Skip to main content

Table 2 Summary of quantitative imaging features

From: Evaluation of prognostic models developed using standardised image features from different PET automated segmentation methods

Type/order of statistics

Feature

Brief definition

Morphological

Volume

Sum of voxels delineated multiplied by the volume of one voxel

Pre-discretisation

SUVmax

Maximum uptake of FDG in the MTV

Energy

Sum squared SUV values in the MTV

First order

Skewness

Measures symmetry of intensity histogram

Kurtosis

Measures flatness of intensity histogram

Entropy

Measures randomness

Second order

Dissimilarity

Variation of grey level pairs (GLCM). Features were calculated for each unique direction and averaged with a distance setting of 1

Higher order

Grey-level non-uniformity

Distribution of zone counts for each intensity value (GLSZM)

Zone percentage

Fraction of recorded zones compared to maximum possible

Coarseness

Measures spatial rate of change in intensity using a distance of 1