Figure 2.

HCC patients stratified into two subtypes by multi-view radiomics features. (A) The process of multi-view feature extraction. After patient screening and multi-view VOIs’ segmentation, multi-view features are obtained from four types of VOIs. (B) Consensus clustering of HCC patients based on CP1 fused features reveals two distinct subtypes. The strength of the blue color is proportional to the frequency at which samples have been clustered together. (C) Principal component analysis of HCC patients based on CP1 fused features separates FIFS1 and FIFS2. (D,E) Kaplan–Meier plots show the prognosis association of subtypes with overall survival (D) and disease-free survival (E) outcomes.