Multi-resolution vision transformer model for histopathological skin cancer subtype classification using whole slide images

The multi-resolution model has the potential to assist anatomical pathologists in automatically detecting, highlighting, and classifying subtypes of melanoma and non-melanoma skin cancer subtypes directly from WSIs. Its capability could lead to more effective clinical decision-making in digital pathology, ultimately improving patient outcomes. For non-melanoma skin cancer, the model could be deployed in resource-limited settings with high incidence of the disease and limited access to experienced dermatopathologists.