Application of artificial neural network and multiple linear regression models for predicting survival time of patients with non-small cell cancer using multiple prognostic factors including FDG-PET measurements.