Logo image
Back
Synthesis of MR fingerprinting information from magnitude-only MR imaging data using a parallelized, multi network U-Net convolutional neural network
Journal article   Open access   Peer reviewed

Synthesis of MR fingerprinting information from magnitude-only MR imaging data using a parallelized, multi network U-Net convolutional neural network

Kiaran P. McGee, Yi Sui, Robert J. Witte, Ananya Panda, Norbert G. Campeau, Thomaz R. Mostardeiro, Nahil Sobl, Umberto Ravaioli, Shuyue Zhang, Kianoush Falahkheirkhah, …
Frontiers in radiology, Vol.4, 1498411
12/01/2024
DOI: 10.3389/fradi.2024.1498411
PMCID: PMC11686891
PMID: 39742349
url
https://doi.org/10.3389/fradi.2024.1498411View
Published (Version of record) Open Access

Abstract

BackgroundMR fingerprinting (MRF) is a novel method for quantitative assessment of in vivo MR relaxometry that has shown high precision and accuracy. However, the method requires data acquisition using customized, complex acquisition strategies and dedicated post processing methods thereby limiting its widespread application.ObjectiveTo develop a deep learning (DL) network for synthesizing MRF signals from conventional magnitude-only MR imaging data and to compare the results to the actual MRF signal acquired.MethodsA U-Net DL network was developed to synthesize MRF signals from magnitude-only 3D T1-weighted brain MRI data acquired from 37 volunteers aged between 21 and 62 years of age. Network performance was evaluated by comparison of the relaxometry data (T1, T2) generated from dictionary matching of the deep learning synthesized and actual MRF data from 47 segmented anatomic regions. Clustered bootstrapping involving 10,000 bootstraps followed by calculation of the concordance correlation coefficient were performed for both T1 and T2 MRF data pairs. 95% confidence limits and the mean difference between true and DL relaxometry values were also calculated.ResultsThe concordance correlation coefficient (and 95% confidence limits) for T1 and T2 MRF data pairs over the 47 anatomic segments were 0.8793 (0.8136–0.9383) and 0.9078 (0.8981–0.9145) respectively. The mean difference (and 95% confidence limits) were 48.23 (23.0–77.3) s and 2.02 (−1.4 to 4.8) s.ConclusionIt is possible to synthesize MRF signals from MRI data using a DL network, thereby creating the potential for performing quantitative relaxometry assessment without the need for a dedicated MRF pulse sequence.
 convolutional neural network magnetic resonance fingerprinting MPRAGE relaxometry U-Net

Details

Metrics

2 Record Views
Logo image