Figure 5
From: Development and clinical application of deep learning model for lung nodules screening on CT images
B–A plots showing agreement tests for positive nodule identification between the expert-reference standard and the DL algorithm. The plots elucidate the difference in the number of positive nodules identified by the experts and the DL algorithm (y-axis) over the mean of two measurements (x-axis) per study. For each plot, the line and value of the mean difference (blue line) and the upper and lower limits of 95% agreement (red dotted lines) are shown. The darker the dot colour was, the denser (more frequent) the data appeared on the specific X–Y axis area. (A) Agreement plot for the multi-centre validation set; (B) agreement plot for the LUNA set. To further assess the agreement for detecting micro-nodules and large nodules, subgroup analyses were performed for a nodule diameter < 4 mm (C) or ≥ 4 mm (D) in the multi-centre study data set.
