Skip to main content
. 2024 Feb 16;14:3932. doi: 10.1038/s41598-024-54489-9

Figure 9.

Figure 9

Indexing and search F1 score for leave-one-patient-out validation for top-5 consensus predictions in skin dataset plotted for each class, deep network, and experimental setup separately (WD well-differentiated, MD moderately differentiated, PD poorly differentiated, NL normal, B base configuration of using no atlas, IF using normal atlas based on Isolation Forest, OS using normal atlas based on one-class SVM).