Introduction

Direct imaging of the skull requires application of ultra-short echo-time (UTE) imaging sequences¹ because compact bone has very short T₂ relaxation times, so short that no signal remains when using most conventional MRI techniques. Although imaging of the skull is conveniently performed with computed tomography; MRI could prove beneficial as an alternative imaging technique in cases such as: (i) minimizing radiation exposure as, e.g., in pediatric imaging, (ii) PET-MRI attenuation correction², or (iii) quantifying relaxation or other parameters within the skull. This work investigates the optimal flip angle for separating the skull from surrounding tissues when using UTE imaging and additionally investigates if the T₂^* values of the skull depend on the flip angle.

Methods

A radial 3D UTE imaging sequence was used with non-selective hard-pulse excitation and center-out readouts using a spikey ball distribution in k-space¹. Four near in-phase echoes were acquired using a multi-echo readout with echo times: 0.17ms, 2.57ms, 4.97ms, 7.37ms. Other parameters were: 128x128x98 acquisition matrix, (200x200x154)mm³ field of view, 1260 Hz/Pixel bandwidth, 10ms TR, 42638 readouts per k-space and fat saturation. This 8 minute long scan was repeated 7 times using flip angles of 5°,10°,15°,20°,30°,40° and 49°. Larger flip angles could not be applied due to limits of the allowed specific absorption rate. Data of 3 adult volunteers with no known pathologies were acquired on a 3T MRI system (PRISMA, Siemens Healthineers) using a vendor supplied 20 channel head coil. Image reconstruction and analysis was performed offline using MATLAB. For image analysis absolute difference images between the magnitude images of the first and second echoes were calculated to separate the fast decaying skull signal from the slower decaying surrounding tissues. A voxel-wise squared mono-exponential fit was performed over all acquired echo times to create T₂^* maps. Regions-of-interests were placed manually in the center of the skull as well as in homogeneous white matter brain regions across several slices.

Results

Figure 1 shows imaging results for a single sagittal slice of the head. The magnitude images of the first two echoes reveal the expected change of signal intensity and tissue contrast with increasing flip angle. The contrast in the absolute difference images between the skull and surrounding tissues is weak for small flip angles, but increases with flip angles in such a way that predominantly compact bone structures with short T₂^* times are visible in the difference images. These observations are also supported by the region-of-interest analysis in Figure 2, which shows the highest increases of the signal ratio between skull and brain for the difference images calculated, ranging from 225% to over 500%. Considering T₂^* as a function of the flip angle (see Figures 1,3,4), a dependency on the flip angle becomes apparent. The decrease of T₂^* with flip angle does not appear to be specific to any particular layer of the skull as shown in Figure 4. The largest changes occurred in the flip angle range between 5° and 20°. For flip angles larger than 20° only smaller changes of T₂^* were observed.

Discussion

Varying the flip angle resulted in distinct image contrast changes between the skull and brain tissue as well as changes of the T₂^* values of the skull. The contrast changes, seen specifically on the difference images, are induced by T₁-based saturation effects of tissues with long T₁ times, such as brain tissue, with higher flip angles. Tissues with very short T₁ times, on the other hand, reach their optimal equilibrium magnetization at higher flip angles. Although we were not able to find exact T₁ values for the skull in the literature, results from investigations on the lower leg indicate T₁ times for compact cortical bone as short as 80ms³ to 223ms⁴. The observed decrease of T₂^* with increasing flip angle can likewise be explained by T₁-based saturation effects. If a voxel contains more than one tissue type with different T₁ and T₂^* times, changing the flip angle will change the relative weighting of the signal of these tissue types contributing to the overall voxel signal, thus modifying the measured sum signal as well as the apparent T₂^* time. As short repetition times and high flip angles over saturate components with long T₁ times their contribution on the measured apparent T₂^* is consequently reduced.

Conclusion

With the shown strong dependency of both difference image contrast and T₂^* on the flip angle acquisition parameters for skull imaging should be chosen carefully. For optimal difference images and T₂^* maps our results suggest the usage of high flip angles, in the present study of 30° and above for a TR of 10ms.

Acknowledgements

No acknowledgement found.