Final Gleason Score Prediction Using Discriminant Analysis and Support Vector Machine Based on Preoperative Multiparametric MR Imaging of Prostate Cancer at 3T
Table 3
The averages and confidence intervals of accuracy, sensitivity, and specificity values for the combinations of two kinds of classifiers and two kinds of PCA averaged after thirty iterations.
Accuracy [CI] (%)
Sensitivity [CI] (%)
Specificity [CI] (%)
All parameters
SVM
Kernel PCA
76.83 75.77–77.90
*
83.38 81.77–85.00
*
68.36 65.88–70.84
*
Standard PCA
52.03 50.52–53.53
58.16 55.94–60.38
47.87 43.93–51.8
LDA
Kernel-PCA
76.56 75.42–77.70
*
81.59 79.87–83.31
*
70.09 67.66–72.52
Standard PCA
60.03 58.88–61.17
55.72 53.98–57.45
67.55 64.19–70.91
DRE eliminated
SVM
Kernel PCA
76.17 75.12–77.21
*
84.59 83.22–85.96
*
66.07 63.67–68.46
*
Standard PCA
51.94 50.17–53.72
58.24 55.92–60.55
46.11 41.70–50.53
LDA
Kernel PCA
75.36 74.01–76.71
*
82.59 80.84–84.34
*
66.27 63.75–68.79
Standard PCA
60.11 59.09–61.13
55.28 53.61–56.94
67.14 63.83–70.46
Age eliminated
SVM
Kernel PCA
75.97 75.07–76.88
*
85.12 83.73–86.51
*
64.15 61.57–66.72
*
Standard PCA
56.03 53.97–58.09
67.85 65.64–70.07
40.55 35.62–45.49
LDA
Kernel PCA
76.33 75.10–77.56
*
83.95 82.10–85.80
*
66.48 63.27–69.69
Standard PCA
57.50 56.11–58.89
51.81 49.34–54.28
65.01 62.26–67.77
Index Lesion Size Eliminated
SVM
Kernel PCA
75.31 74.19–76.42
*
87.88 86.31–89.45
*
56.83 53.94–59.72
*
Standard PCA
53.64 52.52–54.76
64.37 62.52–66.22
39.90 36.66–43.15
LDA
Kernel PCA
76.75 75.64–77.86
*
86.51 84.94–88.08
*
64.00 61.22–66.77
*
Standard PCA
59.69 58.36–61.03
51.20 48.89–53.50
72.71 69.60–75.81
Statistically significantly different ( < 0.05/24).