Towards Faster Fetal 3D CINE Cardiac MRI Reconstruction Using Region‐Optimized Virtual Coils ( ROVir )

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Purpose: To develop and evaluate, in a feasibility study, a computationally efficient image‐reconstruction pipeline for 3D balanced steady state free precession (bSSFP) CINE fetal cardiac MRI that reduces radial streaking artifacts, increases effective temporal resolution, and improves diagnostic image quality compared with previous work. Methods: Three‐dimensional bSSFP CINE data from 10 volunteers were used. Coil compression was performed using the region‐optimized virtual coils (ROVir) technique and compared with singular value decomposition (SVD) coil compression. For ROVir, the region of interest encompassed the fetal thorax, while the region of interference was positioned over the maternal subcutaneous fat to suppress unwanted signal contributions. Image reconstruction was performed for multiple temporal samplings. Two observers evaluated diagnostic image quality using a visual‐analog scale, and results were analyzed using visual grading characteristics (VGC). Computational efficiency was evaluated using reconstruction runtime and memory usage. Results: Average 3D CINE acquisition time was 5 min, with a reconstruction time of approximately 5 h for both SVD (8channels and 8 phases) and ROVir (3 channels and 16 phases), with approximately equal memory usage. ROVir enabled higher temporal resolution by reducing the channels required. VGC analysis showed a preference for ROVir across all diagnostic criteria (AUC > 0.5), including edge sharpness, perceived image contrast, artifact suppression, papillary muscles, great vessels, and extracardiac diagnostic image quality, with comparable median scores across temporal samplings. Conclusion: ROVir coil compression improved diagnostic quality and reduced reconstruction time. By reducing memory demands, reconstruction with a higher number of cardiac phases was also possible.
Funding Information
  • Region Skåne
  • Skånes universitetssjukhus
  • Hjärt-Lungfonden (20180510)
  • Lund University Medical Faculty Foundation
  • Vetenskapsrådet (2019‐05506)
  • Siemens Healthineers

This publication has 35 references indexed in Scilit: