Assessing diagnostic complexity: An image feature-based strategy to reduce annotation costs

Dual-energy CT-based radiomics in predicting EGFR mutation status non-invasively in lung adenocarcinoma

“…The grey-level cooccurrence matrix (GLCM) reflects the spatial relationship of grey levels and thereby describes the texture distribution and characteristics inside the tumor. GLSZM (Grey-Level Size Zone) and GLRLM (Grey-Level Run Length Matrix) are closely related to texture heterogeneity [ 24 ] and thus were hypothesized to be used for predicting LUAD EGFR mutations and establishing prediction models. The results showed that the predictive performance of the four models involving radiomics features, namely, M R , M C-R , M R-D , and M C-R-D , was good, with AUC values more than or equal to 0.8 in the training group.…”