neurovlm.data.load_dataset

Contents

neurovlm.data.load_dataset#

neurovlm.data.load_dataset(name)[source]#

Alias to _load_* functions in retrieval resources.

Parameters:

name (str) – Name of dataset.

Return type:

dataset

Notes

  • “pubmed_text”: dataframe that includes pubmed dois, pmids, pmcids, titles, abstracts

  • “pubmed_summaries”: dataframe with PubMed pmids, generated/curated summaries, and optional train/val/test flags

  • “pubmed_coordinates”: pubmed coordinate tables

  • “pubmed_images”: tensor containing pubmed images

  • “wiki”: wikipedia article titles and descriptions

  • “cogatlas”: all cogatlas terms

  • “cogatlas_task”: cogatlas task terms

  • “cogatlas_disorder”: cogatlas disorder terms

  • “cogatlas_graph”: how cogatlas terms are related

  • “llm_neuro_terms”: novel LLM-extracted neuroscience terms filtered against CogAtlas and MeSH

  • “networks”: dict that contains atlas and network name keys, and .nii.gz keys.

  • “network_test_set_labels”: labeled network rows with mapped terms and short/long definitions

  • “pubmed_mesh_annotations”: dict mapping PMID strings to MeSH gold terms

  • “pubmed_mesh”: PubMed MeSH term/definition table

  • “publications_neurovault”: neurovault-linked publication metadata (titles, abstracts, dois)

  • “neurovault_text”: publication data for each neurovault image

  • “neurovault_images”: tensor containing neurovault images

  • “neurovault_images_meta” dataframe the maps each image to a study