Source: UT-News, September 28th, 2015 by Paul de Kuyper
Due to noise on medical photos it is sometimes difficult to diagnose if a patient has rheumatism. Applied Mathematics student Rien Lagerwerf developed a method to drastically enhance these images.
With Lagerwerfs method it will become a lot easier to determine inflammations in the joints. Whereas these inflammations might indicate that a patient has rheumatism. The mathematician graduated half September on this subject.
Doctors use several methods to determine inflammations in the joints; one of these methods is photoacoustic imaging. "You send light into a finger such that the finger heats up. This heating will result in waves, which you can measure and from these measurement one can create an image", summarises Lagerwerf. On the resulting images inflamed joints will be a different colour than the healthy joints.
Eliminating the noise
A mathematical model is needed to translate the physical process of photoacoustic imaging to an image that is useful for a doctor. Up to now the models used by the researchers did not take the prior knowledge about the finger into account. Lagerwerf: "We know for instance what kind of intensity the blood vessels should have. Using this we can say before constructing the images how they should look like".
However the methods Lagerwerf used cannot remove all the noise, but the resulting images do become a lot sharper. In the figures below a (reconstruction of a) finger is shown, the difference in methods is clearly visible; left is an image made with the old model, right is an image reconstructed with the model used by Lagerwerf.
Lagerwerf expects that using these kinds of methods will be a great help for the doctors. "The more certain a doctor is that a dot in the figure is an inflammation, the more certain he can be about his diagnosis."
The mathematician expects however that his model will not be implemented in hospitals in the near future. The program that makes the reconstructions is at the moment quite hard to use. Moreover the methods will have to be tested more extensively in practice. Another master student is at the moment working on applying these methods to real patient data.
Lagerwerf will possibly keep working on his graduating project, as he is looking for a PhD position in mathematical imaging.