Table 2.
Summary of results for the different fitting methods, with and without an external proton-density-weighted fat fraction (FF) constraint.
| Method | FF Constraint | # echoes | wT2 (ms) | FF (%) |
Global FF error (p.p.) |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Volunteers | Patients | Vol. | Pat. | |||||||||
| Average | Global SD | Pooled SD | Intrasubject SD | Average | Global SD | Pooled SD | ||||||
| EPG | No | 3 | 28.6 | 1.3 | 8.1 | 1.0 | 31.4 | 6.2 | 12 | 17.3 | 22.5 | 6.4 |
| 5 | 34.2 | 1.8 | 5.2 | 1.1 | 37.7 | 7.7 | 8.8 | 2.9 | 8.1 | −8.0 | ||
| 8 | 33.8 | 1.8 | 4.4 | 1.0 | 36.1 | 4.9 | 7.6 | 3.2 | 9.8 | −7.0 | ||
| 17 | 33.0 | 1.3 | 3.3 | 0.9 | 32.7 | 2.5 | 4.1 | 4.6 | 12.8 | −4.7 | ||
| Yes | 3 | 34.5 | 1.5 | 5.9 | 1.5 | 34.5 | 3.6 | 8.9 | 6.4* | 17.8* | ||
| 5 | 32.3 | 1.3 | 4.9 | 1.1 | 32.2 | 3.2 | 6.4 | |||||
| 8 | 31.9 | 1.4 | 4.9 | 1.3 | 32.4 | 2.6 | 5.9 | |||||
| 17 | 32.1 | 1.6 | 5.1 | 1.4 | 33.1 | 2.5 | 6.2 | |||||
| Double exponential | No | 5 | 25.3 | 2.9 | 8.3 | 3.1 | 24.7 | 2.5 | 10.1 | 17.9 | 23.9 | 8.6 |
| 8 | 29.4 | 1.7 | 5.8 | 1.8 | 28.4 | 3.1 | 6.7 | 15.1 | 21.9 | 6.2 | ||
| 17 | 33.2 | 1.6 | 4.6 | 1.6 | 31.7 | 3.7 | 9.5 | 11.9 | 19.0 | 3.2 | ||
| Yes | 5 | 36.1 | 2.1 | 6.9 | 2.3 | 35.4 | 3.6 | 8.3 | 6.4* | 17.8* | ||
| 8 | 38.6 | 2.5 | 7.7 | 2.7 | 38.4 | 3.7 | 9.3 | |||||
| 17 | 41.3 | 3.5 | 10.4 | 3.5 | 41.2 | 6.2 | 14 | |||||
The pooled standard deviation (SD) is associated with image noise, whereas the intrasubject SD is associated with homogeneity in different anatomical regions. The FF error with respect to the multi-echo gradient-echo acquisition (MEGE) is given in percentage points (p.p.). Values marked with an asterisk (*) are not fitted by the algorithm but are derived from the MEGE acquisition.