In-vivo cardiac diffusion weighted imaging (DWI) parameters such as mean diffusivity (MD) and fractional anisotropy (FA) reflect myocardial tissue status. However, these parameters are sensitive to transient physiological changes such as perfusion (1–3), they depend on acquisition parameters, e.g. diffusion weighting strength and diffusion time (4,5), and their accuracy correlates with SNR (6). Interestingly, a recent study by McGill et al. (7) showed in-vivo spatial and transmural heterogeneity of cardiac diffusion parameters derived from single short axis slice STEAM DWI, which could be related to variations in myocardial microstructure, but also to SNR, partial volume, perfusion, and strain heterogeneity. The aim of this study was to investigate the origin of DWI-parameter heterogeneity using whole heart in- and ex-vivo SE-DWI. We hypothesize that observed heterogeneity is related to regional differences in perfusion signal fraction and transmural organization of the cardiac fiber architecture.

In-vivo cardiac DWI data was acquired with b-values of 0, 10, 20, 30 ,50, 100, 200, and 400 s/mm² with 6, 3, 3, 3, 3, 3, 3, and 24 gradient-directions, respectively. This gradient scheme allows for analysis using both Intra voxel incoherent motion (IVIM) and Diffusion Tensor Imaging (DTI) models (3,8,9). A noise map was acquired by switching off the RF and imaging gradients and used to calculate SNR values for the b=0 s/mm² images. In total nine healthy volunteers (5 Female, mean age 24 [range 22-34]) were imaged with a 3T scanner (Philips, Achieva, Release 5.1.7) using a 32-channel cardiac coil and a cardiac-triggered SE-EPI sequence in free breathing with asymmetric bipolar gradients (10) and additional flow compensation (11). Other imaging parameters were: TR: 14 heart beats, FOV: 280x150 mm², slices: 14, voxel size: 7x2.5x2.5 mm³, acquisition matrix: 112x48, SENSE: 2.5, partial Fourier: 0.85, trigger delay: 220 ms, and acquisition time: 12 min. Additionally nine datasets of formalin fixed porcine hearts were acquired (5) using a multi shot STEAM-EPI sequence (mixing time = 30ms) using Stejskal-Tanner gradients, a spatial resolution of 6x2x2 mm³, and b-values of 500, 1000, 2000, and 3000 s/mm² (30 gradient-directions per b-value).

Data processing was done using DTITools (Mathematica 10) and comprised the following steps: registration to correct for subject motion and eddy current deformations (in vivo: 2D b-spline, ex-vivo: 3D affine), Rician noise suppression, manual segmentation of the left and right ventricle (Figure 1A), calculation of the local myocardial coordinate system (LMCS) (Figure 1 B and C) and segmentation using the AHA-17 segment model (Figure 1D). Diffusion parameters were estimated using four methods. The first three methods were weighted linear least squares (WLLS) estimation using b=0 and 400 s/mm², b=200 and 400 s/mm² (to eliminate perfusion effects) and all b-values, respectively. The fourth method was IVIM corrected tensor estimation using all b-values (3,8). Transmural profiles of the diffusion parameters and SNR were obtained for 15 points along the myocardial wall using first order interpolation along the radial axes of the LMCS (green vectors in Figure 1C). 180 radial profiles were used per slice (Figure 1D).

Fits of the average whole heart signal for each of the nine in-vivo datasets using all four methods are shown in Figure 2. The green dashed line indicates the MD corresponding to the “perfusion signal free” fit using b=200 and 400 s/mm². Example per voxel fits of one in-vivo and one ex-vivo dataset are shown in Figure 3. Transmural profiles of the diffusion parameters, perfusion signal fraction (f) and helix angle (HA) are shown in Figure 4. Similar to results shown by McGill et al. (7) we found transmural heterogeneity of all parameters. However, estimation of the eigenvalues, MD and FA are strongly related to perfusion signal fraction but not SNR. With perfusion correction transmural heterogeneity decreases but remains present and is also visible in ex-vivo data. Minimal transmural values of the eigenvalues and MD and maximal transmural values of FA coincide with the position where the helix angle equals 0 (blue lines Figure 4). Figure 5 shows the spatial heterogeneity of the diffusion parameters and perfusion signal fraction of the in-vivo data before and after IVIM correction.In this study we have validated the transmural and spatial heterogeneity of diffusion parameters in whole heart cardiac DWI as first described by McGill et al. (7). The SE-DWI acquisition is insensitive to strain which can thus be excluded as an potential origin. Similar to previous results there is no correlation with SNR. Based on our results, transmural heterogeneity is partially explained by variations in transmural and spatial perfusion signal fraction and, partially seems to have its origin in cardiac architecture during systole.