Ordinal Multi-modal Feature Selection for Survival Analysis of Early-Stage Renal Cancer Wei Shao 1 , Jun Cheng 2 , Liang Sun 1 , Zhi Han 4 , Qianjin Feng 3 , Daoqiang Zhang 1(B) , and Kun Huang 4(B) 1 College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China dqzhang@nuaa.edu.cn 2 School of Biomedical Engineering, Shenzhen University, Shenzhen 518073, China 3 School of Biomedical Engineering, Southern Medical University, Guangzhou 4 School of Medicine, Indiana University, Indianapolis, IN 46202, USA kunhuang@iu.edu Abstract. Existing studies have demonstrated that combining genomic data and histopathological images can better stratify cancer patients with distinct prognosis than using single biomarker, for di erent biomarkers may provide ff complementary information. However, these multi-modal data, most high- dimensional, may contain redundant fea-tures that will deteriorate the performance of the prognosis model, and therefore it has become a challenging problem to select the informative features for survival analysis from the redundant and heterogeneous fea-ture groups. Existing feature selection methods assume that the survival information of one patient is independent to another, and thus miss the ordinal relationship among the survival time of di erent patients. To solve this issue, we make use of the important ordinal ff survival informa-tion among di erent patients and propose an ordinal sparse ff canonical correlation analysis (i.e., OSCCA) framework to simultaneously identify important image features and eigengenes for survival analysis. Specifi- cally, we formulate our framework basing on sparse canonical correlation analysis model, which aims at finding the best linear projections so that the highest correlation between the selected image features and eigen-genes can be achieved. In addition, we also add constrains to ensure that the ordinal survival information of di erent patients is preserved after projection. We evaluate the ff e ectiveness of our method on an early-stage renal cell carcinoma dataset. ff Experimental results demonstrate that the selected features correlated strongly with survival, by which we can achieve better patient stratification than the comparing methods. ___________________________________________________________________ This is the author's manuscript of the article published in final edited form as: Shao, W., Cheng, J., Sun, L., Han, Z., Feng, Q., Zhang, D., & Huang, K. (2018). Ordinal Multi-modal Feature Selection for Survival Analysis of Early-Stage Renal Cancer. In A. F. Frangi, J. A. Schnabel, C. Davatzikos, C. Alberola-López, & G. Fichtinger (Eds.), Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 (pp. 648–656). Springer International Publishing. https://doi.org/10.1007/978-3-030-00934-2_72