MIL-Adapter: Coupling multiple instance learning and vision-language adapters for few-shot slide-level classification | Publicación