Bndicte Mortamet1, Matt A. Bernstein2, Clifford R. Jack Jr2, Jeffrey L. Gunter2, Chadwick Ward2, Paula J. Britson2, Reto Meuli3, Jean-Philippe Thiran4, Gunnar Krueger1
1Advanced Clinical Imaging Technology, Siemens Suisse SA, Healthcare Sector IM&WS - CIBM, Lausanne, Switzerland; 2Mayo Clinic, Rochester, MN, USA; 3Centre Hospitalier Universitaire Vaudois and University of Lausanne; 4Ecole Polytechnique Fdrale de Lausanne (EPFL), Signal Processing Laboratory (LTS5), Lausanne, Switzerland
Quality assessment of MRI is of great importance to derive reliable diagnostic information. As automated quantitative image analysis is being increasingly used in routine, automated measures of quality are needed. Based on a single magnitude image, we propose a procedure that automates the classification of data quality and allows detecting patient-/scanner-related artifacts. Validated on 750 datasets, the approach proofs to be a very promising candidate to perform quality assurance analysis for clinical practice and research. It could greatly improve clinical workflow through its ability to rule-out the need for a repeat-scan while the patient is still in the magnet bore.