Deirdre C. Sheahan MD, Presenter: Nothing to Disclose

Shilpy Chowdhury MD, MPH, Abstract Co-Author: Nothing to Disclose

Peng Lai PhD, Abstract Co-Author: Employee, General Electric Company

Frandics Pak Chan MD, PhD, Abstract Co-Author: Nothing to Disclose

Shreyas Shreenivas Vasanawala MD, PhD, Abstract Co-Author: Research support, General Electric Company Consultant, Zimmer Holdings, Inc Consultant, ArthroCare Corporation

Assess image quality and accuracy of k- & adaptive-t-space (kats ARC) data synthesis as a method of accelerating MRI cardiac imaging compared to standard parallel imaging and k-t GRAPPA .

We use k-t-space correlation with modified acquisition and reconstruction. K-space is sampled in a time-interleaved fashion to minimize distance between unsampled lines at a cardiac phase and sampled lines at its immediate temporal neighbors. Data is Fourier transformed in kx hybrid space. Data synthesis uses the nearest two neighbors along PE at the same phase (tn ) and the sampled neighbor with the closest PE at each adjacent phase covered by a phase-specific time window (Wn ). Wn was determined for each phase based on local motion derived from low-resolution images from autocalibration signals. Wn is thus narrower at systole. 14 pediatric patients referred for cardiac MRI (1.5 T, 8 channel coil) had fully sampled short axis stack cine data stored. Images were reconstructed from fully sampled data and undersampled data to simulate acceleration. Undersampling factors were 4x for ARC, and 4x/6x/8x for kats-ARC. Images were assessed for quality, reconstruction error, and ventricular volume/ejection fraction.

Visual quality was higher for 8x kats-ARC than 4x ARC. Root mean square reconstruction error was calculated for first 5 patients. 4x ARC averaged 19%, and 6% at 4x, 8% at 6x and 9% at 8x for kats-ARC. RV end-diastolic volume mean error vs. full sampling (FS) was 5.5% for 4x ARC, 1.7% for 4x kat-ARC, 3% for 6x kat-ARC, and 7.7% for 8x kat-ARC. RV end-systolic volume mean error vs. FS was 9% for 4x ARC, 6% for 4x kat-ARC, 8.5% for 6x kat-ARC, and 9.4% for 8x kat-ARC. RV ejection fraction mean error vs. FS was 17% for 4x ARC, 11% for 4x kat-ARC, 14% for 6x kat-ARC, and 14% for 8x kat-ARC. LV end-diastolic volume mean error vs. FS was 4.3% for 4x ARC, 1.9% for 4x kat-ARC, 2% for 6x kat-ARC, and 4% for 8x kat-ARC. LV end-systolic volume mean error vs. FS was 8.7% for 4x ARC, 4.3% for 4x kat-ARC, 5.6% for 6x kat-ARC, and 8.7% for 8x kat-ARC. LV ejection fraction mean error vs. FS was 8.7% for 4x ARC, 6% for 4x kat-ARC, 6.5% for 6x kat-ARC, and 10% for 8x kat-ARC.

kats-ARC (8 channel coil) preserves image quality, ventricular volumes, and ejection fractions at higher acceleration.

Increased acceleration would enable a short axis functional assessment in a single breath-hold.

Sheahan, D, Chowdhury, S, Lai, P, Chan, F, Vasanawala, S, Adaptive Spatio-Temporal Acceleration for Faster Cine Bright-blood Cardiac MRI.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11013506.html