honeybee-samples / README.md
Aakash-Tripathi's picture
Upload README.md with huggingface_hub
8939f10 verified
|
raw
history blame
2.98 kB
metadata
license: apache-2.0
task_categories:
  - image-classification
  - feature-extraction
tags:
  - medical
  - pathology
  - radiology
  - clinical
  - oncology
  - whole-slide-image
  - dicom
  - multimodal
size_categories:
  - n<1K
pretty_name: HoneyBee Sample Files

HoneyBee Sample Files

Sample data for the HoneyBee framework — a scalable, modular toolkit for multimodal AI in oncology.

These files are used by the HoneyBee example notebooks to demonstrate clinical, pathology, and radiology processing pipelines.

Paper: HoneyBee: A Scalable Modular Framework for Creating Multimodal Oncology Datasets with Foundational Embedding Models Package: pip install honeybee-ml

Files

File Type Size Description
sample.PDF Clinical 70 KB De-identified clinical report (PDF) for NLP extraction
sample.svs Pathology 146 MB Whole-slide image (Aperio SVS) for tissue detection, patch extraction, and embedding
CT/ Radiology 105 MB CT scan with 2 DICOM series (205 slices total) for radiology preprocessing

CT Directory Structure

CT/
├── 1.3.6.1.4.1.14519.5.2.1.6450.4007.1209.../ (101 slices)
└── 1.3.6.1.4.1.14519.5.2.1.6450.4007.2906.../ (104 slices)

Usage

Install HoneyBee:

pip install honeybee-ml[all]

Pathology — Load a whole-slide image

from huggingface_hub import hf_hub_download
from honeybee.loaders.Slide.slide import Slide
from honeybee.processors.wsi import PatchExtractor

slide_path = hf_hub_download(
    repo_id="Lab-Rasool/honeybee-samples",
    filename="sample.svs",
    repo_type="dataset",
)

slide = Slide(slide_path)
slide.detect_tissue(method="otsu")
patches = PatchExtractor(patch_size=256).extract(slide)
print(f"Extracted {len(patches)} patches from {slide.dimensions}")

Clinical — Process a clinical PDF

from huggingface_hub import hf_hub_download
from honeybee.processors import ClinicalProcessor

pdf_path = hf_hub_download(
    repo_id="Lab-Rasool/honeybee-samples",
    filename="sample.PDF",
    repo_type="dataset",
)

processor = ClinicalProcessor()
result = processor.process(pdf_path)
print(result["entities"])

Radiology — Download CT DICOM series

from huggingface_hub import snapshot_download

ct_dir = snapshot_download(
    repo_id="Lab-Rasool/honeybee-samples",
    repo_type="dataset",
    allow_patterns="CT/**",
)

Citation

@article{rasool2024honeybee,
  title={HoneyBee: A Scalable Modular Framework for Creating Multimodal Oncology Datasets with Foundational Embedding Models},
  author={Rasool, Ghulam and others},
  journal={arXiv preprint arXiv:2405.07460},
  year={2024}
}

License

Apache 2.0 — see the HoneyBee repository for details.