Synopsis

MRI of tissues with short T2* can be realized with ultra-short echo time (UTE) sequences. The effect of lowering TE on relative SNR is calculated and evaluated. The limiting factor for dead time and, consequently, low TE is determined for hardware components of clinical MR system and compared to benchmark results.

Introduction

Materials and Methods

Measurements were performed on a 3T clinical MR system (PRISMA, Siemens Healthcare, Erlangen, Germany) using a mummified ancient hand sample (Figure 1) with a T2* between 100 and 600 µs³. It was measured with a 3D UTE sequence with following acquisition parameters: TE=70µs, TR=3ms, FA=16°, 80000 evenly spaced radial projections with 128 points per spoke. The measurement hardware consisted of a custom-built quadrature TxRx birdcage coil. The coil components were characterized on the workbench in terms of transient behavior (Figure 2). In particular, coil ring down and TxRx switching time were measured. To compare the measured properties to actual performance, the measurement setup was transferred to the MR system (Figure 3). An optical signal was used to trigger the oscilloscope at the start of the RF excitation pulse, which was measured with a sniffer coil. To characterize ON/OFF switching time and transient behavior of the TxRx switch, DC bias was also monitored to determine hardware dead time between RF pulse and signal acquisition. Minimum available TE can be obtained from measuring duration between the center of RF pulse and begin of acquisition. To determine effects of relative SNR of short T2* tissue for different TEs, the relative SNR values were estimated with a simplified SNR equation⁴

$$SNR\approx \gamma B_0 \cdot \frac{B_1}{I}\cdot\frac{{ }V_{0} \cdot M_{T}}{\sqrt{4k_{B}T(R_{coil}+R_{sample})\Delta f}}\cdot\sqrt{t_{0}/t_{R}}\cdot\sqrt{t_{scan}}.$$

The essential parameters are coil sensitivity $$$\frac{B_1}{I}$$$, induced losses by coil resistance $$$R_{coil}$$$ and especially transverse magnetization $$$M_{T}$$$ , where signal decay due to T2* is considered during RF excitation⁵. Due to inherently low conductivity of mummified tissue, the sample losses $$$R_{sample}$$$ can be neglected.

Results and Discussion

The acquired MR image of the mummy hand (Figure 1) allows for discrimination of tissue such as cortical bones (T2*=150µs), tendons (T2*=220µs) and skin with embalming remnants (T2*=300-370µs). These calculated relative SNR values are plotted in Figure 4 for three different TEs (35, 70 and 150µs) and compared to measured mummified tissue at TE=70µs. The overall trend of the curve fits to SNR values in the regions of interest of selected tissue components (Figure 4). The slightly lower SNR might be attributed to neglected losses in either $$$R_{coil}$$$ or in other hardware components. Overall, the plot shows higher SNR gain for T2* values close to selected TE and remains on a slightly higher SNR value for increasing T2*. Total dead time for switching and coil ring down was within microsecond range (2.5 -3µs) which is far below the applied TE values. However, applying the same setup at the actual MR system (Figure 3a) shows an additional DC bias delay that prolongs the ON state of the switch to 20 µs (Figure 3b). To compare workbench and MR system results, the output voltages of workbench experiments need to be scaled up to the MR system’s level which requires a higher excitation voltage of $$$V_0=15V$$$ and results in a coil ring down time of⁶

$$t_{rd} = \frac{2\cdot Q}{\omega_0}\cdot \ln (\frac{V_0}{V_{noise}}).$$

Here a value of $$$t_{rd}=3.8µs$$$ was determined. This increases dead time of the workbench measurement to around 6µs, which is much lower than the 20µs delay of the actual system. Even lowering TE and duration of RF pulse did not affect this delay, which seems to be an intrinsic hardware or software delay.

Conclusion

Acknowledgements

References