TABLE 1.
Characteristics of the 12 evaluated denoising pipelines
| Model number | Name | Trends | Band‐pass filter | RP | WM/CSF‐derived signals | ICA‐AROMA | GSR | T/P‐censoring | Total | Residual tDoF |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 12RP | 2/4 | 343/190 | 12 | — | — | — | — | 357/206 | 203/98 |
| 2 | 24RP | 2/4 | 343/190 | 24 | — | — | — | — | 369/218 | 191/86 |
| 3 | 24RP+8WM&CSF | 2/4 | 343/190 | 24 | 8 | — | — | — | 377/226 | 183/78 |
| 4 | 24RP+aCompCor | 2/4 | 343/190 | 24 | 10 | — | — | — | 379/228 | 181/76 |
| 5 | 24RP+aCompCor50% | 2/4 | 343/190 | 24 | 60.7 ± 4.0/35.3 ± 4.2 | — | — | — | 429.7 ± 4.0/253.3 ± 4.2 | 130.3 ± 4.0/50.7 ± 4.2 |
| 6 | ICA‐AROMA | 2/4 | 343/190 | — | 2 | 81.7 ± 20.7/29.8 ± 10.9 | — | — | 428.7 ± 20.7/225.8 ± 10.9 | 131.3 ± 20.7/78.2 ± 10.9 |
| GSR based | ||||||||||
| 7 | 24RP+8WM&CSF+4GSR | 2/4 | 343/190 | 24 | 8 | — | 4 | — | 381/230 | 179/74 |
| 8 | 24RP+aCompCor + 2GSR | 2/4 | 343/190 | 24 | 10 | — | 2 | — | 381/230 | 179/74 |
| 9 | 24RP+aCompCor50% + 2GSR | 2/4 | 343/190 | 24 | 61.9 ± 3.8/35.5 ± 4.1 | — | 2 | — | 432.9 ± 3.8/255.5 ± 4.1 | 127.1 ± 3.8/48.5 ± 4.1 |
| 10 | ICA‐AROMA + 2GSR | 2/4 | 343/190 | — | 2 | 81.7 ± 20.7/29.8 ± 10.9 | 2 | — | 430.7 ± 20.7/227.8 ± 10.9 | 129.3 ± 20.7/76.2 ± 10.9 |
| Censoring based | ||||||||||
| 11 | 24RP+8WM&CSF+4GSR+Tcens | 2/‐ | 343/‐ | 24/‐ | 8/‐ | — | 4/‐ | 36.3 ± 31.8/‐ | 417.3 ± 31.8/‐ | 142.7 ± 31.8/‐ |
| 12 | 24RP+8WM&CSF+4GSR+Pcens | 2/‐ | 343/‐ | 24/‐ | 8/‐ | — | 4/‐ | 140/‐ | 521/‐ | 39/‐ |
Note: For each pipeline the table shows the number of explanatory variables used by each set of confounders, along with the total number of variables (including the number of censored volumes) and the residual nominal tDoF. When more than one number is present in a cell, the first refers to the CF dataset, the second to the CNP dataset; when this distinction is absent, the number is common to both datasets. When the number of confounders is variable across the dataset, the mean and the standard deviation are reported.