Fig. 3: Diagnostic Performance of the proposed Liver Artificial Intelligence Diagnosis System (LiAIDS) in All Cohorts.
From: A multicenter clinical AI system study for detection and diagnosis of focal liver lesions
a ROC curve of the internal validation cohort; b ROC curve of the external validation cohort of ZZH; c ROC curve of the external validation cohort of QZH; d ROC curve of the external validation cohort of NBH; e ROC curve of the prospective validation cohort of SRRSH; f t-SNE plot of the prospective validation cohort. Scatter plot illustrating the clustering of lesion images in the prospective cohort. Each point represents an image of a lesion, and the color indicates its true class; g Performance of the binary classification across all validation cohorts, evaluated using all five metrics; h Confusion matrix of the internal validation cohort; i Confusion matrix of the external validation cohort of ZZH; j Confusion matrix of the external validation cohort of QZH; k confusion matrix of the external validation cohort of NBH; l Confusion matrix of the prospective validation cohort of SRRSH’s. Source data are provided as a Source Data file (Source_data_Figure_3.xlsx).
