 This study demonstrated that using tensor deep features, DFs, combined with appropriate machine learning methods improved survival prediction accuracy over conventional DFs, conventional and tensor radio mic features, and end-to-end convolutional neural network, CNN, frameworks. The tensor DF approach was found to be superior to conventional DFs, tensor radioomics, and CNN frameworks for predicting patient outcomes. This article was authored by Muhammad Ar-Sulmanpour, Sa'id Masood Razejo, Mahdi Haseen Zayda, and others.