Deep learning radiomics (DLR) is a novel technique in the field of medical imaging combined with artificial intelligence analysis. It can solve several disadvantages of traditional radiomics

including low degree of automation and standardization

tedious feature extraction steps

and time-consuming and labor-consuming. DLR further improves the accuracy and reliability of radiomic models in tumor diagnosis and prognosis prediction. This paper firstly introduced the principle and workflow of the DLR method; then introduced its applications in tumor diagnosis

staging and typing prediction

and survival prognosis evaluation; and finally made the summary and prospect on DLR.