MROnics used in MR Linac applications
Nominating Supervisor: Frank Van den Heuvel
Second Supervisor: Sam Vinko
The advent of clinically implemented MR Linacs generates an excess of available imaging data. This can be leveraged to predict the outcome of the treatment and provide valuable information in a biological adaptation of a patients’ treatment.
In this project we will use the latest Artificial intelligence and machine learning tools combined with a more classical model based technique to find imaging markers, singular or rate based which can be predictors.
1) Increase potential biomarker candidates by including information from MR imaging obtained during treatment;
2) Develop a strategy to combine biomarkers which enables the inclusion of quantitative and semi--quantitative data (e.g. patient condition, general assessment);
3) Increase the data set used to investigate the mechanisms of radiation--induced tissue damage and the evolution of oxygenation, and
4) Create a clinical data set which can inform a Generative Adversary network (GAN) which can, in turn, be used to train neural networks that have been built with less detailed information.
It is clear that a suitable candidate is well versed in clinical medical physics but also would need to have a good underpinning in machine learning techniques and modern computer languages (Python, Julia, TensorFlow).
Our group combines basic physics, computer science and clinical medical physics and could later lead to opportunities in either clinical or medical environments.
Teoh, F. Fiorini, B. George, K. A. Vallis, and Van den Heuvel, F. Proton vs photon: A model-based approach to patient selection for reduction of cardiac toxicity in locally advanced lung cancer.Radiotherapy and Oncology , 2019/08/19 In Press.
Van den Heuvel Frank , George Ben, Schreuder Niek, and Fiorini Francesca. Using stable distributions to characterize proton pencil beams. Medical Physics, 45(5):2278–2288, 2018.
Francesca Fiorini, Niek Schreuder, and Van den Heuvel, Frank . Technical note: Defining cyclotron based clinical scanning proton machines in a FLUKA monte carlo system. Medical Physics , 45(2):963–970, 2018.