Introduction

Cardiovascular diffusion tensor imaging (cDTI) has emerged as a technique for investigating the micro-architecture and orientation of the myocardium¹. Its clinical utility has been somewhat impeded by long scan times and the common practice of performing cDTI under breath-hold (BH) conditions. To improve the clinical utility of cDTI, a control system (CS) was implemented in a free breathing cDTI sequence to perform in vivo slice tracking with 100% respiratory efficiency². The sequence was tested on a clinical MRI scanner with standard gradients (45mT/m).

Methods

Briefly, the CS method uses a series of 90°-180° crossed-pair navigators to repeatedly measure the position of the diaphragm during non-imaging segments. This is used as an input to a control system, which then predicts the diaphragm position and prospectively corrects the slice position during imaging segments (Figure 1). The slice position is computed assuming a linear factor between diaphragm and heart displacement³.
Six healthy volunteers underwent cDTI using a 3 T Skyra (Siemens, Erlangen, Germany). Subjects were scanned using an ECG-gated second-order motion compensated spin echo (M2SE) sequence with two different respiratory motion compensation techniques – multiple BH and free breathing (FB) using the CS. For both techniques, diffusion weighted images (DWI) were acquired using b-values of 50 and 450s/mm² with 12 diffusion directions and 5 repetitions. The order in which the FB and BH scans, and b-values, were acquired was randomised for each subject. Scan parameters: TR/TE – 1 R-R/86ms; matrix size – 136x50; pixel size – 2.57x2.57mm² (interpolated: 1.29x1.29mm²); slice thickness – 8mm; bandwidth – 2298Hz/pixel; GRAPPA – 2. Cine imaging was used to position three short axis slices in the basal, mid, and apical regions of the heart and optimise the trigger delay to acquire the DWI in late systole.
All image analysis and diffusion tensor calculations were computed offline using a set of in-house MATLAB tools⁴ (Mathworks, Natick, USA). Image registration using affine transformations, heart-rate correction, and repetition averaging were performed. The mean diffusivity (MD), fractional anisotropy (FA), helix angle (HA), and secondary eigenvector angle (E2A) were calculated voxelwise from the diffusion tensor for the entire myocardium. All statistical analyses were performed with SPSS v27 using a multifactor ANOVA corrected for repeated measurements.

Results

Each set of 12 DWIs were collected in single breath-hold acquisitions of approximately 15 seconds for a total of 30 breath-holds, approximately 20 minutes including short breaks between scans. For the CS, all DWIs were collected in one continuous scan per slice under free-breathing conditions and completed in approximately 15 minutes including pre-training time³.
There were no significant differences found between the two respiratory motion compensation techniques for MD, FA, E2A, or HA (Figures 2 and 3).

Discussion and Conclusion

The M2SE cDTI sequence was implemented on a 3 T clinical scanner using a prospective motion correction control system.
The results of the diffusion parameters were similar to previously published results. The BH and CS techniques produced very similar results with no significant differences (Figure 4).
These results show that the prospective respiratory motion correction control system technique can play a part in improving the clinical viability of cDTI, allowing for the reduction of exam times, enabling long exams to be performed under free-breathing conditions, as well as opening the door to performing multi-slice acquisitions and increasing the number of diffusion directions.