La présentation sera assurée par Mateo Villa:

The use of x-ray imaging in interventional procedures overexposes medical staff to harmful ionizing radiation. Computational techniques can improve radiation safety by providing realistic radiation exposure feed-back during the intervention. However, existing methods are either inaccurate analytical approximations or unadaptable in real-time, i.e., Monte Carlo simulation (MCS). We propose an alternative approach based on deep convolutional neural networks (DCNN) trained on MCS databases. This allows a near real-time execution with a radiation exposure estimation close to MCS.