Skip to main content

Table 5 Performance of different feature selection algorithm achieved by F-test, MIC, RFE, Lasso, HAFS (\(\alpha =0.5\)) using Random Forest

From: Pulmonary lesion subtypes recognition of COVID-19 from radiomics data with three-dimensional texture characterization in computed tomography images

Method Label Precision (%) Recall (%) Accuracy (%) F-measure (%)
F-test 1 86.59 91.66 87.32 89.05
2 78.32 71.99 90.51 75.02
3 74.62 64.67 97.2 69.29
4 71.43 67.52 88.66 69.42
MIC 1 87.4 93.09 88.69 90.16
2 76.75 75.04 90.91 75.89
3 88.0 70.06 97.98 78.01
4 73.42 65.59 88.27 69.28
RFE 1 84.82 93.32 86.99 88.87
2 81.29 77.26 92.28 79.23
3 88.07 61.15 97.59 72.18
4 75.43 63.97 88.53 69.23
Lasso 1 87.69 93.44 89.05 90.47
2 77.84 73.85 91.0 75.79
3 80.45 68.15 97.52 73.79
4 74.69 67.69 88.85 71.02
HAFS 1 92.21 95.52 93.06 93.84
2 93.17 91.58 96.84 92.37
3 95.14 95.8 99.58 95.47
4 88.43 80.75 94.3 84.42
  1. Bold values indicate the maximum value of each type of lesion classification index