Skip to main content
. Author manuscript; available in PMC: 2022 Oct 15.
Published in final edited form as: Neuroimage. 2022 Jul 13;260:119474. doi: 10.1016/j.neuroimage.2022.119474

Table 4.

Skull-stripping accuracy across datasets, as measured by the mean Dice overlap (± SD) between computed and ground-truth binary brain masks. p-values comparing baseline with SythStrip results are presented below each score. Across each dataset, SynthStrip significantly outperforms most baselines except those with p-values in bold, for which p > 0.05.

Dice (%)
SynthStrip ROBEX BET 3DSS BEaST FSW DMBE
IXI T1w 97.0 ± 0.5 96.2 ± 0.8 96.1 ± 3.1 95.4 ± 1.4 93.4 ± 0.8 92.6 ± 4.4 93.7 ± 3.0
4.5 × 10−8 5.1 × 10−2 1.3 × 10−9 1.1 × 10−32 5.8 × 10−9 8.0 × 10−10
FSM T1w 97.8 ± 0.3 95.9 ± 0.7 65.8 ± 11.2 92.0 ± 3.0 92.1 ± 0.7 93.8 ± 3.7 89.5 ± 3.3
3.3 × 10−18 1.7 × 10−19 1.4 × 10−13 1.9 × 10−35 2.2 × 10−7 5.6 × 10−18
ASL T1w 97.3 ± 0.5 96.8 ± 1.1 94.9 ± 1.1 90.8 ± 3.5 89.0 ± 5.5 95.5 ± 0.9 92.7 ± 1.7
3.0 × 10−3 5.7 × 10−17 1.5 × 10−15 7.7 × 10−13 1.1 × 10−11 1.1 × 10−19
QIN T1w 96.3 ± 0.8 92.8 ± 4.6 93.8 ± 2.7 92.4 ± 3.1 85.2 ± 17.1 79.9 ± 13.3 89.0 ± 7.3
3.7 × 10−7 5.2 × 10−11 2.3 × 10−15 1.5 × 10−5 5.2 × 10−12 7.4 × 10−10
IXI T2w 96.4 ± 0.7 91.3 ± 2.7 91.0 ± 6.2 94.9 ± 1.7 63.0 ± 12.4 - 6.7 ± 8.0
2.9 × 10−19 2.0 × 10−7 1.4 × 10−9 7.8 × 10−24 - 1.1 × 10−52
FSM T2w 97.7 ± 0.3 93.2 ± 1.7 95.7 ± 1.6 94.9 ± 2.0 69.8 ± 15.0 - 3.4 ± 7.2
4.9 × 10−18 9.8 × 10−9 4.1 × 10−10 7.3 × 10−13 - 1.1 × 10−40
QIN T2w 95.2 ± 1.1 87.3 ± 5.3 89.6 ± 4.0 71.4 ± 21.2 57.7 ± 19.1 - 61.5 ± 16.5
1.1 × 10−10 1.7 × 10−9 6.3 × 10−7 9.1 × 10−14 - 2.9 × 10−15
QIN FLAIR 96.4 ± 0.5 93.7 ± 1.3 95.9 ± 0.9 93.8 ± 1.1 90.3 ± 4.7 83.4 ± 6.2 87.9 ± 3.0
5.7 × 10−7 1.3 × 10−2 4.2 × 10−9 8.7 × 10−5 2.6 × 10−7 1.5 × 10−9
IXI PDw 96.4 ± 1.0 94.6 ± 1.3 95.5 ± 1.4 95.1 ± 1.6 78.3 ± 15.8 81.3 ± 10.9 90.0 ± 3.8
5.3 × 10−17 1.9 × 10−7 1.4 × 10−8 1.8 × 10−10 2.5 × 10−12 3.8 × 10−16
FSM PDw 97.2 ± 0.5 95.8 ± 1.0 95.7 ± 6.5 95.5 ± 1.4 88.8 ± 8.4 67.4 ± 9.8 79.1 ± 4.4
1.1 × 10−6 2.3 × 10−1 4.7 × 10−8 5.1 × 10−6 4.9 × 10−17 3.4 × 10−21
IXI MRA 97.7 ± 0.5 82.0 ± 4.3 62.3 ± 14.9 95.1 ± 1.0 91.5 ± 5.3 - 18.1 ± 19.5
4.5 × 10−30 1.3 × 10−21 3.0 × 10−24 6.7 × 10−10 - 4.7 × 10−32
FSM qT1 97.7 ± 0.3 63.1 ± 19.6 47.2 ± 13.5 44.6 ± 3.8 36.3 ± 15.6 49.5 ± 10.7 3.1 ± 7.1
1.1 × 10−10 8.8 × 10−20 4.9 × 10−36 9.4 × 10−20 1.7 × 10−21 1.6 × 10−36
ASL EPI 95.2 ± 1.4 88.3 ± 4.6 94.2 ± 1.8 95.2 ± 1.4 67.4 ± 19.6 69.4 ± 11.9 92.8 ± 1.5
3.6 × 10−11 5.2 × 10−4 5.8 × 10−1 2.0 × 10−11 8.9 × 10−17 2.1 × 10−15
Infant T1w 96.1 ± 1.4 87.4 ± 15.4 66.8 ± 20.3 71.6 ± 17.7 63.2 ± 38.4 61.3 ± 31.3 84.4 ± 7.3
3.5 × 10−2 3.0 × 10−5 4.4 × 10−5 3.9 × 10−3 7.9 × 10−4 4.9 × 10−6
IXI DWI 95.5 ± 1.3 85.3 ± 5.7 93.4 ± 2.1 90.4 ± 2.3 79.5 ± 11.4 75.1 ± 8.8 80.5 ± 3.1
2.4 × 10−10 1.2 × 10−11 1.6 × 10−17 6.8 × 10−9 1.5 × 10−13 1.4 × 10−27
CIM PET 95.4 ± 1.1 89.5 ± 4.8 78.3 ± 8.3 93.7 ± 6.1 4.6 ± 8.8 51.6 ± 10.1 42.3 ± 14.3
1.8 × 10−5 2.9 × 10−8 2.1 × 10−1 1.4 × 10−18 4.1 × 10−13 1.6 × 10−12
CIM CT 94.3 ± 0.9 73.7 ± 2.0 45.4 ± 4.0 60.4 ± 2.9 2.3 ± 3.3 48.8 ± 5.1 58.2 ± 6.5
3.8 × 10−23 4.8 × 10−22 4.0 × 10−22 3.5 × 10−26 1.4 × 10−16 2.1 × 10−15