Light sheet fluorescence microscopy (LSFM) is a well-established technique for volumetric imaging of large samples with high speed and good resolution and is compatible with tissue optical clearing techniques that enhance light penetration in highly scattering media and improve image quality by reducing optical aberrations. We built the di2CLSFM according to the optical scheme shown in panel A in the attached figure (photo in panel B), it works reliably and acquires new data daily. It has two identical optical pathways that alternately serve as fluorescence excitation and detection arms to capture two orthogonal views of a sample, obtaining in post-processing by computational fusion and deconvolution an improved almost isotropic micrometer resolution. We developed the image processing pipeline and will release it as open source. We optimized a tissue preparation protocol based on a refined SWITCH method to image samples with lateral size up to tens of centimeters and height up to 0.5 mm, greatly reducing the traditional need to cut smaller blocks. Up to four target neuronal populations can be rendered fluorescent through immuno-staining and then detected in different emission bands, one plane per image, when illuminated by the digitally scanned light sheet (DSLM), realized via a galvo mirror from a single beam coming from the matching laser. We developed a custom quartz holder to store and sequentially image multiple samples, by scanning them through the light sheet by moving the motorized precision three-axis translation stages in a serpentine pattern.
We leverage confocal line detection, for improved contrast and reduced background, by exploiting the rolling shutter of a fast sCMOS camera and have demonstrated a dual beam illumination scheme that doubles the imaging speed without affecting the image quality. We have also demonstrated two illumination schemes that suppress illumination artefacts and we have studied the effects of excitation light polarization on fluorescence emission in LSFM.
The acquired images are inspected with an advanced artificial intelligence-based (AI) image analysis algorithm to automatically recognize the complex biological content, discriminating the shapes of neurons and blood vessels from the background fluorescence. In the attached figure, panel C presents the algorithm’s image analysis steps, while panel D displays an example of a classified image and panel E of a whole tissue. Three large human brain regions with volume of several cubic centimetres were imaged with the di2CLSFM: a whole left hippocampus, part of the Broca’s area and of the motor cortex (examples of representative slices are shown in panels F, G and H). Their digital anatomical reconstruction is in progress and it will greatly advance the medical and neurobiological understanding of the healthy brain tissue structure. The study of tissues affected by FCD, unfortunately, was not realized due to the impact of the COVID-19 pandemic and related lock-down, making also impossible to do a comparative cytoarchitectonic investigation between healthy and dysplastic brain tissues. The obtained data was contributed to open repositories as a foundation for brain models.
These results have been presented to the scientific community at five conferences and through four publications and three conference proceedings, and more publications are in preparation. Moreover, we disseminated them among the general public through social media and at the Scienz’Estate 2019 and European Researchers’ Night BRIGHT 2020 outreach events.
