Table 2.
FODF prediction assessment
| Model | Shell configuration | Tissue proportion prediction | Scan-rescan consistency |
|---|---|---|---|
|
| |||
| Single model | 1K | 8.45E-04 | 0.862 |
| 2K | 7.92E–04 | 0.865 | |
| 3K | 8.63E–04 | 0.857 | |
| 1K, 2K | 7.32E–04 | 0.856 | |
| 2K, 3K | 7.49E–04 | 0.86 | |
| 1K, 3K | 8.02E–04 | 0.862 | |
| 1K, 2K, 3K | 6.38E–04 | 0.865 | |
|
| |||
| DH w. SC | 1K | 7.27E–04 | 0.855 |
| 2K | 7.12E–04 | 0.86 | |
| 3K | 7.35E–04 | 0.858 | |
| 1K, 2K | 7.01E–04 | 0.858 | |
| 2K, 3K | 6.79E–04 | 0.86 | |
| 1K, 3K | 6.82E–04 | 0.864 | |
| 1K, 2K, 3K | 5.92E–04 | 0.861 | |
|
| |||
| Silver standard: MSMT-CSD [15] | 0.856 | ||
Reconstruction results from msmt-CSD are applied as silver standard in the evaluation. Wilcoxon signed-rank test is applied as a statistical assessment for scan-rescan consistency evaluation. It has a significant difference (p < 0.001) compared with WM fODF. The MSE is reported for evaluation of VF predictions. The ACC between scan/rescan DW-MRI over WM regions is reported.