Researchers from the Institut Langevin at the ESPCI ParisTech in France have developed a new method for ultrasound-based fibre angle estimation – elastic tensor imaging (ETI) – and compared its performance to that of the gold standard MR diffusion tensor imaging (DTI). In vitro and in vivo studies of porcine and ovine hearts reveal that fibre orientation estimated by ETI is comparable to that measured by DTI (Phys. Med. Biol. 57 5075).
"The ultrasound ETI proposed in this study offers key features of conventional ultrasound imaging, including real-time acquisition, patient compatibility and wide availability in clinical practice," explained research associate Wei-Ning Lee. "In addition, ETI may provide both the tissue elasticity and fibre structure, whereas DTI only maps fibre architecture."
Shear wave imaging
ETI uses shear waves, which are polarized perpendicular to their direction of propagation. In anisotropic media such as myocardium, shear waves travel faster along the fibres than across them, with a speed dependent upon the angle between the propagation and fibre directions. Lee and colleagues previously demonstrated that ultrasound-based shear wave imaging could characterize transmural (through the heart wall) fibre orientation. In this latest work, they employ tensor-based ETI to improve the robustness and precision of fibre angle estimation.
ETI comprises three key elements: generation of transient shear waves by focusing an ultrasound beam inside the tissue; shear wave speed estimation; and estimation of fibre angles. In this study, three focused beams were sequentially transmitted at different depths to generate shear waves across a large myocardial wall thickness. The generated waves were then imaged using a 2.8 MHz phased-array probe at an ultrafast frame rate of between 8000 and 12,000 frames/s.
To create shear waves in different directions, the ultrasound probe was mounted on an MR-compatible rotation device, which allowed rotation from –90° to 90° in 5° increments. At each of these 37 probe angles, the shear wave speed was estimated throughout the imaged myocardium, and corresponding fibre angles were determined using a least-squares method.
Immediately after ETI acquisition, the myocardial sample (still in the rotation device) was placed inside a 7T MRI scanner and diffusion was encoded in six directions. This co-registered imaging set-up enabled easy comparison between the fibre orientation determined by the two techniques.
Technique comparison
The researchers examined 10 myocardial samples from excised porcine and ovine hearts. One sample, for example, revealed transmural variation in fibre orientation from roughly 70° in the subendocardium (0–33% cardiac wall depth) to –50° in the subepicardium (67–100% depth). Good correlation was seen between ETI- and DTI-estimated fibre angles in all 10 myocardial samples.
They also computed fractional anisotropy (FA) values at each wall depth. "Fractional anisotropy estimated by ETI may indicate the transmural variation of tissue structure and stiffness," Lee explained. "Larger FA values indicate that the examined tissue has a preferential fibre direction and/or higher contrast in elasticity in principal directions. This may provide an additional diagnostic index."
Average ETI-estimated FA values were higher than DTI-estimated ones in all wall regions: 0.51 versus 0.30 in the subendocardium; 0.38 versus 0.29 in the midwall region (33–67% depth); and 0.34 versus 0.27 in the subepicardium. Average ETI-estimated FA values decreased from subendocardium to subepicardium by 33%, while the corresponding DTI-estimated FA values decreased by 10%.
This disparity was attributed to the different physical phenomena underlying each method. DTI measures water diffusion rate within tissue, which is related to cellular arrangement, while ETI estimates shear wave speed, which is associated with both fibre orientation and the local tissue stiffness. The higher FA values estimated from ETI suggest that elastic anisotropy is greater than diffusion anisotropy in the normal myocardium assessed in this study.
The team also performed in vivo ETI on a beating ovine heart, following which, the heart was excised and examined in vitro by DTI and ETI. Differences were noted between in vivo and in vitro fibre-angle estimations, possibly due to physiologic discrepancies between the two configurations, which may also cause the fibre orientation to alter after excision. The fibre angles estimated by in vitro ETI and DTI were well correlated.
Finally, the researchers examined the number of probe angles needed for robust tensor-based fibre angle estimation. Seven angles (sampled at every 30° between –90° and 90°) were sufficient to estimate fibre orientation with comparable accuracy to that estimated from 37 rotation angles.
The authors conclude that myocardial fibre orientation estimated by ETI correlates well with that measured by DTI. They are now working to evaluate ETI's clinical value, by characterizing pathological myocardium in the presence of fibre disarray. "We have performed several pilot in vivo experiments on normal and transgenic mice with myocardial hypertrophy," said Lee.
• Related articles in PMB
Ultrasound elastic tensor imaging: comparison with MR diffusion tensor imaging in the myocardium
Wei-Ning Lee et al Phys. Med. Biol. 57 5075
Lamb wave dispersion ultrasound vibrometry (LDUV) method for quantifying mechanical properties of viscoelastic solids
Ivan Z Nenadic et al Phys. Med. Biol. 56 2245