Table 1 Overview of quantitative imaging features and their descriptions.

From: Radiomics-enhanced modelling approach for predicting the need for ECMO in ARDS patients: a retrospective cohort study

Features

Name

Description

HU based characteristics

Mean

Mean value of the distribution of HU values in a ROI.

Standard deviation

Standard deviation of the distribution of HU values in a ROI.

Kurtosis

Kurtosis of the distribution of HU values in a ROI.

Skewness

Skewness of the distribution of HU values in a ROI.

Gas volume

Gas volume of a ROI (in \(\:{mm}^{3}\)) according to formula (3).

Water volume

Water volume of a ROI (in \(\:{mm}^{3}\)) according to formula (4).

Tissue weight

Tissue weight of a ROI (in g) according to formula (5).

Relative position of the center of mass; transversal

Position of the HU-weighted center of mass of a ROI along the transversal axis from right to left, relative to the boundary of the lung mask.

Relative position of the center of mass; sagittal

Position of the HU-weighted center of mass of a ROI along the sagittal axis from ventral to dorsal, relative to the boundary of the lung mask.

Relative position of the center of mass; longitudinal

Position of the HU-weighted center of mass of a ROI along the longitudinal axis from caudal to cranial, relative to the boundary of the lung mask.

Geometric characteristics

Volume

Volume of a ROI (in \(\:{mm}^{3}\)) according to formula (1).

Transversal dimension

Spatial dimensions of a ROI (in \(\:mm)\) along the transversal axis.

Sagittal dimension

Spatial dimensions of a ROI (in \(\:mm)\) along the sagittal axis.

Longitudinal dimension

Spatial dimensions of a ROI (in \(\:mm)\) along the longitudinal axis.

  1. Quantitative imaging features. The table shows both HU-based and geometric characteristics, both of which are extracted from CT images via quantitative image analysis. Formulas (1, 3–5) referred to in the table are found within the methods section. HU: Hounsfield unit; ROI: region of interest.