Skip to main content
. 2023 Sep 7;23:83. doi: 10.1186/s40644-023-00605-3

Fig. 1.

Fig. 1

Workflow of the study. Preoperative chest CT images of patients were retrospectively collected and pre-processed, and then segmented for features extraction. Six radiomic signatures were constructed after feature selection. Postoperative haematoxylin and eosin-stained sections of patients were reviewed and assessed for STAS status and combined with radiomic signatures to construct RAISm. RAISm performance was evaluated in the training set, test set and validation set, respectively