Performance data and derived metrics of a repeated, 3-fold cross validation of radiomics data for classification of myxoid tumours into low/intermediate and high grade.

This dataset contains the raw performance results and the derived performance metrics of a 10x repeated, 3-fold cross validation of a Random Forest classifier, trained to differentiate between low/intermediate and high grade myxoid liposarcoma tumours, based on radiomics extracted from MR volumes using the PyRadiomics software package. These values are available here so that performance metrics can be validated, or additional metrics can be computed from the class probabilities and true labels provided for each run and fold of the cross validation, if required. Metrics, such as Positive Predictive Value and Negative Predictive Value, were calculated using a threshold of 0.5. The F1 score is a weighted average.