In 2019 Radboud University Medical Centre (Radboudumc) established a new electron microscopy centre, providing services within Radboud University Medical Centre and also for colleagues from industry.
Nico Sommerdijk, Professor in bone biochemistry at Radboud University Medical Centre says:
We’re bringing together really advanced electron microscopy techniques to push beyond the limits of what currently we’re able to do. And one of the ways that we’re doing that is the use of liquid phase electron microscopy. In liquid phase electron microscopy, we’re putting our samples inside a small aquarium so that we can see the processes in the microscope while they are ongoing. The aquarium prevents the sample from drying out in the microscope. and have electron transparent windows so that we can actually see what is happening inside.
The problem
The problem is that the electron microscope uses a concentrated electron beam that is not only giving us information, but it’s also damaging the sample. So we have to be very careful in how many electrons that we can use. The more we use the clearer the images are, but also the less images we can record because every image is damaging our sample. So what we want to do is to reduce the damage and we want to reduce the number of electrons that are needed.
The solution
We are working with SenseAI, who have developed a methodology where we can use only 20% or 10% of the pixels and therefore of the electrons that we’re using to create an image. Originally you don’t see much, but after filling in the missing spots to reconstruct the image. When we compare these images with the ground truth, it’s very, very difficult to see the difference. This is how we think we can reduce the electron dose that we need, a factor of 5 maybe factor 10, so that we can actually record more images or get better resolution in the images that we’re recording.
Dr Luco Rutten, a postdoc in the group of Nico Sommerdijk and part of the electron microscopy center at Radboudumc adds:
My research focuses on the development of workflows to enable the imaging of dynamic biological processes using liquid phase electron microscopy. As Nico already mentioned, these processes are very sensitive to the electron beam so the subsampling helps us to reduce the damage on our samples. One additional challenge when imaging biological processes, is that you need a large volume to look at the process. We are now able to reduce the electron dose to one electron per square angstrom, still reaching a resolution of five nanometer using STEM.
Future applications with SenseAI
Recently, we have started a project where the mineralisation of other issues is being investigated, specifically where you do not want it. We’re looking currently at heart valve calcification where the collagen of our heart valves also becomes mineralized for reasons that we still do not very much understand. So we’re using a focused ion beam scanning electron microscope to make a lamella that we take out and we take that to the TEM in a frozen state. There we image that sample in its frozen state, but then we’re going to heat it up inside the microscope and then we’re going to look at restarting that process again. But with SenseAI, we’re hoping to look at the restarting processes with nanometer, or nanoscale, resolution.
The the big dream is that we can restart frozen processes in the microscope, but now look at them live, like a in a real tissue sample. There we will have the same challenges as we have in looking at the dynamics in liquid phase electron microscopy, so the sparse imagingfrom SenseAi will be a really important tool.