Fig. 4: Integration testing of the DL system for tumor and MSI diagnosis in unannotated WSIs. | Laboratory Investigation

Fig. 4: Integration testing of the DL system for tumor and MSI diagnosis in unannotated WSIs.

From: Interpretable tumor differentiation grade and microsatellite instability recognition in gastric cancer using deep learning

Fig. 4

a Original HE staining WSI. b Differentiation grade recognition based on HE stating WSI. WSI was segmented into image tiles to achieve differentiation grade recognition. Green squares denote WDA. Yellow squares denote PDA. c Heatmap of MSI probability for tumor tiles calculated by the MSI status recognition model. The tumor tiles consisted of WDA and PDA tiles in (b). d Patient-level MSI recognition in the integration testing dataset. Upper, the confusion matrix plot. Down, evaluation of patient-level MSI recognition performance. TNR true negative rate, TPR true positive rate.

Back to article page