Table 1.
Noise reduction performance, indicated by the observed changes when using the proposed method compared to (otherwise identical) regression analysis without the noise-estimate regressor.
| Volunteer | t-score change [%] | amplitude change [%] | residue variance change [%] | number of active voxels change [%] | number of active voxels change | number of voxels in RAct | number of voxels in RRef |
|---|---|---|---|---|---|---|---|
| 1 | 14.1 (0.3) | −0.06 (0.00) | −12.1 (0.2) | 16.7 | 304 | 1819 | 440 |
| 2 | 9.4 (0.3) | −0.03 (0.00) | −7.3 (0.2) | 10.0 | 201 | 2005 | 960 |
| 3 | 12.3 (0.4) | −0.13 (0.01) | −10.2 (0.3) | 57.7 | 390 | 676 | 963 |
| 4 | 14.7 (0.5) | −0.03 (0.00) | −11.2 (0.3) | 29.5 | 388 | 1314 | 2008 |
| 5 | 6.0 (0.2) | −0.33 (0.01) | −4.8 (0.1) | 7.2 | 195 | 2726 | 103 |
| Mean | 11.3 (1.6) | −0.11 (0.1) | −9.1 (1.4) | 24.2 (9.2) | 296 (43) | 1708 (343) | 895 (323) |
| Mean (threshold=0.4) | 11.0 (1.7) | −0.12 (0.1) | −9.1 (1.5) | 24.3 (9.0) | 301 (39) | 1710 (343) | 1987 (581) |
| Mean (threshold=0.6) | 12.2 (1.2) | −0.13 (0.1) | −9.3 (1.1) | 24.7 (8.9) | 309 (39) | 1706 (344) | 327 (154) |
| Mean (30 s ref. data) | 11.0 (2.5) | −0.09 (0.0) | −9.1 (2.0) | 22.3 (5.3) | 322 (49) | 1708 (343) | 1135 (384) |
| Mean (270 s ref. data) | 10.9 (1.5) | −0.07 (0.0) | −8.8 (1.5) | 23.4 (5.8) | 337 (56) | 1710 (342) | 1632 (666) |
Values were obtained using a voxel selection threshold for the reference region of 0.5 and 120 s worth of reference data, unless otherwise noted (lower four rows). Values between parentheses are standard errors.