In the biological sciences, data from fluorescence and electron microscopy is correlated to allow fluorescence biomolecule identification within the cellular ultrastructure and/or ultrastructural analysis following live-cell imaging. High-accuracy (sub-100 nm) image overlay requires the addition of fiducial markers, which makes overlay accuracy dependent on the number of fiducials present in the region of interest. Here, we report an automated method for light-electron image overlay at high accuracy, i.e. below 5 nm. Our method relies on direct visualization of the electron beam position in the fluorescence detection channel using cathodoluminescence pointers. We show that image overlay using cathodoluminescence pointers corrects for image distortions, is independent of user interpretation, and does not require fiducials, allowing image correlation with molecular precision anywhere on a sample.

The SECOM platform enables accurate and straightforward correlative light and electron microscopy (CLEM) by integrating a fluorescence microscope in a scanning electron microscope (SEM) 1 . The SECOM platform can be fitted on a SEM by replacing the door to the vacuum chamber of the SEM. This replacement supports a motorized stage, the objective and the light path for the optical microscope. The integration of the two imaging modalities eliminates the need to switch between the optical and electron microscope and allows for faster imaging and a more logical workflow. More importantly, the SECOM platform makes it possible to correlate two different types of information on the exact same cell, tissue or structure of interest 2 .

Superresolution fluorescence has pushed the resolution of light microscopy (LM) towards length scales traditionally accessible only with electron microscopy (EM) [1]. Also, developments in scanning EM (SEM) have moved image dimensions for EM to typical LM fields-of-view [2] and into the third dimension [3]. By correlating data from both techniques [4], molecules can be localized within the context of cells and tissue and with reference to their live dynamics, but throughput and quantification are hindered by elaborate, expert procedures involving separate microscopes. We have developed an integrated approach with high-numerical aperture LM inside an SEM, such that the electron beam can be positioned anywhere within the fluorescence field of view [5,6]. Here, we will show that this approach allows for automated light-electron overlay, i.e. without fiducials or user data interpretation, with the same high accuracy anywhere on the sample [7].Our integrated microscope consists of an inverted fluorescence microscope with sample translation stage replacing the original sample stage in an SEM. Using vacuum compatible immersion oil, numerical aperture of the LM can be up to 1.4. The axes of both microscopes can be aligned to about 1 µm [5]. Samples containing fluorescence can be prepared either via standard chemical fixation procedures followed by post-embedding immuno-labelling, or using an adapted sample preparation protocol for maintaining genetic expressed fluorescence during EM sample preparation [4]. In all cases, integrated microscopy allows for rapid identification of regions of interest for SEM based on fluorescence expression and seamless exchange between both fluorescence and electron microscopy acquisition [6]. However, both modalities are still separated imaging systems, each with their own field distortions. Accurate correlation needs a registration procedure to determine both the relative position and orientation of both image fields, as well as to correct for microscope distortions.Image acquisition in the integrated microscope is schematically indicated in Fig. 1. For automated overlay, we exploit the fact that the focused electron beam generates (cathodo-)luminescence in the ITO-coated glass substrate that supports the sample. When the electron beam is scanned over several well-separated positions, this leads to an array of circular intensity spots, or pointers, on the LM camera ( Fig. 1-III). We obtain a discrete set of LM coordinates by fitting pointer centre positions for each pointer. These LM pointer coordinates can be linked to the a priori set electron beam positions, and thus the full LM-SEM coordinate transformation can be determined. To achieve a high overlay accuracy, the distortions between LM and SEM imaging systems need to be mapped. This we achieve by repeating this procedure for a large number of pointers and extracting the non-linear contribution to the coordinate transformation ( Fig.1 right panel). Moreover, for each region of interest, the LM-SEM overlay accuracy now depends on the ...