Artificial neural networks to filter events in ion computed tomography

Context

Ion computed tomography (CT) is an imaging technique developed in the context of ion beam therapy which is proposed to complement X-ray CT in the future [1]. Prototype scanners record kinematic properties (position, energy etc.) of individual protons traversing a patient. As a prerequisite for the reconstruction, it is necessary to identify and categorise these events based on the type of interaction they had in the patient and/or the detector. So far, this is done through cuts applied to the energy and angular distributions of recorded particles.
 

Objective

The purpose of this master internship is to develop new algorithmic tools to filter ions according to their measured properties using artificial neural networks. Data will be generated in Monte Carlo simulations which are already available in our group. We plan to exploit a recently published approach [2], which is particularly suited for heavier ions such as helium and carbon, and extend it by including other measured parameters such as the particles’ propagation angle. The results of the study might also provide clues on how to optimise future ion CT scanner prototypes.
 

Tasks

  • Identify possible input parameters for the neural network.
  • Implement a neural network using a readily available open source toolkit.
  • Study and optimise the network architecture.
  • Apply the network to Monte Carlo simulated data.
  • Evaluate the reconstructed image quality.
 

Required skills

  • Education: master student in applied maths or image processing.
  • Scientific interests: applied maths, neural networks, medical imaging, particle physics.
  • Programming skills: Python, C++.
  • Languages: Command of English required, French optional.
 

Practical information

  • Supervision: Nils Krah, David Sarrut
  • Location: Centre Léon Bérard, Lyon, France.
  • Period: 2019 (duration negotiable).
  • Please send a CV, master marks and a brief statement of interest by email to Nils Krah (nils.krah@creatis.insa-lyon.fr).
 

References

[1] Katia Parodi. Heavy ion radiography and tomography. Physica Medica, 30(5):539–543, 2014.
[2] Lennart Volz, Pierluigi Piersimoni, Vladimir A Bashkirov, Stephan Brons, Charles-Antoine Collins-Fekete, Robert P Johnson, Reinhard W Schulte, and Joao Seco. The impact of secondary fragments on the image quality of helium ion imaging. Physics in Medicine & Biology, 63(19):195016, 2018.