Table 3.
Group 1a (N = 14) | DL estimation | Reference | P value | |
---|---|---|---|---|
T2 (ms) | Non-Tumor region | 111.9 ± 14.2 | 128.1 ± 23.9 | 0.045* |
Tumor region | 83.7 ± 16.5 | 82.1 ± 13.5 | 0.788 | |
P value | <0.001*** | <0.001*** | ||
Group 1b (N = 24) | ||||
T2 (ms) | Non-Tumor region | 103.8 ± 16.5 | ||
Tumor region | 78.4 ± 12.6 | |||
P value | <0.001*** | |||
Group 1 all (N = 38) | ||||
T2 (ms) | Non-Tumor region | 106.8 ± 16.3 | ||
Tumor region | 80.4 ± 14.4 | |||
P value | <0.001*** |
DL, deep learning.
Two series of paired t-tests were included in the analysis: one is between deep learning estimation and ground truth (only on Group 1a), and the other is between tumor and non-tumor regions with the significance level of *P < 0.05; ***P < 0.001.