loki.annotate.annotate_with_marker_genes

loki.annotate.annotate_with_marker_genes(classes, image_embeddings, all_text_embeddings)

Annotates tissue image with similarity scores between image features and marker gene features.

Parameters:

• classes – A list or array of tissue type labels.

• image_embeddings – A numpy array or torch tensor of image embeddings (shape: [n_images, embedding_dim]).

• all_text_embeddings – A numpy array or torch tensor of text embeddings of the marker genes (shape: [n_classes, embedding_dim]).

Returns:

• dot_similarity: The matrix of dot product similarities between image embeddings and text embeddings.

• pred_class: The predicted tissue type for the image based on the highest similarity score.