Towards applying deep learning to predict rigid motion-induced changes in Q-matrices from UHF-MRI pTx simulations

Blanter, Katherine, Plumley, Alix, Malik, Shaihan and Kopanoglu, Emre ORCID: https://orcid.org/0000-0001-8982-4441 2023. Towards applying deep learning to predict rigid motion-induced changes in Q-matrices from UHF-MRI pTx simulations. Presented at: 2023 ISMRM & ISMRT Annual Meeting & Exhibition, Toronto, Canada, 3-8 June 2023.

PDF - Accepted Post-Print Version Download (274kB)

Abstract

Patient motion affects the specific absorption rate (SAR), a safety parameter in MRI. SAR is often calculated using so-called Q-matrices. We used conditional generative adversarial networks (cGANs) to estimate the effect of motion on magnitude from Q-matrices, which were extracted from body models simulated in a parallel-transmit (pTx) coil tuned to operate at 7T. Networks trained on Q-matrices from two positions were extrapolated to nine others. Network-predicted Q-matrices corresponded well with simulated ground truth motion-affected Q-matrices.

Item Type:	Conference or Workshop Item - unpublished
Date Type:	Published Online
Status:	Unpublished
Schools:	Schools > Psychology
Date of First Compliant Deposit:	5 March 2024
Last Modified:	22 Apr 2026 11:33
URI:	https://orca.cardiff.ac.uk/id/eprint/166877

Actions (repository staff only)

Edit Item