Improved CMR imaging around cardiac implantable devices at 0.55T using phase-cycled spiral bSSFP

Presenting Author(s)

• 

  Ye Tian, PhD

  Research Assistant Professor
  University of Southern California
  Alhambra, California, United States

Primary Author(s)

• 

  Ye Tian, PhD

  Research Assistant Professor
  University of Southern California
  Alhambra, California, United States

Co-Author(s)

• JD

  Jon A. Detterich, MD

  Professor
  Children's Hospital Los Angeles
  Los Angeles, California, United States
• 

  John C. Wood, MD, PhD

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

  Krishna S. Nayak, PhD

  Dean's Professor
  University of Southern California
  Los Angeles, California, United States

Background: Modern mid-field MRI systems, including 0.55T¹, provide unique opportunities for improved CMR imaging of patients with cardiac implanted devices (CID)^2,3. These systems reduce susceptibility effects and specific absorption rate, resulting in substantially reduced imaging artifacts and relaxed RF heating concerns. In this work, we investigate bSSFP imaging around CID at 0.55T and evaluated a phase-cycled spiral bSSFP sequence for artifact suppression.

Methods: Acquisition and Reconstruction: We implemente a golden-angle spiral bSSFP pulse sequence with 8 phase cycles (0^o to 360^o, linear increments)⁴ and acquire data for 1.5s per phase cycle (12s per slice) while recording synchronized ECG. Each phase-cycle was reconstructed using spatiotemporal constrained reconstruction⁵ at 40ms/frame. One cardiac cycle was selected from each phase cycle and phase-cycled images were combined using root sum-of-squares (SoS) combination⁶.

Experimental

Methods: Experiments were performed on a 0.55T prototype MAGNETOM Aera (Siemens Healthineers) equipped with high-performance gradients (45mT/m amplitude, 200T/m/s slew rate). Five healthy volunteers without implanted devices (29-63Y, 1M/4F) and one volunteer with two 20-mm stents (80Y M) were scanned. A cardiac pacemaker (Accolade MRI Pacemaker L311, Boston Scientific) and a defibrillator (Claria MRI Quad CRT-D SureScan DTMA1QQ, Medtronic) were externally placed under the left clavicle to approximate the perturbation caused by implanted devices. Short-axis stacks were acquired during breath-holding using the phase-cycled spiral bSSFP in RTHawk (Vista.ai). Reference short-axis stack was also acquired without devices, using a Cartesian bSSFP cine product sequence (Siemens Healthineers). The study was approved by our Institutional Review Board, and all participants provided written informed consent.

Results: Figure 1 shows representative SoS combined and single-phase cycle images at end diastole for both the pacemaker and the defibrillator. Single phase-cycled image suffers from banding artifacts, while phase-cycled bSSFP effectively mitigated artifacts. Figure 2 shows GRE, standard bSSFP, and phase-cycled bSSFP images from a volunteer with stents. GRE images provided insufficient contrast to delineate endocardial boundary, regardless of flip angle. Standard bSSFP has banding artifact, however, phase-cycled bSSFP had minimal artifact in the heart. Figure 3 shows LV volumes and EF measured from phase-cycled spiral bSSFP with devices and Cartesian bSSFP without device. All measurements agreed except LV EF measured with defibrillator (p=0.02). However, phase-cycled bSSFP provided overall lower volumes, likely due to image distortion caused by the large B0 inhomogeneity.

Conclusion: Cardiac implanted devices can generate severe artifacts at 0.55T, including bSSFP banding and image distortion. The proposed phase-cycled spiral bSSFP sequence can effectively reduce banding artifacts. Image distortion caused by devices can alter the cardiac volume measurements.

Figure 1. Representative phase-cycled bSSFP images with a pacemaker and defibrillator. Square root of sum-of-square (SoS) combined phase-cycled bSSFP (left) and individual phase-cycled images (right) are shown at end-diastole. Note that the bSSFP artifact is affected by the type of device, distance from the heart, and device orientation. In this volunteer, the pacemaker produced minimal artifacts, and carefully chosen phase cycles can avoid bSSFP banding in the heart (phase cycles of 180o to 315o). For defibrillator, none of the individual phase-cycled images is artifact-free in the heart, however, SoS-combined image effectively suppresses artifacts. In addition, spatially varying spiral blurring and image distortion were observed, likely due to the large B0 variation generated by the device. For both devices, SoS combined images improved the apparent SNR compared to single PC images.
Figure 2. Comparison of GRE, standard bSSFP, and SoS PC bSSFP in one volunteer with two 20-mm stents. End systolic phase is shown for all images. All images exhibit signal void around the stent that extends into the myocardium. GRE images suffer from poor contrast regardless of the flip angle, likely due to signal saturation of static blood, confounding the delineation of endocardial border. Standard bSSFP (phase cycle of 180o) provides improved contrast compared to GRE but with increased banding artifacts. Phase-cycled bSSFP shows a similar extent of signal void as GRE and without banding artifacts in the heart. Note that the Phase-cycled bSSFP image was acquired at a slightly different slice location than GRE and standard bSSFP, as the sequences were implemented on different console systems (Siemens and RTHawk).
Figure 3. Comparison of left ventricular (LV) end-systolic volume (ESV), end-diastolic volume (EDV), and ejection fraction (EF) between phase-cycled spiral bSSFP (with device) and reference Cartesian bSSFP (without device). Placement of CIDs resulted in lower volume measurements compared to reference, but the differences were not statistically significant overall. For defibrillators, EDV was on average 22 mL lower than reference, likely due to image distortion. Pacemaker EF did not differ significantly from reference EF (P = 0.08), whereas defibrillator EF was significantly reduced compared to reference (P = 0.02).