Table 4 Prediction performance of radiomics features for RF, SVM, and LR models in the training, internal and validation Cohorts in the Prostate Cancer.

From: The value of MRI-based radiomics and clinicoradiological data for the detection of forkhead box protein A1 gene mutated prostate cancer

Models

Internal validation Cohort

External validation Cohort

AUC

Sens (%)

Spec (%)

Accu (%)

AUC

Sens (%)

Spec (%)

Accu (%)

RF

0.82 ± 0.05 (0.70–0.94)

85.32 ± 5.62 (0.63–0.96)

76.36 ± 5.01 (0.62–0.86)

79.01 ± 4.44 (0.67–0.87)

0.81 ± 0.05 (0.73–0.93)

63.30 ± 5.25 (0.45–0.78)

91.86 ± 2.31 (0.84–0.96)

83.38 ± 2.18 (0.76–0.90)

SVM

0.60 ± 0.06 (0.44–0.76)

43.11 ± 8.90 (0.26–0.66)

84.50 ± 5.54 (0.70–0.92)

72.21 ± 5.01 (0.60–0.82)

0.65 ± 0.07 (0.53–0.77)

48.62 ± 7.47 (0.32–0.66)

89.53 ± 3.01 (0.81–0.95)

77.38 ± 2.96 (0.70–0.85)

LR

0.74 ± 0.06 (0.59–0.89)

66.97 ± 7.31 (0.43–0.84)

78.29 ± 5.11 (0.64–0.88)

74.93 ± 4.07 (0.63–0.84)

0.71 ± 0.06 (0.59–0.83)

59.63 ± 6.31 (0.42–0.76)

88.37 ± 2.72 (0.80–0.94)

79.84 ± 2.83 (0.73–0.87)

  1. Senc sensitivity, Spec specificity, Accu accuracy, RF random forest, SVM support vector machine, LR logistic regression, AUC area under the ROC curve, CI confidence interval.
Back to article page