Tissues such as cortical bone, tendons, ligaments and menisci have short T₂ and have little or no signal with conventional clinical MR sequences. Water diffusion in these tissues is inaccessible. Ultrashort echo time (UTE) sequences with TEs down to 8 µs have been employed to image the short T₂ tissues. The combination of UTE and stimulated echo diffusion preparation (UTE-DWI) allows water diffusion to be evaluated in short T2 tissues. In this study we report the use of a novel stimulated echo (STE) diffusion prepared 3D UTE Cones sequence¹ (3D UTE-Cones-DWI) for assessment of water diffusion in the Achilles tendon using a clinical 3T scanner. The magic angle dependence of apparent diffusion coefficients (ADC) determined through this sequence was also investigated.

The remarkable feature of STE-diffusion preparation is that it allows use of long diffusion time which can be much longer than the T₂ of the tissue being studied². This is extremely important not only for imaging but also for the measurement of ADCs in short T₂ tissues, such as the Achilles tendon. The sequence diagram of the 3D UTE-Cones-DWI is shown in Figure 1. The STE-diffusion preparation pulse cluster is composed of two pairs of tip-up and tip-down RF pulses. The composite pulse is insensitive to B1 inhomogeneity. Useful b-values can be obtained by using a large Tmix time and a short diffusion gradient duration. This is of benefit when imaging short T₂ tissues. After the STE diffusion preparation, a train of cones k-space spokes is employed for fast data acquisition. Before each STE-diffusion data acquisition, a long wait time is needed for longitudinal signal recovery.

Cadaveric Achilles tendon samples dissected from human ankle specimens (n=3) were harvested for this study. Data were acquired with the 3D UTE-Cones-DWI on a clinical 3T scanner (GE Healthcare Technologies, Milwaukee, WI). The Tmix and duration time of each diffusion gradient were fixed at 120ms and 4ms, respectively. A slice-direction diffusion gradient with four amplitudes (i.e. 10, 20, 30, 40 mT/m) was used to generate four b-values (i.e. 14, 57, 127, 227 s/mm2). Other imaging parameters were as follows: flip angle=10°, TE=32 µs, 5.4 ms per-spoke, 32 spokes per-TR, TR=1500 ms, FOV=8cm, matrix=128*128, slice thickness=4mm, slice number=10. In addition, multiple TE data were acquired for fast measurement of T₂* values using a 2D radial-UTE sequence³ (flip angle=10°, FOV=8×8cm², matrix=128×128, slice thickness=4mm, 18 TEs = 10µs, 0.1, 0.2, 0.3, 0.4, 0.6, 0.8, 1, 1.5, 2, 3, 4, 5, 6, 7, 8, 9, 10 ms, TR=100ms). The above two protocols were applied to the tendon sample five times with five different angular orientations (i.e. 0°,30°,57°,69°,90°) between the tendon fiber direction and the B0 field.

Both the multiple b-value diffusion data and multiple TE data were processed using a non-linear curve-fit function (i.e. lsqcurvefit) in Matlab to obtain ADC and T₂* values.

Upper row in Figure 2 shows sagittal views of the tendon images acquired with a clinical GRE sequence (TE/TR = 4/16.7 ms) at the five angular orientations. The arrows besides the specimen indicated the orientation of the fibers. Lower row in Figure 2 shows corresponding axial UTE-Cones-DWI images of the same tendon, where the imaging plane is perpendicular to the fiber orientation. High signal is observed in the Achilles tendon at different angles with the 3D UTE-Cones-DWI sequence, demonstrating the feasibility of diffusion weighted imaging of short T₂ tissues.

Figure 3 shows the results of the T₂* and ADC measured at the five angular orientations. T₂* increased more than 9 times when the fibers were oriented from 0° to 54° relative to the signal with fibers parallel to the B0 field. A strong but smaller magic angle effect was observed in the ADC, with an increase roughly half compared with T₂*.

The study showed that it was possible to obtain high quality diffusion weighted images of the Achilles tendon which has a short T₂ around 2ms when the fibers are parallel to B0. A marked magic angle effect was seen in ADC. Here, the signal equation for DWI based on the spin-echo EPI sequence was employed for data fitting. This simplification was not be very accurate due to T₁ recovery effects during the Cones data acquisition time. Multiple cones spokes were acquired with each STE preparation, with each spoke covering both the k-space center and periphery, resulting in a mixed contrast. More complicated model may be needed to compensate for the affects due to T₁ recovery and multi-spoke acquisition to improve the estimation of ADC in short T₂ tissues such as the Achilles tendon.