Feasibility of Multi-Thin-Slab Whole-Heart 3D Cine with Isotropic Resolution and High Contrast in Free Breathing
Peng Lai1, Haonan Wang2, Anja C.S Brau1, and Martin A Janich3

1Global MR Applications and Workflow, GE Healthcare, Menlo Park, CA, United States, 2Global MR Applications and Workflow, GE Healthcare, Waukesha, WI, United States, 3Global MR Applications and Workflow, GE Healthcare, Munich, Germany


Conventional 2D cine is hindered by its needs of repeated breathhold and imaging at each individual view, while breathheld 3D cine suffers from limited slice coverage and resolution. This work developed a new 3D cine sequence with free-breathing capability for whole-heart coverage and isotropic resolution and multi-thin-slab acquisition for high blood-to-myocardium contrast.


Conventional 2D cine MRI, although routinely used for cardiac function assessment, is subject to known limits. Its lengthy series of breathholds needed for slice position and imaging at each individual cardiac view slows down workflow and is challenging for many patients. Also, the inevitable scan-to-scan variation in breathholding position in-between short-axis slices makes the so-obtained ventricular volumes error prone. Although advances in fast imaging have enabled 3D cine in a single breathhold [1], its slice coverage remains limited to cardiac ventricles only with large slice thickness. Recent efforts have attempted to resolve cardiac motion of the entire heart in high resolution during free breathing for arbitrary offline reformatting [2, 3]. However, its whole cardiac blood pool coverage significantly compromises in-flow blood enhancement and results in low blood-myocardium contrast. In this work, we aimed to evaluate the feasibility of free-breathing whole-heart 3D cine with high contrast using multiple thin slab acquisitions.


The entire heart is imaged in multiple slabs sequentially, while each slab covers a relatively thin volume to gain blood enhancement from through slab blood flow. Variable density k-t sampling was used for highly accelerated acquisition. For robustness to respiratory motion, radial vieworder with golden angle increment was employed both in [ky, kz] and along time. Respiratory gating was performed prospectively on a per cardiac phase basis during acquisition.

In reconstruction, kat ARC [4] was used to complete k-space utilizing spatiotemporal correlation. Respiratory motion was further suppressed by incorporating motion error of each k-space acquisition into the k-t reconstruction as a regularization term [5] based on simultaneously recorded bellow signal. Specifically, such regularization balances between data synthesis accuracy and motion suppression, effectively recovering k-space more reliant on data with less motion, while utilizing all acquired data for high gating efficiency.

4 healthy volunteers were scanned on a GE 1.5T 450w scanner. 3D cine imaging was performed with isotropic spatial resolution of 2.0 mm and a temporal resolution of ~50ms. 6 slabs were used to cover the entire heart with ~20 slices/slab and 9 overlapping slices to reduce venetian shading. Data were acquired in free-breathing with a gating efficiency of 60% and an acceleration factor of 8. To study the impacts of orientation to contrast and scan time, imaging was performed at 3 different orientations, 1. short-axis with slabs perpendicular to the main ventricular flow, 2. axial with the smallest phase-encoding and slice FOV and 3. axial-sagittal (an axial plane tilted in a coronal localization view to be perpendicular to the ventricular long axis) with the smallest phase encoding FOV and relative strong through slab flow. Signal in left (LV) and right ventricular (RV) blood pool and septum was measured in a 4 chamber reformat and contrast was calculated as the average blood pool signal at systole and diastole divided by the signal in septum.


Figure 1 compares images obtained from the same subject. Breathheld 3D cine covers only ventricles and suffers from poor slice resolution in reformat. Whole-heart 1-slab 3D cine exhibits low contrast. The proposed multi-thin-slab acquisition greatly improves the contrast and enables arbitrary reformats to different views in the whole heart. The average blood-to-myocardium contrast is 6.30, 5.35 and 6.23 for LV and 5.78, 5.26, 5.98 for RV at short axis, axial and axial-sagittal, respectively, all much higher than the contrast with conventional 1-slab acquisition (LV: 2.39, RV: 2.40). The average scan time is about 6:30min, 5:00min and 5:40min at short axis, axial and axial-sagittal orientations, respectively.


This work preliminarily evaluated a free-breathing multi-thin-slab method for whole-heart 3D cine imaging with isotropic resolution. Our results show that the proposed respiratory gating method effectively suppressed respiratory motion for free breathing imaging and the proposed multi-thin-slab scheme addressed the low contrast issue with single slab 3D cine. A tilted axial (axial-sagittal) acquisition provided relatively the best compromise between contrast and scan time and did not require additional scouting scans for slice positioning.


No acknowledgement found.


1. Kozerke S, MRM 2004. 2. Uribe S, MRM 2007. 3. Li F, MRM 2016. 4. Lai P, ISMRM 2009. 5. Lai P, ISMRM 2016


Figure 1. From left to right: cine images acquired using 2D cine, Breathheld (BH) 3D cine, whole-heart (WH) 1-slab 3D cine, multi-thin-slab (MSlab) 3D cine in short axis, axial and axial-sagittal. The first two and last two rows show the short axis and 4 chamber view images, respectively.

Proc. Intl. Soc. Mag. Reson. Med. 26 (2018)