Skip to main content

Table 5 Ten top features selected by mRMR algorithms for each model

From: Radiomics predictive modeling from dual-time-point FDG PET Ki parametric maps: application to chemotherapy response in lymphoma

H_DTP

H_Static

H_DTP + Static

Non-H_DTP

Non-H_Static

Non-H_DTP + Static

GLCM_Homogeneity

GLCM_Contrast

DTP

GLCM_Energy

GLCM_Homogeneity

GLCM_Dissimilarity

DTP

GLCM_Energy

GLCM_Energy

GLCM_Dissimilarity

DTP

GLRLM_LRE

GLCM_Energy

GLRLM_SRE

DTP

GLRLM_SRE

GLCM_Entropy_log2

GLRLM_SRHGE

DTP

GLRLM_RP

GLCM_Entropy_log2

GLRLM_RLNU

DTP

GLRLM_LRE

GLCM_Entropy

GLRLM_RLNU

DTP

Uniformity

GLCM_Entropy

GLZLM_SZHGE

DTP

GLRLM_RP

GLRLM_SRE

GLZLM_LZE

DTP

Conventional_Skewness

GLRLM_SRE

Uniformity

DTP

Shape_Surface

GLRLM_LRE

GLZLM_SZHGE

Static

GLCM_Contrast

GLRLM_LRE

Conventional_std

DTP

Conventional_Kurtosis

GLRLM_RP

Conventional_std

Static

GLRLM_RLNU

GLRLM_RP

Conventional

Skewness

DTP

Uniformity

Shape_Surface

Conventional Skewness

Static

GLZLM_LZE

SHAPE_Surface

Discretized_Q3

Static

GLCM_Dissimilarity

Conventional

Skewness

Discretized_std

Static

Discretized_std

Conventional

Kurtosis

Discretized_std

Static

GLRLM_RLNU

Uniformity

Discretized

Skewness

Static

Discretized_Skewness

Uniformity

Discretized

Skewness

Static

Discretized_Skewness