Erik Bekkers, one of DSC/e’s researchers in the Mathematical Image Analysis group, has received two awards for his recent work on deep learning in medical imaging. In a successful collaboration with Maxima Lafarge, a PhD student in the Medical Image Analysis group, he developed a deep learning technique that exploits the symmetry to be found in image data in order to make better use of available training samples. Moreover, by adding geometric structure to learning architectures, the machines can focus specifically on learning relevant feature representations without also having to learn the underlying geometry.

In this fruitful collaboration between the departments of Mathematics and Computer Science and Biomedical Engineering, Erik Bekkers showed on a wide range of medical imaging problems the benefits of the approach. Their paper “Roto-Translation Covariant Convolution Networks for Medical Image Analysis” has gained very positive attention by their peers in the form of various awards:

1) In July, Erik presented his work at the (sold-out) international conference on Medical Imaging with Deep Learning (MIDL 2018) in Amsterdam. There, he and Maxime received the Philips Impact Award, which is a prize to reward work with great promise to be of significant benefit to society and patients.

2) This month, Erik presented his work for an audience of over 1500 people at the biggest conference in his field: Medical Image Computing and Computer Assisted Interventions (MICCAI 2018). The MICCAI Society awarded him for his contributions to the field with the highly prestigious MICCAI Young Scientist Award.

3) Most recently Erik was nominated for the 2017 Best TU/e thesis award. The thesis was put forward as one of the best theses of the year by both the Biomedical Engineering and Mathematics and Computer Science departments. Although he could not turn his award winning streak into a hat-trick – the award went to the excellent work by Fons van der Sommen of Electrical Engineering – he is very honored by the nomination and all the positive attention.