Table 3.
Sequence | Interpreter | Sensitivity | 95% CI | Specificity | 95% CI | Accuracy | 95% CI | AUC | 95% CI | p-Value for AUC (Versus CNN) |
---|---|---|---|---|---|---|---|---|---|---|
T2WI | CNN | 0.77 | 0.68–0.84 | 0.85 | 0.76–0.91 | 0.81 | 0.72–0.87 | 0.83 | 0.74–0.91 | - |
Reader 1 | 0.63 | 0.54–0.68 | 0.90 | 0.82–0.96 | 0.77 | 0.69–0.82 | 0.89 | 0.82–0.95 | 0.127 | |
Reader 2 | 0.85 | 0.77–0.91 | 0.87 | 0.79- 0.92 | 0.86 | 0.78–0.92 | 0.91 | 0.85–0.97 | 0.048 * | |
Reader 3 | 0.75 | 0.66–0.82 | 0.85 | 0.76–0.91 | 0.80 | 0.71–0.86 | 0.88 | 0.81–0.94 | 0.305 | |
DWI | CNN | 0.85 | 0.77–0.91 | 0.85 | 0.77–0.90 | 0.85 | 0.77–0.91 | 0.88 | 0.81–0.95 | - |
Reader 1 | 0.71 | 0.61–0.79 | 0.77 | 0.68–0.84 | 0.74 | 0.65–0.82 | 0.81 | 0.72–0.89 | 0.151 | |
Reader 2 | 0.65 | 0.55–0.73 | 0.71 | 0.62–0.79 | 0.68 | 0.58–0.76 | 0.74 | 0.64–0.84 | 0.004 * | |
Reader 3 | 0.79 | 0.70–0.87 | 0.75 | 0.66–0.82 | 0.77 | 0.68–0.84 | 0.82 | 0.73–0.90 | 0.135 | |
ADC map | CNN | 0.85 | 0.76–0.92 | 0.77 | 0.69–0.83 | 0.81 | 0.72–0.87 | 0.89 | 0.83–0.96 | - |
Reader 1 | 0.81 | 0.72–0.88 | 0.75 | 0.66–0.82 | 0.78 | 0.69–0.85 | 0.84 | 0.77–0.92 | 0.263 | |
Reader 2 | 0.92 | 0.83–0.97 | 0.36 | 0.29–0.41 | 0.63 | 0.55–0.68 | 0.79 | 0.70–0.88 | 0.023 * | |
Reader 3 | 0.77 | 0.68–0.84 | 0.79 | 0.70–0.86 | 0.78 | 0.69–0.85 | 0.85 | 0.78–0.93 | 0.356 | |
CE-T1WI | CNN | 0.81 | 0.73–0.86 | 0.92 | 0.85–0.97 | 0.87 | 0.79–0.92 | 0.86 | 0.78–0.94 | - |
Reader 1 | 0.73 | 0.65–0.78 | 0.92 | 0.85–0.97 | 0.83 | 0.75- 0.88 | 0.87 | 0.79–0.94 | 0.903 | |
Reader 2 | 0.75 | 0.66–0.82 | 0.83 | 0.74–0.89 | 0.79 | 0.70–0.86 | 0.85 | 0.77–0.92 | 0.730 | |
Reader 3 | 0.75 | 0.65–0.83 | 0.71 | 0.62–0.79 | 0.73 | 0.64–0.81 | 0.82 | 0.73–0.90 | 0.416 |
ADC: apparent diffusion coefficient; AUC: area under the receiver operating characteristic curve; CE-T1WI: contrast-enhanced fat-saturated T1-weighted imaging; CI: confidence interval; CNN: convolutional neural network; DWI: diffusion-weighted imaging; T2WI: T2-weighted imaging. * p < 0.05.