Meeting Banner
Abstract #2598

Deep learning based multi-organ detection: a step closer to liver T1 map quality assurance      

Charles E Hill1,2, Ferenc Emil Mozes2, Liam AJ Young2, Luca Biasiolli2, Matthew D Robson3, and Vicente Grau1
1Engineering Science, University of Oxford, Oxford, United Kingdom, 2Oxford Centre for Magnetic Resonance Research (OCMR), University of Oxford, Oxford, United Kingdom, 3Perspectum Diagnostics LTD., Oxford, United Kingdom

Quantitative Liver T1 mapping techniques are becoming more prevalent in the clinic, for the diagnosis and prognosis of various liver diseases. However, time pressure and technique complexity can often lead to incorrect acquisition due to sub-optimal slice location. We implement a faster Region Proposal CNN for the detection of multiple organs in the body, and report back the findings promptly via a local user interface, to inform the radiographers if they need to re-acquire. We show that this network has a high localisation accuracy and high speed, such that it will be applicable in the clinical environment.

This abstract and the presentation materials are available to members only; a login is required.

Join Here