Kang Wang PhD, Presenter: Employee, General Electric Company

Scott K. Nagle MD, PhD, Abstract Co-Author: Stockholder, General Electric Company

Harald Kramer MD, Abstract Co-Author: Nothing to Disclose

Tao Zhang, Abstract Co-Author: Nothing to Disclose

Philip Beatty, Abstract Co-Author: Nothing to Disclose

Mahdi Rahimi, Abstract Co-Author: Nothing to Disclose

Courtney K. Morrison, Abstract Co-Author: Nothing to Disclose

Frank Richard Korosec PhD, Abstract Co-Author: Research Support, General Electric Company

Scott Brian Reeder MD, PhD, Abstract Co-Author: Nothing to Disclose

Dan W. Rettmann BS, Abstract Co-Author: Employee, General Electric Company

Ersin Bayram PhD, Abstract Co-Author: Employee, General Electric Company

James H. Holmes PhD, Abstract Co-Author: Employee, General Electric Company

Coil-by-coil (CBC) data-driven auto-calibrating parallel imaging has become more widely used for dynamic MR applications, such as dynamic contrast-enhanced (DCE) MR angiography (MRA). However, for high spatial resolution, high parallel imaging factors and high channel count coil array protocols, the image reconstruction time can be clinically unacceptably long. This work demonstrates an initial comparison of combining Coil Compression and Direct Virtual Coil (CCDVC) to significantly reduce reconstruction times vs. the currently used conventional method of CBC in the setting of dynamic contrast enhanced peripheral runoff MR angiography.

Twenty-four volunteers (7 healthy, 17 with pathology) were imaged and informed consent was obtained prior to all scanning. All scans were conducted on a clinical scanner (3.0T MR750, GE Healthcare), with 48cm FOV, 1.0mm isotropic resolution, 32-channel coil array and parallel imaging factor of 3 (phase) × 2 (slice) = 6. Temporal view-sharing was used to generate raw data for each time frame. The same raw data sets were then reconstructed twice: once with CBC and once with CCDVC. Time-resolved images reconstructed using CBC and CCDVC were randomized and blindly scored by two radiologists using a 5-point scale: image 1 much better (clinically significant); image 1 slightly better (not clinically significant); equivalent; image 2 slightly better (not clinically significant); image 2 much better (clinically significant).

The reconstruction times for CCDVC were about 18× faster than the conventional CBC method (17 sec vs. 311 sec per phase on an offline computer) in the parallel imaging synthesis/FFT/coil combination module. One radiologist scored 18 cases as “equivalent”, 3 cases as CCDVC “slightly better”, and 3 cases as CBC “slightly better”; the second radiologist scored 17 cases as “equivalent”, and 7 cases as CCDVC “slightly better”. 

The combined CCDVC technique was shown to significantly reduce the reconstruction time for high spatial and temporal resolution peripheral MRA, with no significant loss in image quality.

The reconstruction speed for a DCE peripheral MRA scan can be significantly improved by using CCDVC to enable more aggressive clinical protocols or improve clinical workflow.

Wang, K, Nagle, S, Kramer, H, Zhang, T, Beatty, P, Rahimi, M, Morrison, C, Korosec, F, Reeder, S, Rettmann, D, Bayram, E, Holmes, J, Improving Reconstruction Speed for Dynamic MRI Using Parallel Imaging with Combined Coil Compression and Direct Virtual Coil.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13022848.html