Teun Huijben’s research is about particles so small they are invisible to the naked eye. In fact, they cannot even be seen through an ordinary microscope. You need a special instrument to observe these tiny particles, such as proteins, in a cell. To study and find cures for many kinds of disease, it is important to be able to scrutinise small, individual parts of cells separately. Huijben has developed a smart algorithm which makes this possible.

Apples and pears

Every microscope has a maximum resolution. Beyond that limit, you cannot see your sample any more clearly. But scientists use tricks to try to do just that. “For example,” Huijben explains, “an optical microscope that uses special fluorescence to make certain proteins emit light.” As a result, they flicker constantly, a bit like the indicators on a car or flashing Christmas lights. “Thanks to that illumination, you notice the proteins and can check what shape they are later. If they look abnormal, that could be a sign of disease. So you want to select for that.”

With the old method, this is not easy, since an average is calculated from a very large number of images. Such an approach is certainly smart, because it gives you a sharper image. But the averaging process also means that you overlook abnormal proteins or other deformed cell parts. You can compare it with distinguishing apples from pears, says Bernd Rieger, Huijben’s supervisor and a professor in the research department of Imaging Physics. Those fruits have quite different shapes. “But we look at particles so small that the resolution is not good enough to tell ‘apples’ and ‘pears’ apart. You need several images to be able to discern that.” If you average the images, though, you end up with a picture of something that looks like neither a pear nor an apple.

Exploratory study

What you really want, then, is a sharper image with more detail. This is where Huijben’s new research comes in. He has created a calculation program, an algorithm, which can distinguish between the shapes of an “apple” and a “pear”. The algorithm divides images into different categories in such a way that you can now tell them apart. “This increases the resolution,” says Huijben, “so you can look at small particles with even more detail. This happens even without the program knowing exactly what it is looking for. So it doesn’t need any prior knowledge of the shape.”

This research takes science a step forward, says supervisor Rieger. “It is great that he’s found this. When Teun came to me to discuss his master project, this topic immediately appealed to him. But we didn’t know at the time whether he could achieve a result. It really was an exploratory study in which, under our supervision, he had to find his own way because it was unclear if it would work.” Huijben simply calls it “high risk, high gain” research.

According to Rieger, the toughest challenge besides the actual research work was the written reporting. “Compared with other students, he really stood out in that respect. Teun has distilled his findings into an article, which we are now submitting to scientific journals. For that you have to write in a very compact way. Which is very difficult, but he has largely done it himself.”