Breath-hold, Four-Dimensional, Ferumoxytol-Enhanced MUSIC: Initial Results in Adult Congenital Heart Disease

Presenting Author(s)

• TY

  Takegawa Yoshida, MD

  Associate Project Scientist
  University of California, Los Angeles
  Los Angeles, California, United States

Co-Author(s)

• JF

  J. Paul Finn, MD, MSc

  Professor
  University of California, Los Angeles
  Los Angeles, California, United States
• 

  Shu-Fu Shih, PhD

  Assistant Project Scientist
  University of California, Los Angeles
  Los Angeles, California, United States
• 

  Kim-Lien Nguyen, MD

  Associate Professor of Cardiovascular Medicine and Radiology
  David Geffen School of Medicine at UCLA and VA Greater Los Angeles Healthcare System
  Los Angeles, California, United States
• AB

  Arash Bedayat, MD

  Assistant Professor
  University of California, Los Angeles
  Los Angeles, California, United States

Primary Author(s)

• TY

  Takegawa Yoshida, MD

  Associate Project Scientist
  University of California, Los Angeles
  Los Angeles, California, United States

Co-Author(s)

• TY

  Takegawa Yoshida, MD

  Associate Project Scientist
  University of California, Los Angeles
  Los Angeles, California, United States
• NJ

  Ning Jin, PhD

  Senior Key Expert
  Siemens Healthineers
  Cleveland, Ohio, United States
• 

  Xiaoming Bi, PhD

  Director, Cardiovascular MR Collaborations
  Siemens Medical Solutions USA, Inc.
  Oak Park, California, United States
• FH

  Fei Han

  Siemens Healthineers
  Siemens Healthineers, United States
• 

  Xiaodong Zhong, PhD

  Associate Professor
  University of California, Los Angeles
  Los Angeles, California, United States

Background: Multi-dimensional magnetic resonance (MDMR) with ferumoxytol enhancement is increasingly used to unravel complex functional anatomy in neonates with congenital heart disease (CHD) (1-3). However, free breathing (FB) MDMR techniques in older children and adults involve sophisticated motion management, usually requiring offline lengthy image reconstruction (2,3), impeding clinical adoption of MDMR. We recently established the technical feasibility of acquiring isotropic 4D cardiovascular MRI within a single breath-hold (BH) (4). We aim to assess the performance of a focused ferumoxytol-enhanced BH 4D-MUSIC in adults with CHD.

Methods: The BH 4D acquisition consisted of a high bandwidth gradient echo readout (TR/TE 3.0 /1.0 ms) with a variable density, Cartesian phyllotaxis k-space trajectory and L1-regularized wavelet-based compressive sensing (CS) inline image reconstruction (4), implemented on 1.5T and 3T MRI systems. Default parameters were adjusted to acquire up to 25 cardiac phases with retrospective ECG-gating and isotropic voxel dimensions in the range 1.5 mm to 2 mm. 33 adult patients (F/M = 22/11; mean age 32 yrs, range 18-65 yrs) with known CHD underwent ferumoxytol-enhanced MRI for clinical indications and gave written informed consent. All imaging was performed in the steady state distribution of ferumoxytol 4 mg/kg. In addition to BH 4D MUSIC, spoiled gradient echo cardiac cine imaging (N=30) was performed. Target acquisition volumes included whole-heart ventricular coverage for cardiac functional assessment as well systemic and pulmonary vascular anatomy. LV and RV volumes were compared between BH 4D MUSIC and 2D cine in 20 subjects and single ventricular volumes in 9 subjects using Bland-Altman and regression analyses.

Results: BH 4D MUSIC data were successfully acquired and reconstructed inline within two minutes for all patients. Breath-hold durations ranged from 17-24 seconds, and 10-25 cardiac phases were reconstructed, depending on heart rate and breath-hold tolerance. Intravascular signal on BH 4D MUSIC remained high and uniform on all cardiac phases (Figs 1,2), even with an acceleration factor of 25X at 1.5T. The limits of agreement between MUSIC and BH 2D cine derived LV and RV volumes were within 6 ml on Bland Altman analyses (Fig 3) and high correlation was observed on regression analyses (Fig 4). The isotropic BH 4D MUSIC acquisitions enabled reconstruction of volumetric cine images in flexible orientations (Figs 1,2).

Conclusion: Breath-hold, ferumoxytol-enhanced 4D MUSIC is clinically applicable to adult patients with CHD and compares favorably to 2D cine for cardiac volumetric measurements, without the risk of slice misregistration. Our findings have implications for addressing workflow challenges in clinical cardiac MR where multiple 2D acquisitions can be collapsed into a single 4D set.

Figure 1. 65 y.o. female with congenitally corrected transposition of the great arteries (LTGA). a) Multiplanar reconstructions on Tempus Pixel software displays one of 22 phases from BH 4D MUSIC at 1.5T. Breath-hold duration was 21 secs with acceleration factor X25. Note the uniform blood signal and clear definition the systemic right ventricular outflow tract and aortic valve leaflets (arrows), as well as the conspicuous trabeculations of the systemic right ventricle (arrowheads). b) Same patient and same acquisition as in Fig 1a. Color volume rendered frame from BH 4D MUSIC (left panel; Vitrea workstation) and corresponding 4D flow (right panel; Tempus cloud) on one of 22 phases shows the outflow from the systemic right ventricle (arrow) and pulmonary left ventricle (arrowhead) with their respective flow patterns.
Figure 2: 42 y.o. female with mild aortic coarctation. a) Multiplanar reconstructions on OsiriX software displays one of 10 phases from BH 4D MUSIC at 3.0T . Breath-hold duration 22 secs with acceleration factor X15. Note the uniform blood signal and clear definition of the coarctation (white arrow) and left coronary artery (red arrow). b) Same patient and same acquisition as in Fig 4a. Color volume rendered frame from BH 4D MUSIC (left panel; Vitrea workstation) and corresponding 4D flow (right panel; Tempus cloud) on one of 10 phases shows the anatomy and flow field of the coarctation (arrow). c) Same patient as in Figs 2a,b. Separate BH 4D MUSIC acquisition with higher temporal resolution for full coverage cine imaging. Breath-hold duration 19 secs with acceleration factor X15. Top panel shows single phase from 2D cine at several short axis positions. Bottom panel shows corresponding short axis cine 18 phase reconstructions from BH 4D MUSIC with 3mm contiguous slices and 50 ms temporal resolution.
Figure 3: a) Bland-Altman plots comparing left and right ventricular end systolic and end diastolic volumes on BH 4D MUSIC and BH 2D cine . Mean values were highly similar between 4D MUSIC and 2D cine, with a mean difference less than 2 ml and a spread less than 6 ml. b) Linear regression plots of left ventricular end diastolic volume (LVEDV) and end systolic volumes (LVESV) and corresponding right ventricular volumes (RVEDV, RVESV) on BH 4D MUSIC (y axis) and BH 2D cine (x axis) in 30 adult patients with congenital heart disease. The correlation is very high, with most values lying close to the line of identity.