Fig. 4: Representative examples of uncertainty quantification in two gastric cancer samples. | npj Digital Medicine

Fig. 4: Representative examples of uncertainty quantification in two gastric cancer samples.

From: Deep Gaussian process with uncertainty estimation for microsatellite instability and immunotherapy response prediction from histology

Fig. 4: Representative examples of uncertainty quantification in two gastric cancer samples.

a, c WSI images with tile-level MSI-H prediction probability heat maps. b, d Histograms quantifying tiles based on predicted probability of MSI-H status. While both slides were considered MSI-H, the sum predictive probability was 0.996 (BCS = 0.871) for the highly confident sample, and the sum predictive probability was 0.503 (BCS << 0.001) for the highly uncertain slide.

Back to article page