmartScan project, we develop a confocal microscope with acquisition strategy driven by real-time data processing algorithms, in order to retrieve the desired information for the object of interest (position, shape, movement, intensity distribution) with the minimum amount of light sent on the sample – and thus an increased acquisition speed. In order to adapt the algorithms and the instrumentation to the requirements of live imaging of dynamical processes, we take into account the spatio-temporal evolution of the object of interest, allowing to estimate the features of the object and to predict their evolution. We thus algorithmically control not only where to make acquisitions to improve the feature estimation accuracy, but also when.
Finally, we demonstrate the benefits of our approach on the imaging of live processes in Drosophila tissues, comparing the smart-scans with a systematic scan of the volume. We expect the smartscans to be much more gentle for the live processes due to the more than order of magnitude reduction in light dose.
The TemporalSmartScan project combines the highly complementary expertise of optical instrumentation development, statistical image and data processing and developmental biology and biomechanics. This combination is essential for the co-design of a microscope able to deal in real time with unsupervised, driven acquisition towards efficient imaging and to demonstrate its value on current biological questions.