Quantitative MRI mOOCT2 Mapping · Exercise 4
How much SNR is left at each echo?
- a. The signal curves in Figure 3.4 were simulated for different SNR values, however these are relative to the maximum signal (i.e. TE=0). Using Binder or MATLAB and the notebook that was used to generate that figure, estimate the SNR of images acquired TE = 12.8, 128, 256, and 384 ms for a maximum signal SNR of 100.
- b. Repeat the same calculations for a T2* of 30 ms.
- c. Comment on these results and on how they would impact your imaging protocol planning. What’s the maximum TE you think is useful for each case?
a. SNR at four echo times, T2 = 109 ms
| T2 (ms) | |
| SNR at TE = 0 | |
| TEs (ms) | |
| Noise trials |
SNR ≈ 89, 31, 9.6, 3.0. The noise does not change with TE; the signal does: SNR(TE) = SNR0 e−TE/T2.
b. The same for T2* = 30 ms
| T2* (ms) | |
| SNR at TE = 0 | |
| TEs (ms) | |
| Noise trials |
SNR ≈ 65, 1.4, 0.02, 0.0003. Only the first echo is usable: at 128 ms the signal is barely above the noise.
c. The longest useful TE
c. How would these results impact your protocol planning? What’s the maximum TE you think is useful for each case?
| T2 / T2* (ms) | / |
| SNR at TE = 0 | |
| Lowest useful SNR |
- Later echoes add time (and SAR in a spin-echo train), not information.
- On magnitude images they sit on a noise floor, which a fit reads as slow decay: T2 is overestimated.
- 4× the averages adds only T2 ln 2: 76 and 21 ms.
Exercise 4: answers
| TE (ms) | 12.8 | 128 | 256 | 384 |
|---|---|---|---|---|
| a · T2 = 109 ms | 89 | 31 | 9.6 | 3.0 |
| b · T2* = 30 ms | 65 | 1.4 | 0.02 | 0.0003 |
c · With SNR ≥ 5 as the limit, echoes are useful up to TE ≈ 330 ms (≈ 3 T2) for T2 = 109 ms, and ≈ 90 ms for T2* = 30 ms.
Questions: github.com/qMRLab/mooc/discussions