Table 2 Details of scout models.
Model | Base image | Feature selection | Average CV AUC | Feature number | Features | |||
|---|---|---|---|---|---|---|---|---|
Classic | Original | ANOVA LR | 0.623 | 6 | glcm_Contrast, glcm_DifferenceVariance, ngtdm_Contrast, shape_MajorAxisLength, shape_MinorAxisLength, shape_SurfaceVolumeRatio | |||
LoG (σ = 1.5) | KW LR | 0.647 | 3 | firstorder_10Percentile, firstorder_Mean, firstorder_Variance | ||||
LoG (σ = 3) | ANOVA LR | 0.655 | 1 | glcm_DifferenceEntropy | ||||
LoG (σ = 5) | Relief LR | 0.601 | 14 | glrlm_LongRunHighGrayLevelEmphasis, firstorder_RootMeanSquared, firstorder_90Percentile, glszm_SmallAreaLowGrayLevelEmphasis, glcm_SumEntropy, firstorder_Maximum, glcm_ClusterShade, gldm_LargeDependenceHighGrayLevelEmphasis, glszm_LargeAreaLowGrayLevelEmphasis, glszm_LargeAreaEmphasis, glszm_HighGrayLevelZoneEmphasis, firstorder_InterquartileRange, glrlm_GrayLevelNonUniformityNormalized, glrlm_LongRunEmphasis | ||||
Delta | Original | ANOVA LR | 0.651 | 6 | glcm_Contrast, glcm_DifferenceVariance, glszm_LowGrayLevelZoneEmphasis, ngtdm_Coarseness, ngtdm_Contrast, shape_SurfaceVolumeRatio | |||
LoG (σ = 1.5) | ANOVA SVM | 0.677 | 4 | firstorder_10Percentile, firstorder_Mean, firstorder_Median, ngtdm_Strength | ||||
LoG (σ = 3) | KW LR | 0.719 | 1 | glcm_DifferenceVariance | ||||
LoG (σ = 5) | KW LR | 0.639 | 4 | glcm_Idmn, glcm_Idn, glcm_InverseVariance, ngtdm_Contrast | ||||
