Skip to main content
. 2021 May 8;8(3):031906. doi: 10.1117/1.JMI.8.3.031906

Table 5.

A comparison of the learned radiogenomic associations extracted from our neural networks and the modules previously identified in the same dataset.12 Each module consisted of a set of Reactome pathways and a set of image features. Shown are the modules that included the radiomic features used in this study. If any module’s set of pathways was ranked among the top 100 in gene masking, the top three pathways were listed.

Radiomic trait Reactome pathway This study Grossmann et al.12
Test AUC Ranka Module # Pathways
GLCM_diffEntro Cross presentation of soluble exogenous antigens endosomes 0.58 197 2 5
Phase II conjugation 0.69 9 12 35
Regulation of mitotic cell cycle 0.66 21
ABCA transporters in lipid homeostasis 0.65 28
LoG_stats_std Regulation of ornithine decarboxylase 0.94 15 2 5
Cross presentation of soluble exogenous antigens endosomes 0.82 89
Antigen processing cross presentation 0.80 106
Cholesterol biosynthesis 0.75 150 6 7
Signaling by TGF beta receptor complex 0.83 81 8 17
Elongation arrest and recovery 0.82 92
mRNA splicing 0.75 142
LoG_stats_entropy Elongation arrest and recovery 0.73 1 7 8
Mitochondrial protein import 0.63 74
RNA pol III chain elongation 0.58 180
Elongation arrest and recovery 0.73 1 13 26
RNA pol II pre transcription events 0.69 7
Formation of RNA pol II elongation complex 0.69 9
Stats_skewness Antigen processing cross presentation 0.62 139 2 5
a

All Reactome pathways were ranked by test AUC.