Human MRI Data to Quantify Cardiac Performance

This data was acquired as part of the NSF funded project: "Quantifying Cardiac Performance by Measuring Myofiber Strain with Routine MRI" Awards No. 2205103 to Daniel B. Ennis and 2205043 to Luigi E. Perotti.

A total of 51 healthy volunteers (mean age 39±14 years) were prospectively enrolled between 2023 and 2025 in an Institutional Review Board (IRB)-approved study. Written informed consent was obtained from all participants prior to imaging. Demographic data are available upon request.

Figure 1: Demographic of Healthy Volunteers

All participants underwent a cardiac MRI examination on a 3T system (Vida Fit, Siemens Healthineers) using identical radiofrequency coils and imaging protocols.

Cine Imaging. Cardiac functional assessment was performed using a breath-held, retrospectively ECG-gated balanced steady state free precession (bSSFP) cine sequence with the following parameters: 6/8 partial Fourier, rate-2 parallel imaging, matrix size = 256x200, in-plane spatial resolution = 1x1 mm², 25 cardiac phases, TE/T_Res = 1.4/3.3 ms, flip angle = 10~30º, bandwidth = 997 Hz/Px, and slice thickness = 8 mm. Contiguous short-axis slices with zero gap were acquired to ensure full left ventricular (LV) coverage. Additionally, five long-axis views were obtained, beginning with a standard 3-chamber orientation (0°) and rotating 37° between successive slices about the long-axis of the heart (from apex to mitral valve plane).

Figure 2: Representative cine CMR images from left to right and from top to bottom are mid-ventricular short-axis slice and five long-axis slices.

Cine DENSE Imaging. LV myocardial motion was evaluated using a breath-held, prospectively ECG-gated Displacement Encoding with Stimulated Echoes (DENSE) sequence, incorporating xyz-displacement encoding (k_e = 0.10 cycles/mm, k_d = 0.08 cycles/mm), 3-point phase cycling, spatial resolution = 2.5x2.5x8 mm³, TE/T_Res = 1.08/15 ms, two spiral interleaves, and a single average. Short-axis and long-axis slices were acquired at locations identical to those used in cine imaging.  xyz-displacement encoding (ke = 0.10 cycles/mm, kd = 0.08 cycles/mm), 3-point phase cycling, spatial resolution = 2x2x8 ~ 2.5x2.5x8 mm3, TE/TRes = 1.08/15 ms, two spiral interleaves, and a single average. 

Figure 3: Representative cine DENSE CMR images from left to right are magnitude-reconstructed images as well as phase images encoded for x-displacement, y-displacement, and z-displacement.

Cine Tagged MRI. All subjects also underwent grid-tagged cine MRI of the mid-ventricular LV short-axis, acquired during breath-hold with retrospective ECG gating. Imaging parameters included: spatial resolution = 1×1×8 mm³, TE/T_Res = 3.01/45 ms, 30 reconstructed cardiac phases, grid tag spacing = 7 mm, Bandwidth = 401 Hz/Px, and GRAPPA-based acceleration (factor = 3; reference lines PE = 45). The slice location was chosen to be three mid- ventricular short-axis slices at the same locations of acquired cine DENSE and cine CMR images.

Figure 4: Representative cine tagged CMR images from left to right are three short-axis mid-ventricular slices.

Native T1 mapping. Pre-contrast myocardial T1 values were measured using the vendor-provided Modified Look-Locker Inversion recovery (MOLLI) 5(3s)3 scheme with motion correction (MOCO). Imaging was performed during breath-hold with ECG retrospective gating. Parameters included: non-selective inversion pulse, SSFP single-shot acquisition, 20º excitation flip angle, 7/8 partial Fourier, rate-2 parallel imaging, matrix size = 192×132, in-plane resolution = 1×1 mm², slice thickness = 8 mm, bandwidth = 1085 Hz/Px, minimum inversion time (TI) of 100 ms (incremented by 80 ms), TE/TR = 1.01 ms/2.44 ms. Three short-axis slices were acquired in the same anatomical positions as the tagged images.

Figure 5: Representative Native T1 mapping images from left to right are three short-axis mid-ventricular slices.

Each subject's data are provided in a compressed (*.zip) file containing de-identified raw imaging datasets (DICOM).

To apply for access to the dataset, applicants should prepare the following:

• Institutional affiliation and contact details (including PIs)
• A brief summary of the intended research

Application Process:

1. Complete the online application form using this link.
2. Applications will be reviewed, and feedback (if any) will be provided via email.
3. Upon approval, download links for the dataset will be shared directly with the applicant.
4. If requested, group demographics information will also be provided.

Researchers utilizing this dataset should cite the following publication:

Coming Soon!

DOI: Coming Soon!

This material is based upon work supported by the National Science Foundation under Awards No. 2205103 and 2205043.

Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

• ML Toolbox for automatically analyze Cine Tagged MRI images:

Loecher, Michael, Luigi E. Perotti, and Daniel B. Ennis. "Using synthetic data generation to train a cardiac motion tag tracking neural network." Medical image analysis 74 (2021): 102223. DOI: https://doi.org/10.1016/j.media.2021.102223