Skip to main content
. 2020 Nov 9;24(12):3585–3594. doi: 10.1109/JBHI.2020.3036722

Fig. 1.

Fig. 1.

The pipeline of merged model construction. CT, computed tomography. For each CT image, we obtained 3D ROI (A), from which the radiomic features were extracted (B). After feature selection, we obtained the optimal feature set (C). Meanwhile, deep learning model was constructed (D). The performance of merged model which integrating radiomics and deep learning features was evaluated (E).