Table 2 Relative feature importance for the 4 best predicted glioblastoma genetic biomarkers.

From: A fully automated artificial intelligence method for non-invasive, imaging-based identification of genetic alterations in glioblastomas

RankFeatureRelative importance
ATRX mutation
1T1gad mean tumor core1.00
2MD kurtosis non-enhancing tumor0.77
3T1 range tumor core0.52
4T1gad mean whole tumor0.51
IDH mutation
1DWI high gray level emphasis non-enhancing tumor1.00
2MD kurtosis tumor edema0.89
3T1gad variance whole tumor0.78
4T1 kurtosis non-enhancing tumor0.78
Chromosome 7/10 aneuploidy
1FLAIR size zone non-uniformity enhancing tumor1.00
2T1 zone variance tumor edema0.56
3Elongation of enhancing tumor0.55
4T1gad 90th percentile whole tumor0.44
CDKN2 loss
1T1 small area low gray level emphasis tumor edema1.00
2ASL contrast tumor core0.96
3T2 zone entropy tumor edema0.77
4T2 small area high gray level emphasis enhancing tumor0.48