This package contains a PyTorch dataset for ARCADE from "Dataset for Automatic Region-based Coronary Artery Disease Diagnostics Using X-Ray Angiography Images". The dataset can be used for binary and semantic segmentation as well as stenosis detection.

pip3 install git+https://github.com/laudominik/torch-arcade

from torch_arcade import ARCADEBinarySegmentation

ds = ARCADEBinarySegmentation(
    "dataset/",
    image_set="train",
    download="true"
)
img, mask = ds[0]

• ARCADEBinarySegmentation
  • input: image
  • label: image binary mask (shape 512x512x1)
• ARCADESemanticSegmentation
  • input: image
  • label: image semantic mask (shape 512x512x26)
• ARCADEArteryClassification
  • input: image binary mask
  • label: 0 - right artery, 1 - left artery
• ARCADESemanticSegmentationBinary
  • input: image binary mask
  • label: image semantic mask (shape 512x512x26)
• ARCADEStenosisDetection
  • input: image
  • label: coco labels for the stenoses
• ARCADEStenosisSegmentation
  • input: image
  • label: binary mask (0, 1)
• ARCADEInstanceDetection
  • input: image
  • label: { "boxes": [], "labels": [], "masks": [], "image_id": int, "area": float, "iscrowd": bool }