Keywords:
Abdomen, Pancreas, Liver, MR, Image manipulation / Reconstruction, Imaging sequences, Technology assessment, Artifacts, Cancer
Authors:
D. Hausmann1, T. Niemann2, D. Kreul1, A. Nocito1, M. Klarhöfer3, M. Nickel4, B. Kiefer4, R. A. Kubik-Huch1; 1Baden/CH, 2Nussbaumen/CH, 3Zürich/CH, 4Erlangen/DE
DOI:
10.26044/ecr2019/C-2002
Methods and materials
So far,
20 datasets (10 fbVIBE/10 bhVIBE) were independently assessed by two experienced readers in this IRB-approved study.
All fbVIBE/bhVIBE sequences (temporal resolution: approximately 11/18 seconds) were performed on state-of-the-art 1.5T clinical MRI scanners (MAGNETOM Avanto Fit/Aera,
Siemens Healthcare,
Erlangen,
Germany).
Initial cs reconstruction using hard-gating to reduce respiratory motion artifacts was carried out on the scanner during the examination (duration: approximately 8 minutes).
Mann-Whitney U-tests were performed to compare image quality (IQ),
delineation of structures (DoS),
artifacts (A) and diagnostic confidence (DC),
which were rated on Likert-type scales (IQ/DoS/DC: 1 (non-diagnostic)-5 (perfect); A: 1 (no artifacts)-5 (severe artifacts)).
Interobserver agreement was assessed using Cohen`s Kappa.
Motion-state resolved (xd) reconstruction with improved image quality has increased reconstruction times proportional to the number of selected motion states (chosen to be 6,
resulting in duration of approximately 45 minutes).
Clinically acceptable reconstruction times can only be achieved on GPU supported scanners.
Therefore the acquired raw data is stored and selectively reconstructed retrospectively.