Microscopy is always stronger together

Correlative microscopy, massive data sets and citizen science – times are really booming for electron microscopy says Lucy Collinson  

In biomedical research, electron microscopes are rarely used in isolation to answer biological questions. Instead, they fit into a suite of imaging tools. Much of our work starts with a cell or a model organism containing a protein that has been fluorescently tagged and imaged in a light microscope to determine where the protein is found, giving clues as to what its function might be.

They may form coats around malaria parasites that enable them to bind to red blood cells and cause infection, or light up the bacteria that cause tuberculosis as they hide from the human immune system, or mark the switch of an immune cell from a resting-state to one that is able to produce antibodies and fight cancer or infection. There is, however, a limit to what we can learn about these proteins using a light microscope, since we only see tiny spots that indicate where they are, but not the structures that surround them. Correlative microscopy images fluorescent molecules in the light microscope, and then tracks that same cell through the complex preparation methods required to image the same cell in the electron microscope.

Nevertheless, correlative microscopy is extremely powerful, helping us to find one cell in a million, the ‘needle in the haystack’

Preparation methods are complex because the sample needs to be treated with heavy metals to be seen in the electron microscope, and the sample must be embedded in plastic or in ice to withstand the vacuum required for the electron beam to operate effectively. Nevertheless, correlative microscopy is extremely powerful, helping us to find one cell in a million, the ‘needle in the haystack’. Nowadays, correlative microscopy is advancing at a fast pace. Inside new ‘volume’ electron microscopes, ion beams or diamond knives shave slices from the surface of plastic-embedded cells and tissues, so that thousands of images can be automatically collected to build up a 3D picture of the sample. Light microscopes have been added to electron microscopes to make a single system that can quickly switch from one imaging function to the other. And cryogenic light and electron microscopes that operate at less than -185°C can locate fluorescent proteins in cells that are embedded in ice that preserves them as close to their living state as possible for imaging in a vacuum.

The new challenge in microscopy is to extract useful information from these massive datasets

These microscopes allow us to delve deep into the microcosm of the cell, but as the technology has leapt forward, so the amount of data that is produced from these systems has sky-rocketed. The new challenge in microscopy is to extract useful information from these massive datasets. For example, in 2008 we collected thousands of electron images from two developing blood vessels fusing to form a new vessel, a process that occurs during the growth of tumours. Though it took only three days to collect the images, it then took two electron microscopists four months to analyse the data, find the blood vessels amongst all the other cells, and produce 3D models.

One way around this problem is to train computers to automatically recognise blood vessels and other cellular structures in electron microscopy images, but to do this requires thousands or millions of images in which the structures have been labelled as examples. This is not practical in electron microscopy – there are not enough experts with enough time to do this work – so we have taken inspiration from the astrophysicists who have been harnessing the power of the general public to analyse images of galaxies.

Over the last three months, using the Zooniverse platform, almost 3,000 citizen scientists have analysed more than 50,000 images of cells, a prodigious effort that will allow us to start training computers to automatically analyse our images and speed up the process of discovery science.

Lucy Collinson is Head of Electron Microscopy at the Francis Crick Institute. She will be speaking at the Microscience Microscopy Congress 2017, held between 3^rd-6^th July in Manchester.