"One of the techniques I have included in my research is contrast-enhanced echo. My supervisors, Professor Mischi and Professor Wijkstra, have done a lot of research into this. The technique uses small air bubbles that are introduced into the bloodstream. Air bubbles bounced back an awful lot of sound and this technique allows you to visualise the character of blood vessels.
This is a valuable application in cancer research because tumours need a certain blood supply in order to develop. Another technique is elastography, in which sound waves are used to see how elastic tissue is. Tumours are often stiff. This is why the technique is sometimes compared to feeling a lump using sound".
What was the added value of machine learning in your research?
"At the beginning of such a study, I used data from a small patient population. When it appeared that a combination of scans indeed had the potential to better demonstrate prostate cancer, it became important to collect more data to train the model.
The more input you put into machine learning, the better the system can learn. With the help of machine learning, we have now developed a model that combines different parameters related to blood vessels or stiffness of tissue to provide a better picture of tumours. However, this needs to be further investigated in the clinic."