lazyslide_models.vision.MadeleineSlideEncoder

class MadeleineSlideEncoder(model_path=None, token=None)

Bases: SlideEncoderModel

madeleine 🤗Hugging Face GitHub Paper Params: 4.7M CC BY-NC-ND 4.0 [Jaume et al., 2024] Multistain Pretraining for Slide Representation Learning in Pathology MADELEINE H&E slide encoder.

Trained on CONCH features of 256 px tiles at 10x; extract them with conch-madeleine, which reproduces MADELEINE’s patch embedder.

encode_slide(embeddings, coords=None, **kwargs)

Encode the slide using the Madeleine slide encoder. The embeddings should be a tensor of shape [B, C, H, W].