Fig. 1

Patient recruitment and study design. The deep-learning-based Radiomic DeepSurv Net was constructed with MRI radiomic features, and was found to be employed for RFS prediction and associated with therapy response and tumor microenvironment (a). This study included three phases to train and validate the RDeepNet model for prediction of RFS and explore the association between radiomics and the treatment or epigenetic biological underpinning. In phase 1, a total of 1113 patients with preoperative MRI from four institutions were enrolled in this study to construct and validate the RDeepNet model for the prediction of recurrence risk. In phase 2, correlation and variance analyses were conducted to examine the change in radiomics in patients before and after neoadjuvant chemotherapy with the response status. In phase 3, 92 of 698 patients from the training cohort underwent RNA-seq with the FFPE samples to obtain lncRNAs data and analyze the association between radiomics with lncRNAs and RFS (b). LncRNAs, long non-coding RNAs; MRI, Magnetic resonance imaging; RFS, recurrence-free survival; T1 + C, contrast-enhanced T1-weighted imaging; T2WI, T2-weighted imaging