In vivo cardiac Diffusion Tensor Imaging (DTI) has evolved significantly in recent years. Imaging of the complex cardiac fiber architecture in healthy^1-3 and diseased human hearts^4-7 has been reported. Ex vivo DTI in animal models of failing hearts has indicated the occurrence of alterations in the cardiac fiber network^8,9. The objective of the present work was to study the dynamic changes of myocardial fiber aggregate architecture in patients with dilated cardiomyopathy (DCM) using dual-phase cardiac DTI and compare findings with data from healthy controls. Nine DCM patients (age 60±14 years, weight 75±16kg, heart rate 68±12beats/min) and 9 healthy controls (age 50±6years, weight 68±13kg, heart rate 63±7beats/min) were studied using dual-phase stimulated echo acquisition mode (STEAM) DTI on 1.5T Philips Achieva systems (Philips Healthcare, Best, The Netherlands) equipped with 32 channel cardiac receiver arrays. The study protocol was approved by the local ethics committees and written informed consent according to institutional guidelines was obtained prior to imaging. DTI acquisition was ECG triggered to peak systole and mid diastole. Quiescent systolic and diastolic time points were determined based on cine images acquired with a temporal resolution of 10ms. Diffusion weighted imaging was performed using respiratory navigator-gated breath-holding with an acceptance window of 5 mm. The imaging plane was placed in short axis view orientation at a mid-ventricular level with the following imaging parameters: field-of-view (FOV): 309×129mm², in-plane resolution: 2.5×2.5mm², slice thickness: 8mm, TE/TR 18ms/2R-R intervals, partial Fourier factor 0.65. Ten optimized diffusion directions¹⁰ with a b-value of 350s/mm² were acquired per breathhold. A total of eight signal averages per diffusion encoding direction was collected resulting in eight breathholds per slice. Prior to tensor calculation, diffusion weighted images were aligned by affine image registration¹¹. Systolic diffusion tensors were corrected for myocardial strain as described previously³. To assess cardiac function, a contiguous stack of short-axis cine images ranging from apex to base (spatial/temporal resolution:1.6x1.6x8mm³/30ms) was acquired. Three-dimensional tagging data were acquired for diffusion tensor strain correction4 and cardiac motion analysis. Motion data were obtained during breath holding using complementary spatial modulation of magnetization (CSPAMM, spatial/temporal resolution: 3.5x7.7x7.7mm³/18ms, FOV: 108x108x108mm³)¹². Geometrical stack alignment of all tagged volume images was performed by incorporating navigator offsets and rigid image registration. Data analysis was performed on the local helix elevation (helix angle) and the corresponding normalized slope from endo- to epicardium was computed (helix angle slope) (Fig1 a,b). Cardiac function was assessed based on left ventricular ejection fraction (LVEF), maximum torsion¹³, longitudinal, radial and circumferential strain as derived from 3D tagging data.Figure 2 compares helix angle maps in diastole and systole in a DCM patient relative to data of a healthy control. In the DCM case steeper helix angle slopes are observed. In Figure 3 and 4 all data points are pooled. DCM patients show steeper diastolic helix angle slopes when compared to healthy controls (1.16±0.35 vs. 0.85±0.10, p<0.01). While a significant increase in helix angle slope from diastole to systole was found in the controls (0.85±0.10 vs. 1.06±0.17, p<0.001), no significant difference was seen in the DCM patient population (1.16±0.35 vs. 1.15±0.36, p=0.66). LVEF in the DCM patients was significantly lower compared with healthy controls (37±12% vs. 60±4%, p<0.001). Similarly, peak cardiac torsion values in patients were found to be significantly lower compared to the control subjects (0.17±0.10 °/mm vs. 0.30±0.05 °/mm, p<0.001). Figure 4 summarizes the results of torsion and strain parameters. All strain metrics reveal a significant attenuation in the DCM patients when compared to data of the control group.Dual-phase cardiac diffusion tensor imaging has successfully been applied to probe the dynamic changes of fiber aggregate helix slopes between diastole and systole in DCM patients relative to controls. The steeper diastolic helix angles in DCM patients relative to controls are associated with larger pre-stretch of the left ventricle. It is speculated that this larger pre-stretch alongside with reduced myocyte shortening compromises the ability to dynamically reorient fiber aggregates during systolic contraction as evidenced by the data of this study. The finding may be seen as a hallmark of overall reduced mechanical work as measured in this patient population. However, further studies and analyses are warranted to elucidate whether the changes in dynamic fiber orientation are compensatory or maladaptive depending on the stage of DCM progression. ES and CvD contributed equally to this work.