Fig. 5 | Scientific Reports

Fig. 5

From: Pan-cancer frozen section classification using a soft mixture of experts vision transformer under weak supervision

Fig. 5The alternative text for this image may have been generated using AI.

Receiver Operating Characteristic (ROC) curves of three baseline models on the test set. ROC curves comparing the performance of (A) Soft MoE-ViT, (B) Standard ViT, and (C) ResNet50 for binary classification of frozen-section WSIs. Soft MoE-ViT achieved the highest AUC (0.957), followed by Standard ViT (0.934) and ResNet50 (0.921). The optimal operating points selected from the validation set are indicated on each curve.

Back to article page