Dual-Phase Cardiac Diffusion Tensor Imaging with Strain Correction

Abstract

PurposeIn this work we present a dual-phase diffusion tensor imaging (DTI) technique that incorporates a correction scheme for the cardiac material strain, based on 3D myocardial tagging.Methods In vivo dual-phase cardiac DTI with a stimulated echo approach and 3D tagging was performed in 10 healthy volunteers. The time course of material strain was estimated from the tagging data and used to correct for strain effects in the diffusion weighted acquisition. Mean diffusivity, fractional anisotropy, helix, transverse and sheet angles were calculated and compared between systole and diastole, with and without strain correction. Data acquired at the systolic sweet spot, where the effects of strain are eliminated, served as a reference.ResultsThe impact of strain correction on helix angle was small. However, large differences were observed in the transverse and sheet angle values, with and without strain correction. The standard deviation of systolic transverse angles was significantly reduced from 35.9±3.9° to 27.8°±3.5° (p<0.001) upon strain-correction indicating more coherent fiber tracks after correction. Myocyte aggregate structure was aligned more longitudinally in systole compared to diastole as reflected by an increased transmural range of helix angles (71.8°±3.9° systole vs. 55.6°±5.6°, p<0.001 diastole). While diastolic sheet angle histograms had dominant counts at high sheet angle values, systolic histograms showed lower sheet angle values indicating a reorientation of myocyte sheets during contraction.ConclusionAn approach for dual-phase cardiac DTI with correction for material strain has been successfully implemented. This technique allows assessing dynamic changes in myofiber architecture between systole and diastole, and emphasizes the need for strain correction when sheet architecture in the heart is imaged with a stimulated echo approach.

Evaluation of non‐Gaussian diffusion in cardiac MRI

“…In clinical diffusion imaging, non‐Gaussian models have the potential to help in assessing myocardial perfusion 19. However, gradient hardware in clinical scanners typically limit b‐values to 500 s/mm 2 46, 47, 48, reducing sensitivity to non‐Gaussian diffusion arising from finer tissue structures as demonstrated here. Stimulated echo approaches afford higher b‐values for a given maximum gradient amplitude due to long diffusion times.…”

“…Longer $\Delta$ would increase the interactions of water molecules with cellular restrictions and likely enhance non‐Gaussian effects, particularly in the directions of the second and third eigenvectors. However, stimulated echo sequences are subject to myocardial strain, necessitating strain correction 48 or limiting imaging to temporal “sweet spots” 49. Higher b‐values result in lower signal, thus higher SNR is needed to avoid noise floor bias.…”

Diffusion Tractography of the Entire Left Ventricle by Using Free-breathing Accelerated Simultaneous Multisection Imaging

“…Acquisition parameters were field of view of 360 × 200 mm, resolution 2.5 × 2.5 mm, section thickness of 8 mm, in-plane generalized autocalibrating partially parallel acquisition rate 2 (12), echo time of 34 msec, b values of 0 and 500 sec/mm 2 , 10 diffusion-encoding directions, and eight magnitude averages (repetitions). Twelve short-axis sections were acquired at the systolic sweet spot (160 msec after the R wave) to mitigate strain effects (3, 13). SMS excitation was followed by a blipped-CAIPI readout by using a section-phase-section zonal excitation scheme (Fig 1a).…”

Myocardial Scar Delineation Using Diffusion Tensor Magnetic Resonance Tractography

“…The entire left ventricle (LV) was imaged without slice gaps by acquiring 12 contiguous short‐axis slices from base to apex 23. Images were acquired in the systolic and diastolic sweet spots of the cardiac cycle to mitigate strain effects 18, 22…”