Aller au contenu principal

Engineer position: Federated deep learning in medical imaging

Research at LTSI (Laboratoire Traitement du Signal et de l'Image - INSERM U1099) is at the
interface between information technologies and health sciences.
The IMPACT team is taking part in the European project PAROMA-MED, the aim of which is to
develop a platform for implementing federated learning methods, i.e. exploiting data from
several clinical centers without transmitting the data.
See https://paroma-med.eu/ for a description of the project.
In conjunction with the project partners, the aim of this position is to exploit the developed
platform to implement a federated deep learning approach for segmenting cardiac CT images.
The activities will be dedicated to:

  • access to data: link with PACS and FHIR; data anonymization
  • data preprocessing
  • federated model training (e.g. Unet with Flower framework)
  • deployment of the use case on the project’s platform
  • tests and evaluation
  • participation to the writing of deliverables

 

Profile required:

  • Engineer, Master
  • Field: information technologies, computer science, health technologies
  • Duration: 1 year, extendable for 1 year
  • Skills in data science and computer science
  • Interest in biomedical research and medical image processing

Contact:
Antoine Simon : antoine.simon@univ-rennes1.fr