Cardiac Phase-resolved Late-Gadolinium Enhancement Imaging
Sebastian Weingärtner1,2,3, Burhaneddin Yaman1,2, Chetan Shenoy4, Marcel Prothmann5, Felix Wenson5, Jeanette Schulz-Menger5,6, and Mehmet Akcakaya1,2

1Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, United States, 2Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, 3Computer Assisted Clincial Medicine, Heidelberg University, Mannheim, Germany, 4University of Minnesota, Minneapolis, MN, United States, 5Working Group on Cardiovascular Magnetic Resonance Imaging, Max-Delbrück-Centrum and Charité -Medical University Berlin, Berlin, Germany, 6Department of Cardiology and Nephrology, HELIOS Klinikum Berlin-Buch, Berlin, Germany


Late Gadolinium Enhancement (LGE) is commonly acquired during a single end-diastolic phase with inversion-recovery contrast that nulls healthy myocardial tissue. In this work, we propose a method for acquisition of cardiac phase-resolved LGE images based on an ECG triggered Look-Locker experiment with continuous FLASH imaging. Semi-quantitative evaluation of this pulsed-inversion recovery allows synthetization of LGE image contrast for all cardiac phases. Accurate functional depiction with temporal resolution up to 60 ms is obtained in healthy subjects at 3T. Images of 20 patients on a clinical 1.5T scanner show promising depiction of focal scar at a temporal resolution of 80ms.


Late Gadolinium Enhancement (LGE) imaging is clinically established as gold-standard for characterization of scar and focal fibrosis. In LGE inversion-recovery imaging is used after contrast agent administration to depict contrast agent retention. However, in order to obtain the desired contrast that nulls the healthy myocardium, imaging is commonly restricted to a single cardiac phase.

In this study, we develop a method for acquisition of cardiac phase-resolved LGE images. We extend a recently proposed dynamic T1 mapping approach1 to semi-quantitative imaging after contrast administration, which allows for retrospective synthesization of LGE contrast for all cardiac phases.


Sequence: The proposed sequence employs multiple ECG-triggered Look-Locker experiments (Fig. 1). First, the magnetization is driven to steady-state. Following an inversion, FLASH readouts are continuously performed, acquiring k-space data in a segmented manner. This pulsed-recovery curve spans multiple (typically 2) heart-beats, enabling the acquisition of multiple inversion times per cardiac phase. Upon re-reaching the steady-state the magnetization is re-inverted, and the next k-space segment is acquired. After completion of these segments, the entire experiment is repeated by varying the inversion pulse position throughout the cardiac cycle, in order to finely sample the recovery. Semi-quantiative information is then obtained using a 2-parameter model (S=A(1–2exp(-t/T1*)) and used on a per-voxel basis to generate LGE contrast as SLGE=A(1–2exp(-TI/T1*)). Here TI is a retrospectively chosen inversion time that is the same for all cardiac phases.

In Vivo Imaging: 4 healthy subjects (39±17years, 3males) were scanned on a 3T scanner (Siemens Prisma), with the following image parameters: TR/TE/FA=5/2.6ms/3°, FOV/resolution=300x225/1.9x1.9mm2, GRAPPA=2, slice=10mm, water-selective excitation, temp. resolution=60ms, breath-hold=17-19s. Additionally, 20 patients (50±16years, 13males, indications: suspected CAD:6, known CAD:5, myocarditis:7, DCM:1, HCM:1) were imaged on a clinical 1.5T scanner (Siemens Avanto Fit). Due to decreased gradient performance and lower baseline SNR, following sequence parameters were adopted at 1.5T: TR/TE/FA=6.7/3.2ms/6°, resolution=2.1x2.1mm2, temp. resolution=80ms, breath-hold=15-18s.


Figure 2 illustrates image synthesization in a healthy subject at 3T. Despite wide variations in the baseline image contrast, a consistent T1* value is obtained across the cardiac cycle. This leads to robust nulling of the healthy myocardium for all phases of the synthesized LGE images.

Figure 3a shows images acquired at 3T, depicting sharp delineation of the myocardium against blood pools without visually apparent temporal blurring. Patient images at 1.5T show thorough nulling of healthy myocardium throughout the cardiac cycle (Fig. 3b), despite lower baseline SNR, while simultaneously representing the myocardial contractility. Across the patient population, 20 images were of diagnostic image quality, while three cases suffered from fold-over artifacts due to sub-optimal FOV adjustment. One case exhibited two corrupted phases, due to sustained arrhythmia. Images of patients with known CAD show clear enhancement of the infarcted area (Fig. 4 lateral, Fig. 5 antero-septal). Retrospective choice of the inversion time allows for optimized scar-myocardium and scar-blood contrast, allowing good delineation against both. Additionally, the phase-resolved images depict scar deformation throughout the heart-beat.


In this work, functional LGE imaging was enabled by semi-quantitative assessment of the longitudinal relaxation time in a phase-resolved manner. As previously proposed, quantitative T1 information can be used on a per-voxel basis to synthesize LGE images2 that provide accurate quantification of scar area3. Retrospective adjustment of the LGE contrast may even circumvent the problem of incomplete myocardial nulling in the presence of incorrect inversion times, as commonly observed in magnitude based LGE images.

Cross-comparison of end-diastolic LGE images with other cardiac phases has been shown to improve diagnostic confidence4, especially for assessing scar transmurality5. Confounding depiction of scar in standard LGE imaging may also occur when a short-axis slice is adversely placed to intersect with the right ventricular blood-pool causing an apparent myocardial hyper-enhancement. While LGE images can be acquired at multiple phases in different heart-beats, this requires long scan times, and image comparison is hampered by differential breath-holding. The proposed technique improves upon these shortcomings by efficiently assessing myocardial viability through the entire cardiac cycle.

The proposed technique may also be used to simultaneously evaluate cardiac function and tissue viability in a single scan, if sufficient temporal resolution is ensured. Advanced regularizations to improve spatio-temporal resolution6 are currently being investigated and warrant further research.


The proposed method for cardiac phase-resolved LGE imaging enables assessment of scar and focal fibrosis through the cardiac cycle. Excellent blood-myocardium contrast at a temporal-resolution of 60 ms is achieved at 3T. Patient images at 1.5T show clear depiction of the scar, while simultaneously displaying myocardial contraction and providing visualization of scar displacement.


Grant support: NIH R00HL111410, NIH P41EB015894 and NSF CCF-1651825.


  1. Weingärtner S, Shenoy C, Rieger B, Schad LR, Schulz-Menger J, Akçakaya M. Temporally resolved parametric assessment of Z-magnetization recovery (TOPAZ): Dynamic myocardial T1 mapping using a cine steady-state look-locker approach. Magn Reson Med. 2017;doi:10.1002/mrm.26887
  2. Varga-Szemes A, van der Geest RJ, Spottiswoode BS, Suranyi P, Ruzsics B, De Cecco CN, Muscogiuri G, Cannaò PM, Fox MA, Wichmann JL, Vliegenthart R, Schoepf UJ. Myocardial Late Gadolinium Enhancement: Accuracy of T1 Mapping-based Synthetic Inversion-Recovery Imaging. Radiology. 2016;278(2):374-82
  3. Varga-Szemes A, van der Geest RJ2 Schoepf UJ, Spottiswoode BS, De Cecco CN, Muscogiuri G, Wichmann JL, Mangold S, Fuller SR, Maurovich-Horvat P, Merkely B, Litwin SE, Vliegenthart R, Suranyi P. Effect of inversion time on the precision of myocardial late gadolinium enhancement quantification evaluated with synthetic inversion recovery MR imaging. Eur Radiol. 217;27(8):3235-43
  4. Schuster A, Chiribiri A, Ishida M, Morton G, Paul M, Hussain S, Bigalke B, Perera D, Nagel E. End-systolic versus end-diastolic late gadolinium enhanced imaging for the assessment of scar transmurality. Int J Cardiovasc Imaging. 2012;28(4):773-81
  5. Matsumoto H, Matsuda T, Miyamoto K, Shimada T, Hayashi A, Mikuri M, Hiraoka Y. Late gadolinium-enhanced cardiovascular MRI at end-systole: feasibility study. AJR Am J Roentgenol. 2010;195(5):1088-94
  6. Moeller S, Weingartner S, Akcakaya M. Multi-scale locally low-rank noise reduction for high-resolution dynamic quantitative cardiac MRI. Conf Proc IEEE Eng Med Biol Soc. 2017 Jul;2017:1473-1476.


The proposed sequence is based on the acquisition of multiple ECG-triggered Look-Locker experiments. After reaching steady-state, an inversion-pulse is played following a pre-defined period after detection of the R-wave. The recovery is then sampled using continuous FLASH pulses to collect segmented k-space data. Once the magnetization recovers to steady-state the next inversion pulse is played to collect the next k-space segment. These look-locker experiments are repeated with varying time delay between the R-wave and the inversion pulse to provide multiple inversion times for each cardiac phase. A semi-quantiative fit allows synthesization of LGE contrast with nulling of the healthy myocardium.

(For electronic version: Click image to play in CINE view) Post-processing pipeline of the proposed phase-resolved LGE sequence, demonstrated with images acquired in a healthy subject at 3T. Various baseline contrasts per cardiac phase are used on a per-voxel basis to generate semi-quantiative T1* maps. The information of the fit is then used to generate LGE images with a retrospectively chosen inversion time, which is kept constant for all phases. The resulting LGE images with a spatio-temporal resolution of 1.9x1.9mm2 with 60ms temporal resolution, depict thorough suppression of the healthy myocardium, while providing sharp delineation against the blood-pool.

(For electronic version: Click image to play in CINE view) Phase-resolved LGE images acquired in a healthy subject at 3T and two patients at 1.5T. All images are acquired with an in-plane resolution of 2.1x2.1mm2. Images acquired at 3T with a temporal resolution of 60ms show good functional depiction with no visually apparent blurring. Despite lower baseline SNR and decreased temporal resolution of 80ms, images of two patients with suspected CAD acquired at 1.5T provide robust LGE contrast throughout all cardiac phases, while functionally depicting the myocardial contraction.

(For electronic version: Click image to play in CINE view) Phase-resolved LGE image of a patient with known history of CAD and myocarditis (41 years, female) acquired at 1.5T at 80ms temporal resolution. Clear depiction of scar in the mid-anteriorlateral segment is achieved in all phases against thoroughly nulled healthy myocardium. Retrospective inversion time choice allows to optimize scar contrast in comparison to the reference LGE image.

(For electronic version: Click image to play in CINE view) Images acquired in a 34 year old man, with known history of CAD at 1.5T and 80ms temporal resolution. Phase-resolved LGE images allow good assessment of the scar in the inferior segment in all cardiac phases, additionally depicting scar displacement through the cardiac cycle.

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