Table 4.
Summary of the average balanced accuracies and the corresponding 95% CI (DATASET2) obtained using all 5 test folds of the cross-validation of the 5 machine learning models (neural network, random forest, support vector machine, logistic regression, naïve Bayes) as a function of the normalization method. For both intensity discretization methods (FBN and FBS), 32 bins were used. For model 4, numbers of robust features as defined using DATASET1 are written in square brackets. BAC balanced accuracy, ROC-AUC area under the receiver operating characteristic curve.
| T1w-gd | T2w-flair | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | Model 1 | Model 2 | Model 3 | Model 4 | ||||||||||
| BAC | ROC-AUC | BAC | ROC-AUC | BAC | ROC-AUC | BAC | ROC-AUC | BAC | ROC-AUC | BAC | ROC-AUC | BAC | ROC-AUC | BAC | ROC-AUC | ||
| FBN | No normalization | 0.67 (0.61–0.73) | 0.74 (0.68–0.80) | 0.80 (0.76–0.83) | 0.86 (0.82–0.89) | 0.76 (0.71–0.81) | 0.83 (0.77–0.88) | 0.73 (0.70–0.77) [23] | 0.83 (0.80–0.86) [23] | 0.62 (0.59–0.64) | 0.64 (0.60–0.68) | 0.65 (0.62–0.68) | 0.70 (0.67–0.73) | 0.63 (0.60–0.65) | 0.70 (0.66–0.74) | 0.60 (0.57–0.63) [23] | 0.66 (0.63–0.70) [23] |
| Nyul | 0.82 (0.79–0.84) | 0.90 (0.87–0.92) | 0.76 (0.72–0.79) | 0.83 (0.80–0.86) | 0.81 (0.77–0.84) | 0.88 (0.86–0.91) | 0.81 (0.78–0.84) [43] | 0.89 (0.86–0.92) [43] | 0.56 (0.52–0.59) | 0.61 (0.58–0.65) | 0.67 (0.64–0.69) | 0.72 (0.70–0.74) | 0.66 (0.64–0.69) | 0.71 (0.69–0.74) | 0.62 (0.59–0.66) [27] | 0.67 (0.63–0.70) [27] | |
| WhiteStripe | 0.79 (0.77–0.82) | 0.88 (0.86–0.90) | 0.80 (0.76–0.83) | 0.86 (0.83–0.90) | 0.80 (0.77–0.84) | 0.89 (0.86–0.92) | 0.79 (0.76–0.83) [28] | 0.89 (0.87–0.91) [28] | 0.57 (0.54–0.60) | 0.63 (0.60–0.67) | 0.65 (0.62–0.68) | 0.70 (0.67–0.73) | 0.65 (0.62–0.67) | 0.70 (0.67–0.73) | 0.62 (0.59–0.65) [24] | 0.67 (0.64–0.71) [24] | |
| Z-Score | 0.82 (0.80–0.85) | 0.91 (0.89–0.93) | 0.80 (0.76–0.83) | 0.86 (0.83–0.90) | 0.82 (0.80–0.86) | 0.90 (0.88–0.93) | 0.81 (0.78–0.84) [32] | 0.91 (0.89–0.94) [32] | 0.60 (0.57–0.63) | 0.65 (0.62–0.69) | 0.65 (0.62–0.68) | 0.70 (0.66–0.73) | 0.67 (0.64–0.70) | 0.72 (0.69–0.75) | 0.63 (0.60–0.66) [24] | 0.68 (0.65–0.72) [24] | |
| FBS | No normalization | 0.67 (0.61–0.73) | 0.74 (0.68–0.80) | 0.68 (0.62–0.72) | 0.75 (0.70–0.79) | 0.69 (0.63–0.74) | 0.75 (0.68–0.81) | 0.58 (0.54–0.61) [9] | 0.64 (0.59–0.69) [23] | 0.62 (0.59–0.64) | 0.64 (0.60–0.68) | 0.60 (0.58–0.63) | 0.64 (0.61–0.68) | 0.59 (0.56–0.62) | 0.64 (0.60–0.67) | 0.56 (0.54–0.59) [7] | 0.61 (0.58–0.65) [7] |
| Nyul | 0.82 (0.79–0.84) | 0.90 (0.87–0.92) | 0.76 (0.74–0.79) | 0.83 (0.80–0.86) | 0.81 (0.78–0.84) | 0.88 (0.85–0.91) | 0.82 (0.79–0.85) [40] | 0.89 (0.86–0.91) [43] | 0.56 (0.52–0.59) | 0.61 (0.58–0.65) | 0.64 (0.60–0.68) | 0.71 (0.67–0.75) | 0.62 (0.59–0.66) | 0.70 (0.66–0.73) | 0.59 (0.55–0.62) [50] | 0.64 (0.61–0.68) [50] | |
| WhiteStripe | 0.79 (0.77–0.82) | 0.88 (0.86–0.90) | 0.76 (0.72–0.79) | 0.84 (0.81–0.87) | 0.79 (0.76–0.82) | 0.87 (0.84–0.90) | 0.79 (0.76–0.82) [20] | 0.88 (0.86–0.90) [28] | 0.57 (0.54–0.60) | 0.63 (0.60–0.67) | 0.63 (0.60–0.66) | 0.69 (0.67–0.73) | 0.61 (0.58–0.64) | 0.69 (0.65–0.72) | 0.61 (0.58–0.64) [36] | 0.68 (0.65–0.71) [36] | |
| Z-Score | 0.82 (0.80–0.85) | 0.91 (0.89–0.93) | 0.78 (0.75–0.82) | 0.86 (0.83–0.89) | 0.80 (0.77–0.83) | 0.90 (0.87–0.93) | 0.83 (0.80–0.85) [45] | 0.91 (0.88–0.93) [32] | 0.60 (0.57–0.63) | 0.65 (0.62–0.69) | 0.64 (0.61–0.67) | 0.70 (0.67–0.73) | 0.64 (0.60–0.67) | 0.71 (0.68–0.74) | 0.61 (0.58–0.63) [36] | 0.66 (0.62–0.69) [36] | |