Highly Accelerated, Single/Multi-Breathhold 3D Cine Imaging with Whole Heart Coverage in Patients at 1.5T

Background: Cine imaging is a cornerstone of CMR and the basis for calculation of parameters such as LVEF [1]. The current standard for cardiac cine MRI is slice-by-slice acquisition of prospectively planned views, leading to long and uncomfortable exams with 20 or more breathholds as well as complex operator input. A method for isotropic 3D cine CMR with whole-heart coverage is desirable for improving patient experience and reducing scan complexity but is difficult to achieve with realistic breathhold times [2]. A 3D cine framework combining data from multiple breathholds has been proposed previously [3], which in this work is evaluated and expanded to a flexible single- or multi-breathhold exam (1-BH or 2/3-BH) that is adjustable depending on resolution requirements and patient breathhold ability.

Methods: This work uses a research sequence and reconstruction framework for 3D cine featuring a 2-region sampling pattern and an iterative reconstruction with temporal regularization and integrated motion-compensation between breathholds [3]. Acceleration factors for inner and outer k-space regions were 2x2 and 6x6 (PExPA), respectively. For 2/3-BH scans, motion compensation between breathholds was based on single-phase 3D images reconstructed from data pooled across phases within each breathhold (see reference [3]). The full 3D cine dataset was reconstructed using spatiotemporal wavelet regularization and integrated motion compensation. 1-BH and 2-BH data were acquired before contrast agent administration, with parameters (1/2-BH): 2.5/2.0mm isotropic resolution, TE 1.1/1.3ms, TR 2.6/2.9ms, FA 40 deg, bSSFP readout, 53/52ms temporal resolution, retro-gating to 20 phases, sagittal orientation, 72-96 slices per slab.

Results: 6 patients (1 Female, age 34±9y, BMI 27±6) were scanned at 1.5T (MAGNETOM Sola, Siemens Healthineers, Forchheim, Germany) with clinical indications of congenital heart disease and Myocarditis. Average breathhold durations were 18.7s and 17.3s for 1- and 2-BH scans respectively. Function parameters LVEF, LVEDV, RVEF and RVEDV as shown in Table 1 were calculated from 3D cine datasets after reformatting to short-axis slices with thickness matched to the clinical reference using Circle (Circle CVI Inc. Calgary, CA). Good agreement with standard clinical assessment was seen for both 1- and 2-BH scans, with slightly lower LV- and RVEF for the 3D approach (not statistically significant). Figure 3 shows an additional 3-BH scan in a patient with limited breathhold capabilities, underscoring the efficacy of inter-breathhold motion compensation.

Conclusion: A new framework for isotropic 3D Cine CMR has been evaluated in patients, allowing for rapid assessment of cardiac function in as little as a single breathhold. While evaluation in a larger cohort is warranted, functional parameters were obtained reliably using both single- and multi-breathhold scans, which were adaptable to resolution and breathhold duration requirements.

Functional Parameters

Cine scan type (reformatted in-plane resolution)	LVEF +- Std Dev [%]	LVEDV [ml]	RVEF [%]	RVEDV [ml]
Reference 2D (1.4 mm)	56.2 +- 4.9	206.2 +- 29.8	48.2 +- 2.1	233.7 +- 38.2
1-BH 3D (2.5 mm)	54.3 +- 4.9	192.3 +- 26.9	43.5 +- 3.1	227.5 +- 46.7
2-BH 3D (2.0 mm)	54.0 +- 4.7	198.0 +- 27.7	45.7 +- 3.1	239.5 +- 40.4

Example for Diastolic (top row) and Systolic frames (bottom row) from a 2-BH whole heart 3D cine scan. Images were reformatted from the full 3D+time dataset, from left to right: Short axis, 2-chamber, 4-chamber, segmented short axis and sagittal MIP.
Diastolic short axis frames for a CHD patient with difficulty holding their breath. While the 15s 1-BH scan at 2.5mm was diagnostic (a), a 3-BH scan was added at 1.7mm isotropic resolution and 12s per BH (b-d). Images (c) and (d) show the 3-BH images in short axis and 4-chamber views with inter-BH motion compensation, and (b) shows the artifacts arising if the latter is omitted during reconstruction.