Skip to main content

Table 4 List of image features in the entire tumour and intratumoural subregions and the correlations with gene expression modules

From: Tumour heterogeneity revealed by unsupervised decomposition of dynamic contrast-enhanced magnetic resonance imaging is associated with underlying gene expression patterns and poor survival in breast cancer patients

Type Module Feature
Entire tumour
 Entire tumour Tan (n = 158) Kurtosis (r = 0.67)
Magenta (n = 198) Volume (r = 0.6), compactness (r = 0.52)
Tumour subregions
 Plasma input Tan (n = 158) Kurtosis (r = 0.7); volume (r = 0.76); energy (r = 0.55)
Magenta (n = 198) Compactness (r = 0.54)
 Fast-flow kinetics Tan (n = 158) Kurtosis (r = 0.68); energy (r = 0.56); Maximum probability (r = 0.59)
Green yellow (n = 166) Energy (r = 0.58); maximum probability (r = 0.52)
 Slow-flow kinetics Tan (n = 158) Kurtosis (r = 0.55); energy (r = 0.6); Maximum probability (r = 0.53)
Magenta (n = 198) Compactness (r = 0.59)
  1. Only image features that have a high correlation (r > 0.5) with the gene module are shown. The image features with prognostic implications are in bold
  2. Precontrast series (S-0)
  3. Subtraction between the early postcontrast and precontrast series (S-1)